Post by: Johnny B on May 20, 2005, 02:40:37 PM
http://members.tripod.com/~gabevee/harmonics.html
Post by: Johnny B on May 21, 2005, 12:52:19 PM
No one wants to do any math homework and check the stated results?
It must be almost time for the summer break, but school is not out yet, is it?
Post by: danlavry on May 21, 2005, 01:50:02 PM
| Johnny B wrote on Sat, 21 May 2005 17:52 |
No one wants to do any math homework and check the stated results? It must be almost time for the summer break, but school is not out yet, is it? |
Johnny,
I did not "check the math" and it is available at a lot of books. It is the math for Fourier SERIES, which is an analysis method suitable for STEADY STATE SIGNALS - that is a signal that is made of never ending repetition of IDENTICAL CYCLES.
Fourier series math is correct, and it was not intended for audio only. It is a very useful tool for many applications.
But the Fourier SERIES is only STEP 1. There is also the Fourier integral, a much further evolved method of enjoying ANY continues wave, over a given time period. The Fourier integral is not restricted to a "forever repetitive" wave, thus it is much better for analysing audio signals.
The math for Fourier integral is more demanding, but the flexibility in terms of ability to deal with all sorts of signals is so much greater.
The Fourier integral (which does not require steady state) is a very powerful math and engineering method to analyse almost any realistic signal, that can be expressed mathematically (be it a sine wave, a step, an FM radio signal, an impulse...
As I already answered you in another thread, the FFT is a computational analysis tool (based on collecting a "chunk of signal data), and FFT (Fast Fourier Transform) is a "subset" of Fourier analysis. It is useful when you are measuring a signal that you do not know (thus can not express mathematically).
I am repeating what I already said in this forum...
Fourier analysis is one of many tools that can be used for designing or understanding audio. I would only consider looking at audio through the eyes of the very limited Fourier series, when music becomes an infinite repetitive identical cycles....
Regards
Dan Lavry
Post by: Johnny B on May 21, 2005, 02:13:58 PM
Perhaps you missed my lame attempt at humor, that's ok, my wife often misses it and takes me far too seriously when I'm just kidding.
However, I'd be interested in your impressions of what the author of that linked article is really "trying" to say.
I sort of admire one who has a seeking spirit and who is asking or re-asking questions which I feel is in keeping with scientific progress and helps keep the technology forever moving forward.
I'd also be interested in hearing what others impressions of the linked article are as well.
Post by: danlavry on May 21, 2005, 02:59:10 PM
| Johnny B wrote on Sat, 21 May 2005 19:13 |
Dan, Perhaps you missed my attempt at humor, that's ok, my wife often misses it and takes me far too seriously when I'm just kidding. However, I'd be interested in your take of what the author of that linked article is really trying say. |
First he is talking about Fourier series. Than he states that a non linear transfer curve would yield harmonic distortion. Nothing new.
After that, I disagree with most of what he says and with his most of his conclusions. I do not want to waste my time explaining, but will be glad to respond to a single point or two if you wish to bring any up.
Regards
Dan Lavry
Post by: Johnny B on May 21, 2005, 03:24:30 PM
I kind of understood him to be trying to get at the linear vs. non-linear phenom that I would guess that most already know about, but I wonder if that might be an area that is worth re-visiting. Of course we know that many people are doing algo's to try to emulate the non-lineraities found in analogue to varying degress of success and failure, but IIRC, he also touches on the harmonic overtone series which is something that has always facinated me.
In addition, if I'm understanding him correctly (I very well may be misunderstanding him completely) I think he is calling into question the ability to accurately represent certain waveforms with existing technology. Again, maybe I misuderstood him.
In any event, however ill- or well-founded a question may be, I always think it's good that people keep asking questions.
Post by: danlavry on May 21, 2005, 04:45:31 PM
| Johnny B wrote on Sat, 21 May 2005 20:24 |
...In any event, however ill- or well-founded a question may be, I always think it's good that people keep asking questions. |
This is how he ended his writeup:
"We have now two factors in favor of the triode regarding harmonic distortions. 1) The Fourier derived harmonic pattern of the triode is similar to natural sources of music and adds to the harmonics of the music in proportion, making for a perceptibly louder output. 2) And the input of the tube does not load down the previous stage such that it alters the harmonic amplitude proportions. Which one you subscribe to is your choice.
But as for the device of choice? The tube wins this contest."
It does not sound like a question. More like a statement, much of it based on previous statements I disagree with completely...
Regrads
Dan Lavry
Post by: Johnny B on May 22, 2005, 01:21:21 AM
I understood him to be possibly getting at the early distortion characteristics of tubes vs. transistors or ICs as discussed by Russell O. Hamm in his famous AES white paper. I may be reading something into the linked article that is not there, or just as bad, assuming something that is not appropriate because the author may not have said it.
For all I know, the author might be a guitar player as well, and you know how they love those tube amps. I'd rather try to take a bone away from a big dog than try to take a tube amp away from a guitar player. In any event, clearly he likes tubes or valves best.
Post by: danlavry on May 24, 2005, 02:12:49 PM
| Johnny B wrote on Sun, 22 May 2005 06:21 |
Dan, I understood him to be possibly getting at the early distortion characteristics of tubes vs. transistors or ICs as discussed by Russell O. Hamm in his famous AES white paper. I may be reading something into the linked article that is not there, or just as bad, assuming something that is not appropriate because the author may not have said it. For all I know, the author might be a guitar player as well, and you know how they love those tube amps. I'd rather try to take a bone away from a big dog than try to take a tube amp away from a guitar player. In any event, clearly he likes tubes or valves best. |
You are talking about a paper that was written in 1972, about technology of the 60's. The paper is based on characteristics of tubes and transistors, used by as poor circuit design that one can imagine. It certainly does not have much to do with electronics today.
I am less than impressed when someone attacks a transistor based on some elementary circuit that no one ever uses, getting everything out of context.
Regards
Dan Lavry
www.lavryengineering.com
Post by: Johnny B on May 24, 2005, 04:16:44 PM
Certainly, the re-created and updated test suite could take into account the technological changes or advancements that have occurred since 1973. Perhaps the tests could be conducted at a facility like Sear Sound in NYC. I could easily see the results being published as a new AES White Paper or possibly leading to a 'zine article.
Post by: danlavry on May 24, 2005, 04:45:01 PM
| Johnny B wrote on Tue, 24 May 2005 21:16 |
I think it would interesting for someone such as yourself (i.e. someone with a nice AP Test Box) to go back and try to recreate and perhaps update Russell O. Hamm's White Paper. Certainly, the re-created and updated test suite could take into account the technological changes or advancements that have occurred since 1973. Perhaps the tests could be conducted at a facility like Sear Sound in NYC. I could easily see the results being published as a new AES White Paper or possibly leading to a 'zine article. |
Johnny,
Not my kind of a project. I come from times when transistors were replacing tubes. I do not long for selenium rectifiers, or the days when a 10000 tube computer could not finish a calculation because a tube would fail every few minutes. I do not long for the days when the alternative to computing was a combination of analog discrete amps, integrators, differentiators, potentiometers and a box of wires with banana plugs...
Regards
Dan Lavry
www.lavryengineering.com
Post by: Johnny B on May 24, 2005, 05:27:34 PM
Now if I had a real working brain and a good AP Test Box, I might undertake such a project myself, but alas, I am sadly lacking in both categories.
I do hope that someone far smarter than I will take up the challenge.
Post by: George_ on May 25, 2005, 12:45:00 AM
| Quote: |
Dan, Perhaps you missed my attempt at humor, that's ok, my wife often misses it and takes me far too seriously when I'm just kidding. |
Dan,
try not to A/D (Eye/Brain) everything with 32bit/7MHz and think too much over it..
Johnny.. maybe I'm wrong and Dan Lavry is bored all the time.. but I think, he has some work to do with his converters..
Dan, if you are really bored.. I wait for the first 12 channel A/D.. there are lots and lots of 8 and 16 channels.. but I need 10 (12 on the safe side).. It's more cost effective..
Drum recording usually takes more than 8 and less than 16 channels(ok.. no dreamtheater please)
I'm waiting for my christmas present;)
HAVE a nice day!!
Post by: trevord on May 27, 2005, 01:41:19 PM
| danlavry wrote on Tue, 24 May 2005 21:45 |
I do not long for the days when the alternative to computing was a combination of analog discrete amps, integrators, differentiators, potentiometers and a box of wires with banana plugs... Regards Dan Lavry www.lavryengineering.com |
hey ..
what's wrong with analog computing. i was in college in the very last days of analog computing and it was capable of some neat stuff. I even had to take a stab a hydraulic (or fluid) logic, now there's some interesting stuff.
Post by: pipelineaudio on May 27, 2005, 08:30:01 PM
Post by: Johnny B on May 28, 2005, 07:31:44 PM
but with music...Analogue is THE alpha and THE omega.
Post by: dayvel on May 30, 2005, 10:24:40 PM
Post by: trevord on June 01, 2005, 01:08:28 PM
i remember a lot of debate.
The main problem is asymptotic relationships. The really interesting things in math happens in the "tends to" region and digital math doesn't cut it.
well..
it does within "acceptable limits"
If you think those problems are solved, talk to the people at Intel about the little floating point snafu that sent quakes thru the audio industry.
I was always amazed at the ease of representing physical systems with circuitry and vice versa.
I am not sure how it would apply to audio tho.
Normally you think of a "computer" replacing an actual circuit, so an analog computer circuit would replace an actual analog circuit?
Maybe it doesn't quite apply.
trevor
Post by: danlavry on June 01, 2005, 02:15:45 PM
| trevord wrote on Wed, 01 June 2005 18:08 |
yeah.. i remember a lot of debate. The main problem is asymptotic relationships. The really interesting things in math happens in the "tends to" region and digital math doesn't cut it. well.. it does within "acceptable limits" If you think those problems are solved, talk to the people at Intel about the little floating point snafu that sent quakes thru the audio industry. I was always amazed at the ease of representing physical systems with circuitry and vice versa. I am not sure how it would apply to audio tho. Normally you think of a "computer" replacing an actual circuit, so an analog computer circuit would replace an actual analog circuit? Maybe it doesn't quite apply. trevor |
Trevor,
The digital math provides a PERFECT representation of the analog signal, as long as you obey Nyquist and sample faster then twice the signal bandwidth. It is not easy to swallow with simple common sense, but it is is a fact. Check my paper "Sampling Theory" on my web site, and you may get a better sense of it.
What ever objections you may have to digital, it is not about the math. The sources of objections may be about resolution, truncation, aliasing, long time delays... all are implementations of perfect math model, and many of the issues are being addressed. The technology is not perfect, but it is better than it was, and will continue to improve.
Regards
Dan Lavry
www.lavryengineering.com
Post by: trevord on June 01, 2005, 05:11:12 PM
I am talking about the quality of the math algorithms' handling of small values.
For example if a formula tends to 0 with a asymptote, most algorithms "give up" quickly and the value of the formula becomes 0. This describes the quality of the algorithm (not the digital representation of an analog signal).
Math algorithms are far from perfect because they are the result of compromises between accuracy and processing time.
These are valid design choices and the "correctness" of the choice depends on the context.
Some of these choices maybe things like..
Smaller terms of the polynomials are usually ignored.
The numbers of "taps" of a filter may be reduced.
etc.
In fact, some more complex scenarios are sensitive to the "initial conditions" and will not produce a solution if the input is not in the expected range.
To sum up.
I was not referring to the representation of an analog signal in the digital domain. It can be a perfect representation - we agree.
I was not referring to the theory (or equation) of a particular formula. Theoretically - the formula is correct - we agree.
I was referring to the implementation (trade-offs) of the formula by design choices made in a digital system.
Two examples...
The classic case is now the infamous pentium "NAN" flaw.
To (very) briefly sum it up.
FP representation changes when a number get very close to 0
The theory behind this works.
BUT
The Intel Pentium 4 implementation did an exception to execute special code to handle this case (thus making the FPU simpler and faster) - again - it works
Of course this meant FP operations which hit this would run orders of magnitude slower, but they (Intel) assumed this happens in only a small percentage of cases.
BUT
This is the bread and butter of most audio math algorithms.
(think reverb tail)
This resulted in the "CPU spike" most of you saw with your DAW, and the subsequent re-coding of most the math algorithms for these apps.
Another example is aliasing noise in virtual analog synths. Again - this is not the aliasing in the digital/analog conversion, this aliasing is the result of poor choices made in the implementation of the synthesis formulas (ae) by DSP programmers.
Which is why you have price differences in virtual analog synths.
(to "ear people" - your ears are right, there is a difference)
So the quality of the implementation of the math makes a very big difference.
Post by: danlavry on June 01, 2005, 05:53:51 PM
| trevord wrote on Wed, 01 June 2005 22:11 |
i am not talking about digital representation of analog as in a to d. I am talking about the quality of the math algorithms' handling of small values.... To sum up. I was not referring to the representation of an analog signal in the digital domain. It can be a perfect representation - we agree. I was not referring to the theory (or equation) of a particular formula. Theoretically - the formula is correct - we agree. I was referring to the implementation (trade-offs) of the formula by design choices made in a digital system.... ...So the quality of the implementation of the math makes a very big difference... |
Thanks for clarifying. Now that I understand what you mean, I agree with all that you said.
Regards
Dan Lavry
www.lavryengineering.com
Post by: Johnny B on June 01, 2005, 07:15:35 PM
| danlavry wrote on Wed, 01 June 2005 19:15 |
[Digital "technology is not perfect, but it is better than it was, and will continue to improve." |
I agree with this.
Post by: Johnny B on June 01, 2005, 07:20:32 PM
| trevord wrote on Wed, 01 June 2005 22:11 |
This "aliasing is the result of poor choices made in the implementation of the synthesis formulas (ae) by DSP programmers. Which is why you have price differences in virtual analog synths. (to "ear people" - your ears are right, there is a difference) So the quality of the implementation of the math makes a very big difference." |
I also agree with much of this, many bad trade-offs have been made, and this may partially explain why the "ear people" continue to complain about digital's bad sound quaility.
Post by: trevord on June 01, 2005, 11:07:29 PM
what is even more disturbing is the fact that some of these issues we solved many years ago.
For example, the orignal korg oasys is a dream to listen to, i know i did not know how good a virtual analog could sound until i got the original oasys pci card. But this was how long ago, 10-15 years ago. I still fight tooth and nail to keep it going despite lack of support.
And with the pentium FPU problem - many programmers are opting for the "easy" way out by adding moise to low level signals when the signal gets low enough to cause the problem. So much for "pristine" mixing "in the box".
I have a lot of respect for "ear people", i am more a "math" person, but i have trained myself to go past the explanations when talking to ear people and listen to what they are really saying.
9 times out of 10 a good ear person will detect a problem before the math people figure it out.
thats why we need more ear people designers
if you throw enough money at the problem a good design is a no-brainer
but when you have to meet cost considerations you need an ear to determine which is the "best" solution
i think right now we seeing the most cost effective designs - not the best designs
(present company excepted
)Post by: RedBus on June 02, 2005, 12:24:57 PM
| Quote: |
And with the pentium FPU problem - many programmers are opting for the "easy" way out by adding moise to low level signals when the signal gets low enough to cause the problem. So much for "pristine" mixing "in the box". |
Stop right there before this misconception gets out of hand.
It is true that adding low level noise is an acceptable way to avoid the Pentium processor from degenerating a floating point signal into "denormal" representation, which incurs a 10x to 100x processing time penalty. BUT, that noise signal can be at -150dB to -200dB and still have the correct effect. Don't let anyone believe that they can hear that signal on top of the audio passing through the DAW's mixing bus.
If you were adding that noise in an algorithm where it will get fed back, then there may be problems, but then that is the wrong choice to make for that design.
RedBus.
Post by: trevord on June 02, 2005, 01:25:57 PM
| RedBus wrote on Thu, 02 June 2005 17:24 | ||
Stop right there before this misconception gets out of hand. It is true that adding low level noise is an acceptable way to avoid the Pentium processor from degenerating a floating point signal into "denormal" representation, which incurs a 10x to 100x processing time penalty. BUT, that noise signal can be at -150dB to -200dB and still have the correct effect. Don't let anyone believe that they can hear that signal on top of the audio passing through the DAW's mixing bus. If you were adding that noise in an algorithm where it will get fed back, then there may be problems, but then that is the wrong choice to make for that design. RedBus. |
i agree..
the point i wanted to make was..
a lot of lower priced (or free) DAW and VSTi vendors say use the "math is math" argument
but they are making choices which can affect the quality of the result.
Usually is goes like
an ear person says "it sounds funny to me"
the free VSTi guy says "math is math, we do it just like the big boys"
but if you probe a little deeper you may find some of these bad choices.
I think you are trying to prevent the usual round of "see i told you so" from ear people, but "math is math" is just as bad a misconception.
Post by: RedBus on June 02, 2005, 04:25:08 PM
| Quote: |
the point i wanted to make was.. a lot of lower priced (or free) DAW and VSTi vendors say use the "math is math" argument but they are making choices which can affect the quality of the result. Usually is goes like an ear person says "it sounds funny to me" the free VSTi guy says "math is math, we do it just like the big boys" but if you probe a little deeper you may find some of these bad choices. |
Part of the misconception is that the mistakes made by the "free" DAW or plug-in creators are duplicated in the commercial DAWs. For confidentiality reasons you'll never see those actual engineers speak up and talk about what choices and why they made them in implementing summing et al.
Lynn Fuston's Awesome DAWsum CD dispells much of the myth when it comes to the current generation of DAWs.
The plug-in creators further down the food chain are trying to gets sales wherever they can (if it's not free), so they'll spin their own limitations however many ways required to get other people to buy.
This is starting to veer a long way off the original topic, but we are talking about the same plug-in creators who put the following kind of text in their press release (from Harmony Central).
ParametrEQ Common Features:
"Low CPU usage"
"Randomize features: choose to randomly set the bandwidth, frequencies or gain separately."
and no mention whether there has been attention paid to word length issues or denormals...
RedBus.
Post by: Johnny B on June 02, 2005, 06:15:43 PM
What DAW Sum "myth" does Lynn's CD "dispell?"
And, perhaps more importantly, can we honestly attribute all the bad sound quality charges to only this narrow cause?
Post by: trevord on June 02, 2005, 06:49:57 PM
But we are referring to the use of math for effects purposes (i don't think you have to use Fourier for summing).
IMHO - the differences between DAWs would show if there was a way to set set EQ (or some other effect) parameters on a track.
I am not sure there is a way to qualify the "correctness" of an effect processing implementation.
take EQ for example
there could be a theoretical output for a given Q,f and cut,
but some DAWs may strive to emulate more "musical" (vintage) EQ boxes, which is not wrong by itself.
Plus its not just the results of the math algorithm, but the ability of the DSP implementation to handle increased processor loads.
For example a particular effect implementation may sound good with no processor load, but if the design choice is too cpu intensive, then running more tracks may result in inferior sound (or outright failure).
Is a more "accurate" cpu intensive effect implementation better
than a less "accurate" implementation which can handle more tracks?
And we haven't even talked about what happens when the lesser accuracy produces a more pleasing / desirable sound.
With virtual analog DSP programming, for example, a major part of the difficulty is reproducing things like random frequency drift and imperfections in oscillator sync'ing
The only way to judge a good or bad math algorithm choice is how it sounds.
Even that may be misleading, because some DAWs use a less accurate effect implementation for real-time and a more accurate cpu intensive version for "printing" the track or for mixdown.
it's a very grey world when things go from theory to implementation.
Post by: RedBus on June 02, 2005, 07:44:39 PM
If we start to discuss the implementation in the digital world of the flaws of a particular analog circuit, then the sky is the limit and people will have a right to their opinion if they say the digital version doesn't sound correct, etc.
Implementation in digital of the non-linear processes that sometimes goes on in the analog world (e.g. opto-coupler in a compressor) does bring us back to the trade-offs, good and bad, that people make.
One cardinal matematical rule about such implementation is that you have to deal with the higher frequencies generated in the process. Usually this means upsampling the incoming data and performing the whole simulation at a higher rate, then filtering it back down to the original rate. The quality of this conversion dictates the accuracy of the model, assuming that your model was mathematically close to the analog process to start with.
But, one shortcut you can take is to not do the upsampling. That guarantees you get aliased audio when you do your analog simulation, but it may end up being an entirely new effect that you want to keep. I'm sure that this wouldn't work for all audio sources you throw at such an algorithm (assuming you were going for a model of a smooth opto-compressor instead of a grungy distorted comp), but it opens up new paths.
RedBus.
Post by: 3D Audio on September 26, 2005, 08:53:57 PM
| RedBus wrote on Fri, 03 June 2005 00:44 |
The Myth is basically that DAWs sound different because of the way their summing is implemented. The CD proves that given the same input files and the same fader levels, you get the same data (bit-for-bit accurate) out of most DAWs. If you still hear a difference, it is because of other parts of your system. |
The part we never figured out (and possibly never will) is why people felt they could clearly hear differences in bit-identical files. I did the math once and not one single value of one 24-bit word among 3 million words (I think it was) was different at all. Yet people still thought they could hear the difference.
I did not read the whole thread so pardon me if this comment is out of line.
Post by: Ronny on September 27, 2005, 03:03:08 AM
| 3D Audio wrote on Mon, 26 September 2005 20:53 | ||
The part we never figured out (and possibly never will) is why people felt they could clearly hear differences in bit-identical files. I did the math once and not one single value of one 24-bit word among 3 million words (I think it was) was different at all. Yet people still thought they could hear the difference. I did not read the whole thread so pardon me if this comment is out of line. |
Very easily explained. The mind tends to fill in blanks when there are none. Our ears are great, but our short term audio memory isn't so hot and we know very little about the audio cortex and how other senses, preconceived thought patterns and mood affect the sense of hearing, or at least the information derived from it. The design of the ear is a great device, but it still has the brain, that has to coherently assemble the information into meaningful data or the lack of data and that's where the problem lies. It's harder to train your ears not to hear differences when there are none, than it takes to train the ears to hear differences when there are some. You have all kinds of non audio factors that enter the equation, one is the egotistical notion of not admitting hearing anything, when there is nothing to hear. I call it golden ear syndrome.
In response to RedBus' statement, if you still hear differences, it's likely that it has more to do with your own system, than differences in the audio one, IMHO.
Post by: danlavry on September 28, 2005, 07:10:24 PM
| RedBus wrote on Fri, 03 June 2005 00:44 |
The Myth is basically that DAWs sound different because of the way their summing is implemented. The CD proves that given the same input files and the same fader levels, you get the same data (bit-for-bit accurate) out of most DAWs. If you still hear a difference, it is because of other parts of your system. If we start to discuss the implementation in the digital world of the flaws of a particular analog circuit, then the sky is the limit and people will have a right to their opinion if they say the digital version doesn't sound correct, etc. Implementation in digital of the non-linear processes that sometimes goes on in the analog world (e.g. opto-coupler in a compressor) does bring us back to the trade-offs, good and bad, that people make. One cardinal matematical rule about such implementation is that you have to deal with the higher frequencies generated in the process. Usually this means upsampling the incoming data and performing the whole simulation at a higher rate, then filtering it back down to the original rate. The quality of this conversion dictates the accuracy of the model, assuming that your model was mathematically close to the analog process to start with. But, one shortcut you can take is to not do the upsampling. That guarantees you get aliased audio when you do your analog simulation, but it may end up being an entirely new effect that you want to keep. I'm sure that this wouldn't work for all audio sources you throw at such an algorithm (assuming you were going for a model of a smooth opto-compressor instead of a grungy distorted comp), but it opens up new paths. RedBus. |
I do not wish to argue with "new path" but let me point out that alaising takes frequencies and moves them to frequency locations that are "highly non musical" in the sense of thier relationship to the intended (non aliased) sound.
A very simple example: Take a 5KHz sound with harmonics at 10KHz, 15KHz, 20KHz, 25KHz and 30KHz. Say you ampled at 44.1LHz so Nyquist is at 22.050KHz. The 25KHz is aliased to 19.1Khz and the 30KHz is now at 14.1KHz.
With an analog system limited to say 22KHz the sound you can hear is made of 5, 10, 15 and 20KHz. Very harmonic sound.
With the aliasing you have 5,10,14.1, 19.1 and 20KHz. I would say that the aliasing is non musicaly related to the energy.
Of course, when having many instruments together, we have a more complex picture, but I think the concept is the same. I have listened to some non oversampling AD's with the aliasing protection (filters) intentionaly removed. It is really bad.
Regards
Dan Lavry
www.lavryengineering.com
Post by: RedBus on September 29, 2005, 03:41:52 PM
Regards,
RedBus.
Post by: Johnny B on October 02, 2005, 03:01:38 AM
"With an analog system limited to say 22KHz the sound you can hear is made of 5, 10, 15 and 20KHz. Very harmonic sound.
"With the aliasing you have 5,10,14.1, 19.1 and 20KHz. I would say that the aliasing is non musicaly related to the energy."
By golly, Dan, I think you are on to something there. So, this begs the question...What if you redesign the entire process so the harmonics are more even...rather than odd?...
That's always been one of the things cited to explain why people love their analogue gear...the pleasant distortion is more musical because it works well with what people find harmonically pleasing...Now wonder if digital sytems could be adjusted or redesigned so that harmonics where more even than odd...
I'm just doing a little green light thinking here...not at the red light stage...(That's the stage where people say "No, that won't work because...." )...The green light stage is where you just dream about things considered impossible to do as being possible...some people call it "Imagineering."
Anyway, let's think about that aliasing thing that Dan mentioned as giving us the nasty odd harmonics instead of the pleasant sounding good even harmonics...Now if any of this were true, what would you do to go about fixing it? How would you nail down that alaising bit to get it to behave itself in a more pleasant way?
The exercise here is not whether you are right or wrong...The exercise is merely to spur some ideas, perhaps some new, fresh ideas, that might be worth pursuing later on...
Post by: Jon Hodgson on October 02, 2005, 04:15:18 AM
| Johnny B wrote on Sun, 02 October 2005 08:01 |
Dan just said, "With an analog system limited to say 22KHz the sound you can hear is made of 5, 10, 15 and 20KHz. Very harmonic sound. "With the aliasing you have 5,10,14.1, 19.1 and 20KHz. I would say that the aliasing is non musicaly related to the energy." By golly, Dan, I think you are on to something there. So, this begs the question...What if you redesign the entire process so the harmonics are more even...rather than odd?... That's always been one of the things cited to explain why people love their analogue gear...the pleasant distortion is more musical because it works well with what people find harmonically pleasing...Now wonder if digital sytems could be adjusted or redesigned so that harmonics where more even than odd... I'm just doing a little green light thinking here...not at the red light stage...(That's the stage where people say "No, that won't work because...." )...The green light stage is where you just dream about things considered impossible to do as being possible...some people call it "Imagineering." Anyway, let's think about that aliasing thing that Dan mentioned as giving us the nasty odd harmonics instead of the pleasant sounding good even harmonics...Now if any of this were true, what would you do to go about fixing it? How would you nail down that alaising bit to get it to behave itself in a more pleasant way? The exercise here is not whether you are right or wrong...The exercise is merely to spur some ideas, perhaps some new, fresh ideas, that might be worth pursuing later on... |
There is nothing new in what Dan just said, it is basic sampling theorem.
Preventing those alias components is what the input filter on an ADC is all about.
Post by: Johnny B on October 02, 2005, 02:08:24 PM
Yeah, but he's talking about those harmonic aspects, isn't he? I could be wrong, I often am.
Post by: Jon Hodgson on October 02, 2005, 02:43:41 PM
| Johnny B wrote on Sun, 02 October 2005 19:08 |
Yeah, but he's talking about those harmonic aspects, isn't he? I could be wrong, I often am. |
He's saying that if you don't bandlimit your signal before you sample it then the harmonics will fold back into non-harmonic frequencies, and sound awful.
Nothing to see here folks, move along please...
Post by: Johnny B on October 02, 2005, 06:07:35 PM
Post by: bobkatz on October 02, 2005, 06:08:50 PM
| RedBus wrote on Thu, 29 September 2005 15:41 |
I only mentioned the aliasing as being an effect in the context of those users seeking a "grunge" sound effect for Guitar or Keyboard parts. There is no doubt that aliasing is horrible for ordinary audio signals, but the non-musical result can be utilized by some people and may be appropriate to some forms of composition (I'd prefer not to call it music!) Regards, RedBus. |
Aliasing is the worst-sounding kind of "grunge" you can imagine. Well, then again, some of the sampled cymbals you get out of synthesizers sound very aliased.
Prior to digital recording, most of the analog distortion that we heard was very harmonically related, shall we say "pleasant sounding grunge". The problem with aliasing is it is very foreign and ugly to the ear. Create your grunge the old fashioned way, please
BK
Post by: Johnny B on October 02, 2005, 09:31:17 PM
| bobkatz wrote on Sun, 02 October 2005 23:08 |
Aliasing is the worst-sounding kind of "grunge" you can imagine. Well, then again, some of the sampled cymbals you get out of synthesizers sound very aliased. Prior to digital recording, most of the analog distortion that we heard was very harmonically related, shall we say "pleasant sounding grunge". The problem with aliasing is it is very foreign and ugly to the ear. Create your grunge the old fashioned way, please BK |
Yes, that's right on the money.
Post by: jimmyjazz on October 04, 2005, 03:28:06 PM
Post by: C-J on October 04, 2005, 04:18:48 PM
| Johnny B wrote on Sun, 02 October 2005 10:01 |
Dan just said, "With an analog system limited to say 22KHz the sound you can hear is made of 5, 10, 15 and 20KHz. Very harmonic sound. "With the aliasing you have 5,10,14.1, 19.1 and 20KHz. I would say that the aliasing is non musicaly related to the energy." By golly, Dan, I think you are on to something there. So, this begs the question...What if you redesign the entire process so the harmonics are more even...rather than odd?... |
Dan is talking about harmonic overtones of the 5k when he mentions 10k (even), 15k (odd) and 20k (even). The aliased 14.1 and 19.1 are neither even, nor odd harmonics, but non-harmonics.
| Quote: |
That's always been one of the things cited to explain why people love their analogue gear...the pleasant distortion is more musical because it works well with what people find harmonically pleasing...Now wonder if digital sytems could be adjusted or redesigned so that harmonics where more even than odd... |
Well,,, if your heads on straight,
with today's equipment you record at 24-bits, and leave 'nuff headroom to completely avoid all unpleasant digital distortion.C.J.
Post by: Johnny B on October 07, 2005, 01:14:09 AM
I suppose I could look up the formulas for the 15th billionth time or see if I could dig up my excel spreadsheets which take them out to the trillionth harmonic...but I'm far too lazy to do that right now.
I can say, the calcs are not that difficult.
Post by: Jon Hodgson on October 07, 2005, 03:36:10 AM
| Johnny B wrote on Fri, 07 October 2005 06:14 |
Hmmm, who here is not conversant with the ins and outs of the harmonic overtone series?...both in the perfect form and the impefect well-tempered form..?. I suppose I could look up the formulas for the 15th billionth time or see if I could dig up my excel spreadsheets which take them out to the trillionth harmonic...but I'm far too lazy to do that right now. I can say, the calcs are not that difficult. |
And your point is??
Anyway it sounds like you are confusing harmonics with scales, not the same thing. Harmmonics are integer multiples of the fundamental, very easy calculations.
Post by: jimmyjazz on October 07, 2005, 10:02:14 AM
| Johnny B wrote on Fri, 07 October 2005 01:14 |
Hmmm, who here is not conversant with the ins and outs of the harmonic overtone series/ |
Who are you referring to?
Post by: Johnny B on October 07, 2005, 11:58:11 AM
| Jon Hodgson wrote on Fri, 07 October 2005 08:36 |
Anyway it sounds like you are confusing harmonics with scales, not the same thing. Harmmonics are integer multiples of the fundamental, very easy calculations. |
No, I'm not at all confused. I know exactly what the harmonic overtone series is about.
I fear, that many others do not. For example, many wrongly believe that the harmonic overtones are rather simple multiples with no decimal places, although the math is not complex, there are trailing decimals.
This all came up in response to Bob K's observation.
Post by: Jon Hodgson on October 07, 2005, 12:32:14 PM
| Johnny B wrote on Fri, 07 October 2005 16:58 | ||
No, I'm not at all confused. I know exactly what the harmonic overtone series is about. I fear, that many others do not. For example, many wrongly believe that the harmonic overtones are rather simple multiples with no decimal places, although the math is not complex, there are trailing decimals. This all came up in response to Bob K's observation. |
Bob K was referring to distortion. If you take a waveform and distort it, say you chop the top off, your fundamental remains the same but your harmonic content changes.
Those harmonics ARE integer multiples of the fundamental (remember Fourier).
Post by: Johnny B on October 07, 2005, 02:53:51 PM
I'll take it out to 16, simply because I don't have time for more.
Decimal Ratio; Nearest Multiple of 1.059; Corresponding Degree of Scale
1. 1; 1; fundamental
2. 1; 1; fundamental
3. 1.5; 1.494; fifth
4. 1; 1; fundamental
5. 1.25; 1.258; Major Third
6. 1.5; 1.494; fifth
7. 1.75; 1.774; minor seventh
8. 1; 1; octave
9. 1.125; 1.121; major second
10. 1.25; 1.258; major third
11. 1.375; 1.411; diminished fifth
12. 1.5; 1.494; fifth
13. 1.625; 1.582; minor sixth
14. 1.75; 1.774; minor seventh
15. 1.875; 1.879; major seventh
16. 1; 1; octave
Now Jon, if you'd like you can explain exactly how digital differs from analogue with respect to handling all these aspects of the harmonic overtone series, I'm all ears.
Post by: Jon Hodgson on October 07, 2005, 02:55:16 PM
| Johnny B wrote on Fri, 07 October 2005 19:53 |
Now Jon, if you'd like you can explain exactly how digital differs from analogue with respect to handling all these aspects of the harmonic overtone series, I'm all ears. |
It doesn't
Post by: Johnny B on October 07, 2005, 02:59:22 PM
Post by: Jon Hodgson on October 07, 2005, 03:09:30 PM
| Johnny B wrote on Fri, 07 October 2005 19:59 |
Well wait a minute, didn't Bob K. or somebdy say there were differences between analogue and digital with respect to the harmonics? |
Ok, lets establish something here...
The harmonics are integer multiples, those decimal values are scale values. That is why there is no "perfect" scale, you always get notes pairs where the harmonics generate dissonance.
This is completely unrelated to any question of analogue/digital.
Ok, so that bit is out of the way...
Now, what Bob and Dan before him were on about was aliasing frequencies. Somebody proposed that aliasing could be a cool effect, Dan then showed that because of the way frequencies fold back in aliasing, previously harmonic components end up enharmonic, and sound like shit.
Bob basically agrees that aliasing distortion is nasty because it is enharmonic, whereas saturation style distortion is harmonic and therefore much nicer sounding (assymetric is usually nicer is this regard, since it generates even rather than odd).
But in a decent ADC system, all the frequencies which would alias are filtered out before conversion, so this is a non-issue anyway.
Post by: Johnny B on October 07, 2005, 03:50:27 PM
If the harmonics, whether or not they are folded back, are not handled correctly, then I could easily see all sorts of problems.
Imagine for a moment, if you could make the foldback more harmonically correct and pleasing...I dunno if anyone is already doing that....but I suspect they are taking a more linear approach and not producing the results that sound as good as they could...
For example, IIRC, I think Daniel Weiss may be using multiple bands and applying some extra DSP to great advantage...
I could be wrong, I often am.
Post by: jimmyjazz on October 07, 2005, 04:31:06 PM
| Johnny B wrote on Fri, 07 October 2005 15:50 |
Imagine for a moment, if you could make the foldback more harmonically correct and pleasing... |
You'd have to vary something, right? The notes are fixed. Their harmonic series are fixed. The sampling rate is fixed. What are you going to vary in order to change the manner in which (unfiltered) harmonics supposedly alias in a more pleasing manner?
And what problem would you be "fixing", anyway? There is no aliasing, because we filter out all harmonics at or above the frequency which is less than half the sampling rate.
Post by: Jon Hodgson on October 07, 2005, 06:20:04 PM
| Johnny B wrote on Fri, 07 October 2005 20:50 |
Yeah, I'm familiar with the foldback problem...and the question remains whether more realistic harmonics, which are a little more messy than overly simplistic ratios, could be used somehow to advantage. |
I guess I must be talking to myself.
Sit down, and re-read this sentence until it sinks in...
HARMONICS ARE INTEGER MULTIPLES OF THE FUNDAMENTAL, THIS INTEGRAL TO THEIR DEFINITION AS HARMONICS.
Take a sine wave, peak level 1.
Now add another sine wave, starting at the same point, twice the frequency, peak level 1/2
Now another, 3 times the frequency of the first peak level 1/3
and so on, add infinitum, you will get a sawtooth.
If your supposed harmonic was anything other than an integer ratio, then you couldn't fit an integer number of cycles into the fundamental cycle, they wouldn't synch up for several cycles... you wouldn't have a single waveform, you'd have a moving timbral content.
| Johnny B wrote on Fri, 07 October 2005 20:50 |
If the harmonics, whether or not they are folded back, are not handled correctly, then I could easily see all sorts of problems. Imagine for a moment, if you could make the foldback more harmonically correct and pleasing...I dunno if anyone is already doing that....but I suspect they are taking a more linear approach and not producing the results that sound as good as they could... For example, IIRC, I think Daniel Weiss may be using multiple bands and applying some extra DSP to great advantage... I could be wrong, I often am. |
Ok, now try to take this one in...
IN A PROPERLY DESIGNED SYSTEM THERE IS NO FOLDBACK OF FREQUENCIES
That's what those anti aliasing filters on the input are all about.
You think you're in a position to motivate people to improve the quality of digital audio, when you haven't even grasped basic Fourier, sampling theory 101, or the present state of converters?
As Confucius said..
"Learning without thought is labour lost,
Thought without learning is perilous"
Right now, you're dangerous
Post by: Jon Hodgson on October 07, 2005, 07:20:20 PM
that list you posted of the "harmonic overtones" appears to be an attempt to calculate the equal temperament scale.
If so, then you're guilty of far worse rounding errors than you're accusing everyone else of.
The multiplier is not 1.059, it should be 1.0594630943593
(to 13 decimal places). If you use that value then you'll find you can at least resolve to an octave correctly (well to 11 decimal places anyway).
Congratulations on demonstrating how recursive formulae will often require higher precision otherwise errors accumulate, if you calculate your scale intervals this way, you do need more than 32 bit floats.
Still doesn't make them harmonics though
Post by: Johnny B on October 07, 2005, 11:46:55 PM
Analogue does not mess around with this in a displeasing manner, but I do not believe the same thing can be said for digital.
And when you look at some of the overtone series, you see what some might call "dissonant" freqs, that is to say, freqs. that are right next to each other...this occurs in the uppers...but take something like a tube amp with iron transformers...the pleasing distortion is often said to emphasize some of the "lower" or "even" harmonics....
I know there are digital models and so-called "so-called" emulators, but they do not behave the same way as the real deal analogue device...in any event, it might be possible to put the vagaries and real world behavior of the overtone series to better use...
Maybe "String Theory" will provide some anwers for Confucius. And he never fly off the ground, did he?
Post by: Jon Hodgson on October 08, 2005, 04:45:12 AM
| Johnny B wrote on Sat, 08 October 2005 04:46 |
Ah, you're getting the idea I think...Yes, they DO go many places out...I did an abbreviated form for a quick post...perhaps like maybe some others make the mistake of doing in code and so on....and yes, the frequencies are not steady state and perfectly linear...they interact and, for lack of a better term, "wander" in real time. Analogue does not mess around with this in a displeasing manner, but I do not believe the same thing can be said for digital. And when you look at some of the overtone series, you see what some might call "dissonant" freqs, that is to say, freqs. that are right next to each other...this occurs in the uppers...but take something like a tube amp with iron transformers...the pleasing distortion is often said to emphasize some of the "lower" or "even" harmonics.... I know there are digital models and so-called "so-called" emulators, but they do not behave the same way as the real deal analogue device...in any event, it might be possible to put the vagaries and real world behavior of the overtone series to better use... Maybe "String Theory" will provide some anwers for Confucius. And he never fly off the ground, did he? |
Johnny,
This whole post is complete rubbish from start to finish, starting with misconceptions and coming up with flawed hypothesis
You don't even know the basics, you're like a caveman pontificating on the next generation of particle accelerator.
If you want to contribute anything to the future of digital audio other than irritation then PLEASE go back to first principles and learn what the hell you're talking about.
Right now all you're doing is making yourself look a fool to those who understand, and confusing those who don't.
Post by: Johnny B on October 08, 2005, 12:54:05 PM
You are certainly entitled to your world view, however, going along with all the old thinking you seem to accept as gospel would have prevented man from going to the Moon and coming back. Sometimes it's: "Out with the old, in with the new."
Post by: Jon Hodgson on October 08, 2005, 01:14:40 PM
| Johnny B wrote on Sat, 08 October 2005 17:54 |
Jon, You are certainly entitled to your world view, however, going along with all the old thinking you seem to accept as gospel would have prevented man from going to the Moon and coming back. Sometimes it's: "Out with the old, in with the new." |
On the contrary, the moon landings were the result of over four hundred years of mathematics, physics and chemistry, combined with high quality engineering. Yes along the way people had great ideas, sometimes evolutionary, sometimes revolutionary, but always based on the knowledge that went before... even those revolutionary ideas that proved previous theories incorrect.
If everyone thought like you, we'd still be dreaming about the possbility of going to the moon, whilst sitting in the mud and probably imagining the moon was made of cream cheese... if indeed we had got as far as inventing cheese.
If you want to pick holes in sampling theory, signal processing algorithms, or anything else associated with digital audio, you first need to understand what it is you're trying to contest.
Post by: Johnny B on October 08, 2005, 01:38:20 PM
How about the way it sounds, they way it behaves, and the way it performs for starters?
C'mon, you know it can be improved.
And chances are, it will be improved.
I also suspect that some people are upset because the old mic and speaker argument is rapidly fading away...
If they want to argue against greater bit-depth and higher sample rates they will no longer be able to cloud the issue with mics and speakers...100kHz mics and speakers appear to be already be here...and that means sample rates of a min. of 200kHz...right?
The focus will most likely become how to make the filters work well to pass the increased frequency band and how to employ better DSP, but I'm just guessing, because as you've pointed out on numerous occassions, I'm the idiot around here.
Cheers.
*
Post by: Jon Hodgson on October 08, 2005, 01:57:54 PM
| Johnny B wrote on Sat, 08 October 2005 18:38 |
How about the way it sounds, they way it behaves, and the way it performs for starters? |
Define your complaints
Specify what it is you hear that you don't like.
What is it about the behaviour you consider inferior to your reference system?
What is your complaint about performance?
| Johnny B wrote on Sat, 08 October 2005 18:38 |
C'mon, you know it can be improved. And chances are, it will be improved. I also suspect that some people are upset because the old mic and speaker argument is rapidly fading away... If they want to argue against greater bit-depth and higher sample rates they will no longer be able to cloud the issue with mics and speakers...100kHz mics and speakers appear to be already be here...and that means sample rates of a min. of 200kHz...right? The focus will most likely become how to make the filters work well to pass the increased frequency band and how to employ better DSP, but I'm just guessing, because as you've pointed out on numerous occassions, I'm the idiot around here. Cheers. * |
But I thought that your aim was to make it sound "as good as analogue" ? (whatever the hell that means).
Since your analogues systems don't have these sorts of bandwidths, if they do produce any improvements, then they won't be ones that take it closer to the goal you hold so dear.
As for bit-depths, at the converter end, none of these new transducers makes one iota of difference, we're already down to the limitations imposed by johnson noise... that's the random movement of electrons that occurs at any temperature above 0 Kelvin, during processing, the situation is no different to what it was before, some places it helps, some it doesn't.
You seem to think that people in my position have some kind of vested interest in keeping the status quo the same, actually the opposite is the case. As long as new technologies and algorithms keep being developed, I get work implementing them. It's better for my wallet, and anyway it's far more interesting for me, if all I cared about was the money I'd switch to programming for the financial markets.
Post by: Johnny B on October 08, 2005, 03:45:48 PM
I'm glad you do not want to hold the tech back, I fear there are some who do.
Post by: Jon Hodgson on October 08, 2005, 03:57:56 PM
| Johnny B wrote on Sat, 08 October 2005 20:45 |
I'm glad you do not want to hold the tech back, I fear there are some who do. |
Just because people who actually know about the subject disagree with you about what improvements can be made and what the paths might be to make them, and point out negative aspects in your proposals doesn't mean that they want to hold the techonology back.
It might actually mean that with their years of study, their daily real world experience, and the many tests and experiments they have made looking for how to improve things, that they know more than you.
Post by: Johnny B on October 08, 2005, 07:20:58 PM
| Jon Hodgson wrote on Sat, 08 October 2005 20:57 |
[P]eople who actually know about the subject disagree...[about] what improvements can be made and what the paths might be to make them... |
Such as? I have not seen many brilliant ideas come from the stuck-in-the-past quarters, in fact, not only do they resist more research, they seem to stubbornly resist any change at all.
I'll here and now issue this challenge:
I dare anyone who comes here to post one or more ideas on how to improve digital, either its performance or its sound quality.
It's big task, let's see what you come with Jon.
Post by: Jon Hodgson on October 08, 2005, 07:31:19 PM
| Johnny B wrote on Sun, 09 October 2005 00:20 |
Such as? I have not seen many brilliant ideas come from the stuck-in-the-past quarters, in fact, not only do they resist more research, they seem to stubbornly resist any change at all. I'll here and now issue this challenge: I dare anyone who comes here to post one or more ideas on how to improve digital, either its performance or its sound quality. It's big task, let's see what you come with Jon. |
Firstly, fix that quote, it is incorrect and misleading... I assume it's simply an error on your part composing that post.
Secondly you don't understand the present state of audio, digital or analogue, how are you going to undestand the improvements?
Thirdly, here is my challenge to you...
You come up with one actual real problem with digital audio, something you can point to and explain and quantify, beyond your repeated claims that digital audio as a whole, as opposed to the possibly poorly implemented systems you've heard, sounds bad, without a single bit of evidence that it is anything other than your own prejudice and imagination, and I will do my best to explain to you what the problem is and how to fix it.
So far I've heard nothing substantial from you, only a bunch of boundless claims, incorrect assumptions, innacurate 'facts' and wild uneducated hypothesis.
Post by: Johnny B on October 09, 2005, 12:13:34 AM
Maybe we can have this conversation again in 5 years...
Maybe we can stay civil and avoid the name-calling..
Maybe based on innovations and changes we see in that 5-year period we might even agree on some points...
Anyway, I'm outta this thread.
See ya all in 5 years.
Cheers
Post by: Jon Hodgson on October 09, 2005, 06:10:02 AM
However you have posed a question that I think people reading may be interested in an answer to, even if you probably won't like it, because it's not "revolutionary" enough.
So my educated but by no means infalible or definitive hopes and guesses on where improvements will be made.
Nyquist and Fourier appear to be, according to overwhelming evidence, fact. They're mathematical formulae which can be proven (it's all in the calculus) and they've been in daily use in the communications industry for decades. With a few hundred million phones making calls every day, and NASA communicating with probes at the end of the solar system, if they were wrong then someone should have noticed by now.
So unless someone makes some discovery in physics which shows that the music hitting the transducer can make those electrons do anything more interesting than sway back and forth in time with the music, which would require capture of more than the two dimensions (level and time) we presently require, we can stick with the sampling theorem as it stands... and I don't personally expect this discovery, it seems very unlikely considering what we know about sound waves in air that there is anything else to capture as it is.
Converters are mostly an analogue design problem, so hopefully these will continue incremental improvement. Hopefully the marketing guys and the peoplw who respond to their spin and pseudo-science won't take over when it comes to bit depths, at 24 bit conversions for audio we're already below any noise floor you'll ever get from the analogue stages. I think we'll progress much further if we concentrate on perfecting the bits we have, rather than adding random values on the bottom. If you want a 32 bit converter you can effectively do it now, take a 24 bit input sample, shift it 8 to the left, and add an 8 bit random number on the bottom... the effect is the same.
As for sample rates, again hopefully good sense will win over marketing, though it doesn't seem like that is happening right now. Ideally sample rates should go as high as neccessary, but no higher. Why no higher? Isn't it better to have more rather than less? ... well quite simply it's a tradeoff, higher sample rates means lower precision in your samples. Good engineering can reduce the problem, but it can never eliminate it, electrons are not nicely behaved things, they wizz around and shuffle about and bump into each other all the time, all electronic circuits do make them move in a general direction, and then they take a while to start moving and when they get where they're supposed to be they mill around a bit finding a place... the less time you give them to sort themselves out, the more likely they'll not be settled when you sample them.
What is the sample rate required?
Well so far I haven't seen anythin compelling saying we need more than 20k bandwidth, and I'll take Dan Lavry's word for it (since this is his area of expertise) that sampling at 60kHz would allow us to get any filter artifacts well out of the audio band. We're already past that, so let's just stop at 96kHz.
Unless of course research actually shows that there is something higher up that is worth capturing... or more precisely something that has greater value than the loss of precision we would get across the whole spectrum due to the tougher time high rate converters have.
Is there anything worth capturing up there? Well as I understand it most research shows that we don't respond to steady state signals in those areas. From some people's anecdotal comments I did wonder if perhaps the ear might respond non-linearly to transients, I haven't seen any research on this (which probably means that it's been done and they found nothing interesting, but not neccessarily)... but if it was the case then perhaps you could do something interesting with the converters, capturing more samples on transients, and oversampling digitally for steady state signals.
If it was found that the higher frequencies were beneficial in steady state form (which I have to repeat seems unlikely at present), then capturing them without making things worse lower down is going to be tough, I did wonder about a dual band converter system, capturing lower frequency components with a lower rate (and more accurate) converter, and higher components with a high rate converter, and then recombining...
But those last two are just idle speculation on my part, mostly a mental exercize, what I've seen so far leads me to believe that 96kHz is more than enough, so we should just concentrate on perfecting it.
All the above refers to capture and playback, where your intention is achieve a "what goes in is what comes out" situation. People talk about "the sound" of analogue, or digital... recording systems should not have a sound, sure tape saturation can be cool, but surely it would be better if the producer could decide is he wants it, and exactly how much he wants? In my opinion recording systems should be totally transparent, and then the producers should be given the tools they need to manipulate the sound as much or as little and in any way they want.
Which leads us on to processing...
More powerful processing chips will make more complex algorithms commercially viable, and practically usable, but they don't in themselves make for any kind of "revolution". I could process 80 bit floating point 384kHz audio on a twenty year old 8 bit 4Mhz processor if I wanted (or even older and slower).... just don't expect it to be real time!!
Linear processes can gain from higher bit depths IN SOME PLACES. We should see more of this as time goes on.
Non Linear processes that introduce harmonic components will gain from oversampling internally to higher rates, since that keeps the components from folding back into the audio band. They can also gain from having more complex filters to keep those harmonics under control.
A lot of the interesting work comes from trying to emulate the complexities of analogue circuis, to replicate the pleasant sounding artifacts some of them create (such as tape saturation), more analysis of the circuits and the components will allow more accurate simulations, more powerful processors will make implementing them viable in the real world.
Generally speaking research into psycho acoustics will continue to generate new ideas, or refine old ones I believe. There is still some way to go on 3d sound for example.
So generally science and engineering will push the algorithms forward, refining those that exist, and adding new ones.
And personally, my WISH is that with the DSP power now going into just about every new domestic playback system, that the transmission format of audio could be made agnostic of the playback situation (2 speakers, 4 speakers, 16 speakers, headphones).
So, I'm afraid no revolution, just a lot of science, research, engineering and listening resulting in incremental improvements.
Post by: blueboy on October 09, 2005, 04:43:13 PM
JL
Post by: C-J on October 10, 2005, 07:56:36 AM
| Johnny B wrote on Sun, 09 October 2005 00:20 |
I'll here and now issue this challenge: I dare anyone who comes here to post one or more ideas on how to improve digital, either its performance or its sound quality. It's big task, let's see what you come with Jon. |
| Jon Hodgson wrote on Sun, 09 October 2005 02:31 |
...here is my challenge to you... You come up with one actual real problem with digital audio, something you can point to and explain and quantify, beyond your repeated claims that digital audio as a whole, as opposed to the possibly poorly implemented systems you've heard, sounds bad, without a single bit of evidence that it is anything other than your own prejudice and imagination, and I will do my best to explain to you what the problem is and how to fix it. |
| Johnny B wrote on Sun, 09 October 2005 07:13 |
Ha, ha...yeah...ok..whatever you say, Jon. Maybe we can have this conversation again in 5 years... Anyway, I'm outta this thread... See ya all in 5 years. |
C'mon Johnny,
First, you make a challenge, and when you're asked to define what you think should be improved, you quit the thread and postpone it for 5 years...
We readers can make only one conclusion out of this: you need five years to "paint yourself out of the corner"?!
C.J., Finland
Post by: bobkatz on October 11, 2005, 09:41:12 PM
| Johnny B wrote on Fri, 07 October 2005 14:59 |
Well wait a minute, didn't Bob K. or somebdy say there were differences between analogue and digital with respect to the harmonics? |
No, I did not. What I had to say had nothing to do with digital audio at all.
All I expressed was that the perceived differences between a high frequency sine, square and triangle wave can be explained by non-linear tweeters... If there's enough IM distortion generated by the higher harmonics you're going to hear problems in the audible band. Non-linear drivers make this whole business of proving whether you need extreme bandwidth very hard.
| Quote: |
And given the non-linear nature of the math, it seems that could easily create some problems for digital. Music is not linear. |
What you are arguing is not going to cause a problem with digital audio.
One thing that digital recorders can do better than analog recorders is reproduce a frequency accurately and without wow and flutter. The harmonic series, perfect thirds and fifths and all will be reproduced far more accurately in a digital recorder than an analog tape recorder. And A/D/A conversion in between won't affect the reproduction of any incoming frequency. If you put in 220.697 Hz you'll get that out, exactly and steadily.
Post by: Johnny B on October 12, 2005, 03:07:55 AM
| danlavry wrote on Thu, 29 September 2005 00:10 |
With an analog system limited to say 22KHz the sound you can hear is made of 5, 10, 15 and 20KHz. Very harmonic sound. With the aliasing you have 5,10,14.1, 19.1 and 20KHz. I would say that the aliasing is non musicaly related to the energy. |
Post by: Jon Hodgson on October 12, 2005, 04:29:56 AM
| danlavry wrote on Thu, 29 September 2005 00:10 | ||
I do not wish to argue with "new path" but let me point out that alaising takes frequencies and moves them to frequency locations that are "highly non musical" in the sense of thier relationship to the intended (non aliased) sound. A very simple example: Take a 5KHz sound with harmonics at 10KHz, 15KHz, 20KHz, 25KHz and 30KHz. Say you sampled at 44.1LHz so Nyquist is at 22.050KHz. The 25KHz is aliased to 19.1Khz and the 30KHz is now at 14.1KHz. With an analog system limited to say 22KHz the sound you can hear is made of 5, 10, 15 and 20KHz. Very harmonic sound. With the aliasing you have 5,10,14.1, 19.1 and 20KHz. I would say that the aliasing is non musicaly related to the energy. |
Johnny,
You have to look at what Dan said in the context of it being a reply to what Redbus said, and then it becomes a simple fourier series (harmonics) and nyquist (aliasing)... both of which you say you understand.
Redbus was suggesting that by sampling (or resampling) without first limiting the signal to the Nyquist limit, then the aliased components generated might sound cool.
Dan responded saying he doubted they would, because they would not be harmonically related to the original signal, and then explained why.
Take a 5 kHz waveform with harmonics at 10, 15, 20, 25, and 30 kHz (Note the integer ratio, these are the fundamental and 2nd to 5th harmonics).
Now on an analogue system bandlimited to 22kHz, the 4th and 5th harmonics will simply be lost, giving you components at 5, 10, 15 and 20 kHz
But on a digital system WITHOUT prefiltering to the Nyquist limit, those components will fold back.
He's done his calculations using a 44100 Hz sample clock,
So the 25 kHz component, which is 25 - 22.05 = 2.95 kHz above the Nyquist frequency now appears 2.95 kHz below the Nyquist frequency, so at 22.05 - 2.95 = 19.1 kHz
The 30 kHz component, using the same maths
30 - 22.05 = 7.95
22.05 - 7.95 = 14.1
(or you could just write the above as 44.1 - 25, and 44.1 - 30).
So in that digital system you'd have components at
5, 10, 14.1, 15, 19.1 and 20 kHz
Two of those aren't harmonically related to the fundamental.
But all of this is moot in a system which has proper filtering to the Nyquist limit.
Post by: Johnny B on October 12, 2005, 11:00:16 AM
Let's say, just for the purposes of discussion that Rupert Neve, David Blackmer, Professor Boyk and others are right...that some..maybe not all...of those higher frequencies turn out to be "important" to the human system, then what?
You'd most likely have to use new DSP and new filters to make sure that the 104kHz came thru in an appropriate manner, right?
Again, you would most likely be presented with some new engineering challenges, but those kinds of challenges are solved everyday by creative minds. Today, I find myself thinking about the recent discoveries in outer space which appear to challenge the Big Bang Theory and how it might relate to String Theory which attempts to present a unifying principle for the universe.
Of course, all these wonderful challenges and opportunities will undoubtedly be solved by far younger and far more creative minds than the limited capacity remaining in my small brain.
And while it is certainly true that analogue is the world class sound quality standard to which digital must eventually meet, I'm not at all certain that continuing to raise the technical deficiencies in analogue serves any useful purpose in the continuing quest to improve digital sound quality.
IOW, maybe it's more "important" for digital to capture and reproduce those higher frequencies than it's been for analogue. And maybe there will be some unexpected benefits from the push to expand digital's "important" frequency spectrum to 104kHz.
Clearly, more advanced research and more applied science using an appropriate multi-disciplinary approach is warranted.
Post by: Jon Hodgson on October 12, 2005, 11:35:50 AM
| Johnny B wrote on Wed, 12 October 2005 16:00 |
You'd most likely have to use new DSP and new filters to make sure that the 104kHz came thru in an appropriate manner, right? |
No, DSP works the same at higher sample rates, you just do more of it because you have more samples.
Higher sample rates are an analogue design problem, not a digital one.
| Johnny B wrote on Wed, 12 October 2005 16:00 |
And while it is certainly true that analogue is the world class sound quality standard to which digital must eventually meet, |
You really haven't been paying attention when people have pointed out what your analogue recorders, EQs and transformers actually do to the signal, have you?
| Johnny B wrote on Wed, 12 October 2005 16:00 |
I'm not at all certain that continuing to raise the technical deficiencies in analogue serves any useful purpose in the continuing quest to improve digital sound quality. |
Ah you have been paying attention.... well kindof
It does serve a useful purpose in illustrating...
1) That differences between the two do not neccessarily point to a failing in the digital recorder, even if the difference is pleasant. You can always add non linear processing (like tape saturation), you can never take it away.
2) That using analogue tape as a reference standard for recording quality is foolish, since it is so inherently flawed. Using analogue tape as a quality reference WILL hold back improvements, and so pointing out its deficiencies is relevant.
2) Lack of logic when people complain that digital doesn't sound "as good" as analogue, and then go on to deduce from this that digital needs improving in areas it has already surpassed analogue in order to match it. Rather like saying "diesel cars vibrate more on idle than petrol ones... so I think we need to make them faster".
| Johnny B wrote on Wed, 12 October 2005 16:00 |
IOW, maybe it's more "important" for digital to capture and reproduce those higher frequencies than it's been for analogue. |
And why would that be?
In what way do you think that adding signal components that you yourself don't appear to consider important to your listening experience (because I haven't seen you complaining vehemently that analogue recording is inadequate for your listening pleasure) could nullify the differences in lower bands that you feel do make it sound worse?
| Johnny B wrote on Wed, 12 October 2005 16:00 |
Clearly, more advanced research and more applied science using an appropriate multi-disciplinary approach is warranted. |
Once again you say this like it's not already being done.
Post by: Johnny B on October 14, 2005, 11:12:10 AM
| Jon Hodgson wrote on Wed, 12 October 2005 16:35 |
Ah you have been paying attention.... |
Thanks for your kind words.
| Jon Hodgson wrote on Wed, 12 October 2005 16:35 |
Higher sample rates are an analogue design problem, not a digital one. |
Not sure if this statement is totally accurate, we now have 100kHz mics and speakers. And, if you ever want to capture and reproduce the 104kHz that Cal Tech's Professor Boyk measured you are looking at a minimum of 208kHz sample rates. Using Dan's conservative design philosophy, you'd need to go a little higher...If you want smooth filters, maybe higher still...so sample rates are indeed a very real digital problem...
It's really both an analogue and DIGITAL problem, is it not?
It's also a total "HUMAN SYSTEM" problem because "How much of the frequency spectrum is "IMPORTANT" for digital to capture and reproduce for the Human System to experience pleasure?" is a very debatable point.
There are some pretty reputable people arguing for those higher rates simply because they want the "Human System" to experience that expanded frequency spectrum, they feel that an expanded frequency system is very important.
For example, the engineers at NHK Science and Technical Research Laboratories say this:
"100kHz microphones do exist in the field of acoustic measurement but the Sanken CO-100k is the first such device for use in the professional recording world. Designed in conjunction with the NHK Science and Technical Research Laboratories, this microphone is Sanken's solution for those exceptional engineers who want to record a complete sonic picture from 20Hz to 100kHz."
And if George Massenburg is a big fan of this mic, well then, what can I say, I'll just have to check this 100kHz mic out for myself.
Post by: Jon Hodgson on October 14, 2005, 11:54:53 AM
| Johnny B wrote on Fri, 14 October 2005 16:12 | ||
Not sure if this statement is totally accurate |
To make an accurate converter you need certain things...
a very stable clock...
that's analogue
very clean input amplifiers...
they're analogue...
a comparator of some kind with very equal quantization steps...
that's analogue...
very smooth power supply...
that's analogue...
High quality filters...
the difficult ones are analogue
In fact all the really difficult bits are analogue, and they all get harder and harder as the clock speed goes up.
Post by: Jon Hodgson on October 14, 2005, 12:05:55 PM
| Johnny B wrote on Fri, 14 October 2005 16:12 |
It's also a total "HUMAN SYSTEM" problem because "How much of the frequency spectrum is "IMPORTANT" for digital to capture and reproduce for the Human System to experience pleasure?" is a very debatable point. |
Take "digital" out of that paragraph and replace it with "an audio system" and we might agree on something. You should also either add a "maximum" between "experience" and "pleaure", or remove the "very" before debatable, since people have been deriving great pleasure from limited bandwidth recordings for the past 100 years.
Post by: Jon Hodgson on October 14, 2005, 12:12:18 PM
| Johnny B wrote on Fri, 14 October 2005 16:12 |
For example, the engineers at NHK Science and Technical Research Laboratories say this: "100kHz microphones do exist in the field of acoustic measurement but the Sanken CO-100k is the first such device for use in the professional recording world. Designed in conjunction with the NHK Science and Technical Research Laboratories, this microphone is Sanken's solution for those exceptional engineers who want to record a complete sonic picture from 20Hz to 100kHz." And if George Massenburg is a big fan of this mic, well then, what can I say, I'll just have to check this 100kHz mic out for myself. |
If George likes the mic, then it's probably a great sounding mic, though that doesn't neccessarily mean he can actually hear anything over 20 kHz.
But as for that first bit... it just screams out "written by someone in marketing", don't attribute it to engineers, especially the customer masterbatory line "for those exceptional engineers". In fact that whole Sanken product sheet has very obviously had the marketing department all over it.
Post by: Ronny on October 14, 2005, 12:54:45 PM
| Jon Hodgson wrote on Fri, 14 October 2005 12:12 | ||
If George likes the mic, then it's probably a great sounding mic, though that doesn't neccessarily mean he can actually hear anything over 20 kHz. But as for that first bit... it just screams out "written by someone in marketing", don't attribute it to engineers, especially the customer masterbatory line "for those exceptional engineers". In fact that whole Sanken product sheet has very obviously had the marketing department all over it. |
Exactly.
Post by: Johnny B on October 14, 2005, 02:49:02 PM
As for GM, IIRC, he has long complained about the unduly constrained sound of lower sample rates...I think he might even be one of those "exceptional" people that the Sanken quote says this 100kkHz mic is attempting to satisfy. In fact, IIRC it was thru George that I first learned about Professor Boyk's 104kHz findings. I do know that GM likes to hear the cymbals correctly and that he's had some problems with slow 96kHz digital in the past. Exactly what GM's current thnking or feelings are will have to await for the time when he can write the words himself.
Now if any of you has an AP test box handy, maybe you could try to put the lie to this so-called marketing hype. If not, then you are condeming something without having neither scientific evidence nor any actual experience which makes those comments nothing but personal opinions based on a blind faith belief system that is now under attack. But if you get the mic and properly test it, you just might find that the following Sanken comment rings true:
"100kHz microphones do exist in the field of acoustic measurement but the Sanken CO-100k is the first such device for use in the professional recording world. Designed in conjunction with the NHK Science and Technical Research Laboratories, this microphone is Sanken's solution for those exceptional engineers who want to record a complete sonic picture from 20Hz to 100kHz."
Post by: jimmyjazz on October 14, 2005, 03:01:41 PM
Post by: dcollins on October 14, 2005, 03:50:37 PM
| jimmyjazz wrote on Fri, 14 October 2005 12:01 |
It's so far beyond "disingenuous" that I wouldn't even know what to call it. |
How about "stupid?" A waste of everyones time? JH deserves some kind of prize for trying, in vain, to explain stuff to Johnny B.
But alas, when it's a mental brick-wall........
There's always the "ignore all posts from this user button"
DC
Post by: Jon Hodgson on October 14, 2005, 05:16:51 PM
| Ronny wrote on Fri, 14 October 2005 17:54 | ||
Exactly. |
Actually my favourite bit is...
NOTE: energy of more than 20kHz especially near
100kHz decrease unexpectedly and severely,
therefore, in order to record music sound,
frequency 20kHz to 100kHz, the mic has peak at
20kHz to 70kHz on axis.
Talk about marketing waffle! That ranks along side the text you used to get on some pre-recorded cassettes which said something along the lines of "To provide a convenient pause in the program, one side of this album is shorter than the other" ... which translates as "We couldn't sort the songs out in a way that both sides matched, but the guy in marketing thought you couldn't handle the truth, so we'll pretend it's for your benefit"
Personally I can imagine how the conversation between marketing and engineering may well have gone...
"So, have you finished that new Mic yet?"
"Yeah we have, have a listen to this"
<marketing guy probably couldn't tell a DPA from a something that comes free with a Creative sound card, but he knows he has to humour the engineer>
"That sounds really nice"
"Yeah, we gave it a 10 dB boost at 20kHz, that really makes it sound bright and airy"
<marketing guys eyes start to glaze over>
"Really, that's good"
"Yeah, and we kept it smooth at 20k by actually making it peak between 30 and 40k"
<marketing guy fights boredom but continues to feign interest>
"That's quite high isn't it?"
"Yeah, actually the mic responds up to 100k, but who cares apart from vampires?"
<makreting guy now has a glint of interest in his eye>
"100k? That's really high isn't it?"
"Well yes, but that's not really the point, the point is that by giving it a p...."
"Does any other mic go up to 100k?"
"Well there are measurement mics that do, but I don't know of any music recording mics that do, it's not really that important th..."
"Do people want 100k mics?"
"Well I guess so, some people think they can hear stuff up there, but act...."
"Great!"
<marketing guy rushes off to write marketing blurb>
And I'd guess that anyone who's ever worked in the engineering department of a company that also had marketing and sales departments has probably had that conversation!
Post by: Jon Hodgson on October 14, 2005, 05:22:52 PM
I have the greatest respect for the engineers who designed this microphone, they have created something which goes beyond typical boundaries in frequency response, whatever their reason for doing so. Fine engineering always gets my personal and professional respect.
I also think that there's every chance this microphone sounds great.
But neither of these facts in itself gives me any reason to believe that people can hear to 100k, or even over 20k
Post by: Johnny B on October 14, 2005, 08:12:34 PM
DC, I'm considering giving KK Proffitt a shout to get her beat you down and put you in your place again. She tore you up before, she *could* do it again.
My advice is that a little courtesy, even when people disagree, is always the best policy.
Glad Jon, at least, recognizes the talent of the NHK people.
Post by: Johnny B on October 14, 2005, 08:27:57 PM
I do understand that some resist speeds like 192kHz, but no longer can mics and speakers be used to defend that position. That's over.
Now if people want to defend slower rates like 96kHz, they will have to stick to the real issues and cannot use the "mic and speaker argument" as a lame fallback position.
I could be wrong tho', I often am.
Post by: kraster on October 14, 2005, 09:22:00 PM
| Johnny B wrote on Sat, 15 October 2005 01:27 |
Oh and BTW, even if the new Sanken Mic effectively only goes to 70kHz, that means you'd need, at a minimum, SRC's cranking along at 140kHz. Of course, these frequencies cannot be captured with slower SRC's working at only 96kHz. You'd probably have to use the currently popular 192kHz chips, right? I do understand that some resist speeds like 192kHz, but no longer can mics and speakers be used to defend that position. That's over. Now if people want to defend slower rates like 96kHz, they will have to stick to the real issues and cannot use the "mic and speaker argument" as a lame fallback position. I could be wrong tho', I often am. |
So are you suggesting we ditch all our U47s and c12s in favour of this marketing nonsense?
I think, contrary to your position, that most people here will have a great deal of justifiable skepticism when confronted with this marketing blurb.
The 'real issues' are that the vast majority of Mics and speakers are limited from 20hz to 20khz and that's were most people want to hear audio gear performing at its greatest potential.
This isn't the first ultrasonic mic ever produced and the previous ones didn't exactly set the world on fire and people stayed away in their droves because they could smell a rat.
I cannot grasp the fact that when you have some exceptionally learned minds in the audio world telling you from their vast experience and knowledge what the facts are you will not accept it. And the way that you imply that these learned people have no interest in improving the way in which we capture music is plain ludicrous. They just know not to follow a dead end.
Karl
Post by: Johnny B on October 14, 2005, 10:57:04 PM
I suggest you try things before commenting. And never forget, many things we once thought was gospel is no longer accepted.
Post by: Jon Hodgson on October 15, 2005, 04:43:03 AM
| Johnny B wrote on Sat, 15 October 2005 01:27 |
Oh and BTW, even if the new Sanken Mic effectively only goes to 70kHz, that means you'd need, at a minimum, SRC's cranking along at 140kHz. Of course, these frequencies cannot be captured with slower SRC's working at only 96kHz. You'd probably have to use the currently popular 192kHz chips, right? |
Firstly I haven't seen anybody claim it only goes to 70kHz
Secondly I think just about anyone reading this is aware of the Nyquist theorem.
Thirdly just because the mic can capture those frequencies doesn't mean we have to. Just look at the response curve, it's almost certainly been designed to make the top end of the audible range brighter, whether that was the original intent and the 100 kHz was simply as side effect that marketing got hold of, or whether it's a more cynical action of giving it that boost so people hear it and go "Wow, it's got a lot more top end, I really can hear over 20 kHz", when in fact what they're hearing is that lift in the top end (about 10dB at 20kHz), I don't know... but one thing this microphone has almost certainly NOT been designed for is perfect unadulterated capture of the sound.
| Johnny B wrote on Sat, 15 October 2005 01:27 |
Now if people want to defend slower rates like 96kHz, they will have to stick to the real issues and cannot use the "mic and speaker argument" as a lame fallback position. |
No, what is REALLY a lame argument is repeatedly pushing the idea that the way to "improve" an audio recording system (read "make it sound more like another system which is proven highly imperfect"), is to extend its performance in areas which are already far beyond that of the reference.
So, if you feel that a C12 recorded onto a Studer sounds better than a C12 recorded onto a RADAR then you won't find the solution in super sample rates, because the C12 isn't generating frequencies that require them.
But then that's simple logic, which seems to elude some people.
Higher sample rates good? - possibly, but unlikely going by the weight of scientific evidence.
Higher sample rates make it sound more like an analogue recorder?
NO WAY
Post by: Jon Hodgson on October 15, 2005, 04:54:24 AM
| Johnny B wrote on Sat, 15 October 2005 03:57 |
Oh, so now you are calling George Massenburg names...he reportedly loves this new Sanken mic. |
No he's not, he didn't say anything about George M's reasons for his opinion of the Mic.
Personally I hope George Massenburg likes the microphone because it is a good sounding microphone in the range that he can hear, and not because he responds to the kind of marketing blowjob we can read on that product sheet.
| Johnny B wrote on Sat, 15 October 2005 03:57 |
And you are also deriding the NHK engineers and their staff... |
No, just the marketing guys.
| Johnny B wrote on Sat, 15 October 2005 03:57 |
I suggest you try things before commenting. And never forget, many things we once thought was gospel is no longer accepted. |
And when did any of those steps forward come from applying LESS science?
Post by: Johnny B on October 15, 2005, 12:04:33 PM
| Jon Hodgson wrote on Sat, 15 October 2005 09:43 |
Higher sample rates good? - possibly, but unlikely going by the weight of scientific evidence. |
No, there is no good hard science to back that claim up, all there is are some math theories, which, incidentally happens to correspond to certain box makers marketing positions, and I mean no disrespect to this position, it's one view, but there are other views and other avenues which must be thoroughly explored and investigated. I suppose their might be economic and emotional reactionaries who seemingly do not want to actually test alternative rates and seem mono-centric in their focus on old ear studies rather than looking at the more holistic human being's entire system. If the needle is wiggling on those brain scans or if the heart and pulse rate monitors are showing changes, then relying solely on old ear studies may be inadequate in light of newer discoveries. I'm in the camp that believes that the old 20-to-20 myth is now dead, but "How much of the frequency spectrum is truly "Important" to the entire human system?" is a legitimate topic for further debate and further study.
And clearly, if the spectrum gets expanded as some would like, it will mean the sample rates will need some adjustment. I think there's your "logic."
I also know I'm not truly happy with the current sound of digital systems, there's a lot I'm not happy about in the digital world, which includes the formats. I suppose there are some who are very happy with the state of digital, perhaps they are doing rap and hip hop for example, but there is quite a large group of people who are not totally satisfied with digital.
For example, here's a post from GM's board from April 2004.
=================================================
"George Massenburg wrote on Tue, 27 April 2004 03:25
"Johnny B wrote on Fri, 23 April 2004 18:10
"Well you can answer if you want, but this is just my way of saying "Hello" to GM. I hope this will be a more serious adventure over here. I'll keep my fingers crossed. Good luck with this.
"BTW, do you know Walter Sear?
===================================================
And George answers:
"Hi JohnnyB, and welcome over...
"No, I don't know Walter Sear well. I know I must have met him, and I sure like what he's done with his studio. It's not what I'd do but that's even better.
"George
"p.s. I have no idea who makes the best A/D/A's...still looking."
===================================================
I do not know if GM has changed his opinion since April, 27, 2004 or if he's "still looking?" But I think it shows that digital has not come as far as it *should* or as far as some would like us to believe with THEIR marketing spin.
Let's just say, "Digital ain't perfect," I think we can all agree on that point.
Post by: kraster on October 15, 2005, 12:37:53 PM
| Johnny B wrote on Sat, 15 October 2005 03:57 |
Oh, so now you are calling George Massenburg names...he reportedly loves this new Sanken mic. And you are also deriding the NHK engineers and their staff... I suggest you try things before commenting. And never forget, many things we once thought was gospel is no longer accepted. |
Johnny B,
Don't put words in my mouth. I never even referred to Mr. Massenburg in my post and didn't even know his thoughts on his mic.
I have been burnt by marketing spin before and think it only prudent to be cynical on claims that are conjured up with the express purpose of selling a product.
I would never buy a mic on the pretext of how it performs at 100khz or 70khz. I would buy a mic on how it sounds in the Audible range.
You seem to forget that many things we once thought as gospel still ARE. You are very selective in your use of science to back up your statements. I too have read Boyk's findings and the research done by Tsutomu Oohashi on ultrasonics. I find it compelling reading but to a large extent impractical and inconclusive. The well established science of filter design, Jitter correction, error correction and application of Dither have much more of a bearing on how we hear what we're trying to capture.
Post by: Jon Hodgson on October 15, 2005, 12:50:35 PM
| Johnny B wrote on Sat, 15 October 2005 17:04 |
No, there is no good hard science to back that claim up, all there is are some math theories |
Your repeated demonstration of your ignorance of how all this works is depressing.
The maths of the sampling theorem and fourier say nothing about what we can hear, they simply state, WITH 100% PROVABLE CERTAINTY, how we can perfectly replicate a signal of given bandwidth.
If you can prove that 1+1 is not equal to 2, or that the integral of a sine wave is not a cosine wave, then you may be able to disprove Fourier or Nyquist.
The theories are to do with what people can hear or perceive in other ways, there are several, including those you support.
The science of what people can hear is the subject of a great deal of research, and so far although there is evidence that things happening at high frequencies can affect our perception of audio, I have yet to see anything that shows conclusively, or even probably, that this perception is not due to effects that occur in lower frequencies (for example non-linearities in air or speakers causing audible difference frequencies) and therefore cannot be captured by a transducer and recording system with a 20k bandwidth
For every one test which had indicated that there may be perception above 20kHz, there are at least 10 which indicate there is not, so whatever the truth may be, I can stand by my statement that the WEIGHT of evidence is in favour of the 20k limit.
| Johnny B wrote on Sat, 15 October 2005 17:04 |
, which, incidentally happens to correspond to certain box makers marketing positions, and I mean no disrespect to this position, it's one view, but there are other views and other avenues which must be thoroughly explored and investigated. I suppose their might be economic and emotional reactionaries who seemingly do not want to actually test alternative rates |
So you act like people are insulting engineers when they point out that a product sheet which says things you like was obviously written by marketing people, but then having first said you mean no disrepect you immediately go on to imply that the professionalism and integrity of a wealth of scientists and engineers is in question when they say things you don't?
Oh and get over yourself...
A lot of them have already tested the rates.
| Johnny B wrote on Sat, 15 October 2005 17:04 |
and seem mono-centric in their focus on old ear studies rather than looking at the more holistic human being's entire system. If the needle is wiggling on those brain scans or if the heart and pulse rate monitors are showing changes, then relying solely on old ear studies may be inadequate in light of newer discoveries. I'm in the camp that believes that the old 20-to-20 myth is now dead, but "How much of the frequency spectrum is truly "Important" to the entire human system?" is a legitimate topic for further debate and further study. |
So I can assume from this statement that you are planning on ditching your analogue systems, which don't get up (or down) to these frequencies and which nobody is planning on upgrading?
| Johnny B wrote on Sat, 15 October 2005 17:04 |
I also know I'm not truly happy with the current sound of digital systems, |
Oh, nope, sorry, take back my last sentence. It seems I can't assume that you would follow your own reasoning through to its logical conclusion, I should have learned by now.
| Johnny B wrote on Sat, 15 October 2005 17:04 |
there's a lot I'm not happy about in the digital world, which includes the formats. |
You say this like you actually know something about the implications of a particular format are... but you've yet to show that you do.
| Johnny B wrote on Sat, 15 October 2005 17:04 |
I suppose there are some who are very happy with the state of digital, perhaps they are doing rap and hip hop, |
Actually there are huge swathes of people who are quite happy with digital, or at the very least find it equal or superior to their experience of analogue.
As for the second part, I can't stand most rap and hip hop myself, but I'm not going to assume that those who like it have worse hearing or care less about sound than I do.
| Johnny B wrote on Sat, 15 October 2005 17:04 |
for example, but there is quite a large group of people who are not totally satisfied with digital. |
Quite true... I am one of that group, and I think I speak for the majority of that group when I say that your contributions to the debate and the future improvements of audio in general, including digital, have so far ranked alongside the chocolate teapot in usefulness.
Post by: AndreasN on October 15, 2005, 01:23:27 PM
JB wrote:
>Let's just say, "Digital ain't perfect," I think we can all agree on that point.
Haven't seen anyone arguing against that notion? Several of the posters have stated N million times that there is indeed room for refinement in digital audio.
The difference is that they have an informed idea of where those improvements may be made. In order to make progress you need to know your current status. How can you expect to push digital audio ahead when you don't know what it is?
| Johnny B wrote on Sat, 15 October 2005 18:04 | ||
No, there is no good hard science to back that claim up, all there is are some math theories, which, incidentally happens to correspond to certain box makers marketing positions, and I mean no disrespect to this position, it's one view, but there are other views and other avenues which must be thoroughly explored and investigated. |
You're taking this a bit too far now. Those 'math theories' you totally ignore *ARE* digital audio. No less.
This is a scientific forum. The sticky thread specifically states that:
'This forum is about the why and how of audio technology. It is about the electronic and electromechanical bridge between musician and listener.' 'This forum is not about subjective opinions.'
Most importantly: 'This forum is not about the mechanics of the ear and brain.'
You have now spent months and hundreds of post making one single point. Which is based on vague theories on our perceptions, claiming the importance of hearing beyond 20khz. I feel this strongly belong to the 'mechanics of the ear and brain' department.
Andreas Nordenstam
EDIT: Sorry! This was very harsh. I hope you do not get personally offended. It's your bold statements that I'm adressing, not your personal being.
Post by: Johnny B on October 15, 2005, 08:22:35 PM
| Jon Hodgson wrote on Sat, 15 October 2005 17:50 |
The maths of the sampling theorem and fourier say nothing about what we can hear, they simply state, WITH 100% PROVABLE CERTAINTY, how we can perfectly replicate a signal of given bandwidth. |
Not quite an accurate statement if you cannot make it work in the real world, since there are no such things as "Perfect Filters," thus this is far from an accurate scientific statement because it has NOT been proven to actually work in the real world.
100 % certainty? Yeah, right. Sounds more like marketing spin, wild exageration, and total BS when you don't have Perfect Filters.
BTW, my opinion has not changed, I still do not the sound of digital.
Post by: Jon Hodgson on October 15, 2005, 09:13:17 PM
| Johnny B wrote on Sun, 16 October 2005 01:22 | ||
Not quite an accurate statement if you cannot make it work in the real world, since there are no such things as "Perfect Filters," thus this is far from an accurate scientific statement because it has NOT been proven to actually work in the real world. 100 % certainty? Yeah, right. Sounds more like marketing spin, wild exageration, and total BS when you don't have Perfect Filters. BTW, my opinion has not changed, I still do not the sound of digital. |
And once again he misses the point!!
Nyquist doesn't require perfect filters to be valid, it talks about bandwidth. The process of getting a bandlimited signal FALLS OUTSIDE OF THE NYQUIST THEOREM.
It doesn't matter whether your filters are good or bad, the Nyquist theorem still stands, in fact it also predicts exactly what happens when your filters are bad.
So, with a good implementation, Nyquist stands, and with a bad implementation - Nyquist still stands!!
Mathematics is a real bummer like that, it doesn't matter which way round you go, you still get the same result.
You see, you actually have to understand what you are referring to when you try to argue your position.
As for "perfect" filters, in THEORY you are correct, but in PRACTISE by giving yourself enough space between your top desired frequency and the nyquist limit, and using oversampling converters, it is possible to implement filters which leave the desired band ruler flat and have attenuated any signal down to less than a sample by the time you get to the Nyquist limit. So to all intents and purposes, perfect filters are possible... it's rather harder to achieve the same thing in an analogue recorder (and yes, you have to bandlimit that signal, or it interferes with the bias signal and everything goes horribly wrong), since you don't have the advantage of oversampling.
Personally I doubt you'd do nearly as well in a properly conducted blind test as you think you would, certainly nothing like your claims of what you can hear imply you would, but as for not liking the "sound" of digital, your subjective opinion is yours to have... just don't misrepresent the science.
Post by: dcollins on October 15, 2005, 10:19:21 PM
| Jon Hodgson wrote on Sat, 15 October 2005 18:13 |
Personally I doubt you'd do nearly as well in a properly conducted blind test as you think you would, certainly nothing like your claims of what you can hear imply you would, but as for not liking the "sound" of digital, your subjective opinion is yours to have... just don't misrepresent the science. |
You say that like Johnny B. doesn't have the experience to back himself up............ Or does the "B" stand for something?
DC
Post by: Johnny B on October 16, 2005, 12:51:07 AM
Jon, you seemingly made the false claim that existing converters are perfectly accurate as some kind of wildly exagerated scientific fact, and that's a bunch of bull...If that's what you are implying you may as well get a job spinning for the Bush Regime.
Nyquist may be fine on paper, but in real life...without perfect filters then you a have a serious problem.
In fact, one of the primary reasons for going with higher sample rates is that the filters can be made to work much better and at a cheaper cost, is that not the case?
Of course it is.
Oh, I know...you are buying into the chip makers being involved in some giant consirapacy theory...is that it?
Post by: dcollins on October 16, 2005, 12:57:33 AM
| Johnny B wrote on Sat, 15 October 2005 21:51 |
No I do not miss the point, the filters have to be perfect for that theory to work, they are not perfect...get a clue. |
JB, I've got it.
You should change your .sig file to:
"In theory, there is no difference between theory and practice, but in practice, there is."
That way we can all have a hearty laugh, instead of just wincing at your posts...
DC
Post by: Johnny B on October 16, 2005, 01:11:13 AM
I suppose you are now claiming that you are expert in all the multi-disciplinary sciences, that you personally lay out the logic in the chips, and that YOU are the greatest know-it-all on earth...because that's how you come across...
Perhaps you missed that lesson in school about good manners...and apparently you learned nothing at all when KK Proffitt tore you a new asshole.
Post by: Johnny B on October 16, 2005, 01:15:08 AM
Where can I buy a "Perfect Filter?" I'd really like to know? Ha Ha
Post by: kraster on October 16, 2005, 02:25:08 AM
| Johnny B wrote on Sun, 16 October 2005 06:11 |
DC, I suppose you are now claiming that you are expert in all the multi-disciplinary sciences, that you personally lay out the logic in the chips, and that YOU are the greatest know-it-all on earth...because that's how you come across... Perhaps you missed that lesson in school about good manners...and apparently you learned nothing at all when KK Proffitt tore you a new asshole. |
I'm literally gobsmacked at this remark. I have never met Mr. Collins, but it's safe to say that he is highly respected in the audio world. Your repeated displays of ignorance have frustrated some of the most experienced and learned members of the audio community who have carefully and methodically explained to you how it all works. To no avail whatsoever.
In light of this has it ever occured to you that YOU are the arrogant and rude one here, mate. Saying that you're often wrong as you so often do and are does not excuse this. Neither does your sycophantic references to George Massenburg, Rupert Neve or anyone else who you can conveniently fit into your argument. In the face of formidable knowledge you dance around like a headless chicken trying to get "one up" on everyone with anecdotal, irrelevant and subjective references.
Manners....Sheesh.......
Post by: Gunnar Hellquist on October 16, 2005, 05:41:46 AM
just maybe it is time to ask mr Jonny B to show his credentials. My guess is that he has never engineered a record that got published, and that he has never really used any top line equipment, analog or digital.
Post by: Malcolm Boyce on October 17, 2005, 11:09:03 AM
That shot at DC... You don't have nearly enough folks around here who share your way of thinking on these topics. Especially some of the most educated people in the field. I hate to say it, but if you took a college course on digital audio... you'd probably FAIL!.... Oh, but then you'd try and convince everyone the theory is flawed, and everyone isn't using their ears, just some crazy "math"....
| Johnny B wrote on Sun, 16 October 2005 02:15 |
Jon, Where can I buy a "Perfect Filter?" I'd really like to know? Ha Ha |
I find it very interesting that someone who says something like that, and apparently finds it funny, also finds it appropriate to say this on another current thread:
| Johnny B wrote on Mon, 03 October 2005 04:06 |
But let's say the OS already was 64-bit, just for purposes of discussion and that it was now stable...say Longhorn was stable and was now at a full 64-bits, OK. Now address the thruput issue... |
OK JB... Let's just say a "perfect" filter exists...
I've never seen such a blatant disregard for facts masquerading as "forward thinking". How could someone hearkening back to analog, be trying so hard to come across as a guy looking forward to how things should be... AND WHY ARE ALL YOU CRAZIES TRYING TO HOLD BACK PROGRESS!!!
Talking about ultrasonics and how vibration in air reacts to our bodies is something very learned people are looking at. Using that as some kind of argument as to why the digital audio format is flawed, is like trying to say that things sound different in different light.... Hey, just because nobody has proven it, doesn't mean it's not true around here.
I find my 16 bit mixes sound a lot better with the lights on, but with my 24 bit mixes, I don't notice as much of a difference. I think it's because of some kind of acoustic dither the light is adding to the sound.......... Discuss...
Of course I could be wrong, I often am...
Malcolm
Post by: jimmyjazz on October 17, 2005, 12:43:37 PM
| Johnny B wrote on Fri, 14 October 2005 22:57 |
Oh, so now you are calling George Massenburg names...he reportedly loves this new Sanken mic. |
I have an idea, Johnny B. Why don't you ring up Mr. Massenburg and ask him some of your questions? I don't think you'll like the answers, but hey, he's your man. Maybe if it comes from him you'll listen.
Post by: Johnny B on October 17, 2005, 01:04:05 PM
1. The formats should change, 16-bit and MP3 are not good enough.
2. The sampling rates need to be adjusted for those new and improved formats
3. The sampling rates should be raised or adjusted to capture the frequencies that Cal Tech's professor Boyk measured.
4. Using the higher sample rates can result in better filters
5. The 20-to-20 myth needs to be allowed to die
6. The old mic and speaker fallback argument is no longer valid
7. The "important" frequency spectrum used to adjust the sample rates higher must not rely upon a constrained view and be allowed to focus solely on the ear alone, it must take into account the impacts on the entire human system including the "brain" and the "body."
8. I supposedly cannot hear 8Hz, but I sure can feel it in my body. More focus should be placed on the low-end where some serious problems seem to occur.
9. The views of KK Proffitt regarding the "important" frequency spectrum and women in general ought to be given substantial weight.
10. Since many people still prefer the sound of analogue, that is strong evidence that digital has some serious problems which need further work. There are people who describe digital as sounding "thin, sterile, and ice cold." One describes the sound of digital as being akin to 'having an ice pick shoved into his ears" when he compares it to analogue, which he finds much more pleasing.
11. Out of respect for Walter Sear, who is one of digital sound's harshest critics, chances are, when Walter Sear and people like him are fully satisfied, then the analogue vs. digital debate will subside.
Now people can argue about a better path to improving digital sound quality, they can make positive suggestions, and they can disagree with me all they want on the best path to take, that's fine, but I think they should do that without the lame ass name-calling. YMMV
BTW, what happened with the test that Steve Albini suggested that some of try? Did any of you actually do it?
Post by: Jon Hodgson on October 17, 2005, 01:56:49 PM
Now he's gone, does anyone fancy an intelligent conversation about this stuff?
Post by: Malcolm Boyce on October 17, 2005, 03:44:50 PM
| Johnny B wrote on Mon, 17 October 2005 14:04 |
5. The 20-to-20 myth needs to be allowed to die 7. The "important" frequency spectrum used to adjust the sample rates higher must not rely upon a constrained view and be allowed to focus solely on the ear alone, it must take into account the impacts on the entire human system including the "brain" and the "body." 8. I supposedly cannot hear 8Hz, but I sure can feel it in my body. More focus should be placed on the low-end where some serious problems seem to occur. |
I think we need to remember that we are talking about "audio", and not some multi-sensory medium. I remember learning a definition of what sound is some years ago, and that still stands up after many years of study.
What's interesting in calling 20-20K a myth, is that in hearing tests, a lot of adults can't even hear THAT bandwidth... and that's not a myth.
What we need to decide, when talking about things such as sampling science, is what we are trying to reproduce, before we can debate the variables and their effects. You can easily say, that an infinite number of factors could affect the perception of sound for a human being. What we have to decide is where we are going to draw the line... today. I'm not saying that we are closing the door on future science, like some may like you to believe. I'm just saying that the "myth" of 20-20K is still the number that the overwhelming number of experts are still measuring performance by, to the approval of an even more overwhelming number of users.
| Quote: |
10. Since many people still prefer the sound of analogue, that is strong evidence that digital has some serious problems which need further work. There are people who describe digital as sounding "thin, sterile, and ice cold." One describes the sound of digital as being akin to 'having an ice pick shoved into his ears" when he compares it to analogue, which he finds much more pleasing. |
Since many people prefer the sound of digital, that is strong evidence that analog has serious problems which need further work. There are people who describe analog as sounding "noisy", or "inaccurate and problematic". Many engineers of note relate stories of analog tape losing bandwidth with each pass of a track.
I ask you, is this any less accurate than the above point #10?
Malcolm
Post by: Johnny B on October 17, 2005, 05:51:54 PM
And the 20-to-20 belief is nothing more than a myth. Even Dan would extend it some...not as much as others may want...but he'd go at least a little higher and seems to be willing to take a second look at it...which is more than some would do...
And please don't forget the important work on the low-end either, unfortunately, whenever the sample rate debates come up, the focus, always shifts to the highs and the lows get orphaned in the heat of the discussion.
To me, "How much?" of the frequency spectrum is "important" to the entire human system (the body, the brain, and the ear) is open for legitimate debate.
Post by: Jon Hodgson on October 17, 2005, 05:56:03 PM
Post by: Johnny B on October 17, 2005, 06:12:38 PM
Post by: Jon Hodgson on October 17, 2005, 06:22:24 PM
| Johnny B wrote on Mon, 17 October 2005 23:12 |
I know that's the claim, but look a little closer. |
I did.
Some systems have coupling filters which take out the DC, but there's nothing inherent in sampling systems that requires them.
Post by: Malcolm Boyce on October 17, 2005, 06:54:40 PM
| Johnny B wrote on Mon, 17 October 2005 14:04 |
Ok, here's my final post on this thread. |
Quit kiddin' around will ya.
| Jon Hodgson wrote on Mon, 17 October 2005 14:56 |
Good, Now he's gone, does anyone fancy an intelligent conversation about this stuff? |
Oh, I so hoped you were right.
| Johnny B wrote on Mon, 17 October 2005 18:51 |
You know M. you have your personal opinion, I disagree with most of what you said, and that's Ok. |
That's probably the nicest compliment you could give me.
Post by: Johnny B on October 17, 2005, 08:53:14 PM
And listen too. Listen very carefully.
Post by: Ronny on October 17, 2005, 10:14:14 PM
| Johnny B wrote on Mon, 17 October 2005 17:51 |
You know M. you have your personal opinion, I disagree with most of what you said, and that's Ok. Let's just agree to disagree. How's that sound? And the 20-to-20 belief is nothing more than a myth. Even Dan would extend it some...not as much as others may want...but he'd go at least a little higher and seems to be willing to take a second look at it...which is more than some would do... And please don't forget the important work on the low-end either, unfortunately, whenever the sample rate debates come up, the focus, always shifts to the highs and the lows get orphaned in the heat of the discussion. To me, "How much?" of the frequency spectrum is "important" to the entire human system (the body, the brain, and the ear) is open for legitimate debate. |
Johnny, you need to run a fixed tone test on yourself, so that you'll be able to recognize frequencies that are inconsequential. It's quite easy to chart your F-M curve. Start with a -10dB 440Hz sine tone, an A at concert pitch, monitor it at a comfortable level, we'll be staying on octave A's for the first part of the exercise. Listen to the A 440Hz for a few seconds and than stopping momentarily on each, go down to 220Hz, 110Hz, 55Hz and finally 27.5Hz. Note the decrease in volume each time the tone goes lower. Going higher it's going to be a lot more profound as the percentage of octaves narrows, more on this later. Now go back to 440Hz, again stopping momentarily on each, increase the tone to 1k, than 2k, 4k, 8k, 10k, 12k, 14k and 15k. Notice the volume curve rising until 2k and than gradually falling at 4k, but substantially more fall at 8, 10 and 12k, 14k pretty low compared to 2k which should be close to your most perceptive frequency. The range between 1k and 2k will be your loudest freq's. Put it back on 2k for a few seconds, than toggle to 15k, "slightly" turn the gain up so that you can hear the 15k signal better and note how many dB's you would need to increase the 15k tone to perceive it at the same volume as the 2k signal. Your tweeters are still hearing the full gain at -10dB, although your ears are barely hearing the tone compared to the 2k, so don't blow your tweeters, just boost the 15k tone slightly so that you can reference how many dB of gain you would have to add to hear it as loud as the 2k tone, do the same at 10k. You really don't need to plot the curve to see how fast your acuity falls below 10k. It increases in percentage of how fast the curve drops as you go above 15k and evermore as you go higher. Now play the -10dB 2k tone and generate another 20k tone and play them both together at -10dB. How much is the 20k tone affecting the 2k tone? Music is made up of the fundamentals between 27.5 and 4.2k. Not coincidentally your acuity falls about half way in between these two frequencies, so at 2k where much of the content of music is at, the harmonics at 20k are very inconsequential in doing anything at all to the most perceptive to human frequencies that reside in the fundamentals that musical instruments play. Now let's go up to 384k where you can faithfully capture 192k freq's. The fall of your frequency acuity curve is way, way down at 192k. Any harmonics at that freq would be way, way inconsequential. Of course this is all theoretical because you can't even hear a 20k tone unless you boost the gain by +70dB above what you hear a 2k tone at to just barely be able to hear it and your tweeters won't take that much gain. So if your hair cells were capable of vibrating at 192k, the gain would have to be several hundred dB above what you percieve 2k at, due to the natural human perception curve. Of course the hair cells don't vibrate that fast in reality. If this exercise doesn't clue you in, I'm afraid that I'm going to have to give up on you.
If this does clue you in, next week I'll show you how to use sweep tones to demonstrate how each ear and side of the brain perceives different frequency ranges louder or lower than the other and how your ears are naturally exhibiting pan trajectories just by the acuity difference between each ear, where frequencies are perceived to shift relative to location of stereo sound source.
Post by: dcollins on October 18, 2005, 12:40:30 AM
| Jon Hodgson wrote on Mon, 17 October 2005 10:56 |
Good, Now he's gone, does anyone fancy an intelligent conversation about this stuff? |
Do you feel minimum phase filters, as used by Wolfson in new D/A products, have merit over conventional FIR implementations?
DC
Post by: Jon Hodgson on October 18, 2005, 03:42:14 AM
| dcollins wrote on Tue, 18 October 2005 05:40 | ||
Do you feel minimum phase filters, as used by Wolfson in new D/A products, have merit over conventional FIR implementations? DC |
Right now, I have no idea!
But give me some links if you have them and I'll get reading, then we can take it from there.
This should make a nice change, talking about something I haven't known the answer to for over a decade
Post by: Jon Hodgson on October 18, 2005, 03:47:11 AM
| Johnny B wrote on Tue, 18 October 2005 01:53 |
Look some more Jon, look more carefully. And listen too. Listen very carefully. |
Why? Have the laws of physics and mathematics changed since the last time I did it?
What possible pseudo science could you have for sampling not being able to capture down to DC? Perhaps we're sampling too quickly now?
Post by: blueboy on October 18, 2005, 04:21:53 AM
| Jon Hodgson wrote on Mon, 17 October 2005 10:56 | ||
Right now, I have no idea! But give me some links if you have them and I'll get reading, then we can take it from there. This should make a nice change, talking about something I haven't known the answer to for over a decade |
Hi Jon,
Here's some links for you....
http://ccrma.stanford.edu/~jos/filters/Linear_Phase_Audio_Fi lters.html
http://www.dspguru.com/info/faqs/fir/props.htm
I had a similar question in relation to a minimum phase filter being used to improve sample rate conversion.
The following are quotes taken from the above links.
| Quote: |
Minimum-phase filters (which might better be called "minimum delay" filters) have less delay than linear-phase filters with the same amplitude response, at the cost of a non-linear phase characteristic, a.k.a. "phase distortion". |
| Quote: |
It is generally accepted that zero or linear phase filters are ideal for audio applications. This is because such filters delay all frequencies by the same amount, thereby maximally preserving waveshape. Mathematically, all Fourier-components passed by the filter remain time-synchronized exactly as they were in the original signal. However, this section will argue that a phase response somewhere between linear- and minimum-phase may be even better in some cases. |
I thought this stuff on minimum phase filters might relate to DC's question.
This quote is from the AES paper that I think DC is referring to.
| Quote: |
P10-6 An Ultra High Performance DAC with Controlled Time-Domain Response—Paul Lesso, Anthony Magrath, Wolfson Microelectronics - Edinburgh, Scotland, UK This paper describes the design of an ultra-high performance stereo digital-to-analog converter (DAC) employing advanced digital filtering techniques. Recently there has been a renewed interest in the time-domain properties of digital filters used for interpolation and decimation. Linear phase FIR filters, which have proliferated digital filter design for the last two decades, have the undesirable properties of pre-ringing and high group delay. Conversely, minimum phase filters, which offer lower levels of pre-ringing, do not have a uniform phase response. This paper describes the trade-offs in the design of filters with controlled pre-ringing, coupled with desirable phase and magnitude characteristics. The paper also describes architectural choices in the implementation of the DAC signal processing chain, required to achieve commensurate analog performance. Convention Paper 6577 |
http://www.aes.org/events/119/papers/session.cfm?code=P10
If you get a chance after answering Dave C, please have a look at my post regarding the Voxengo R8Brain Pro sample rate converter that has a minimum phase filter option. I'm just curious as to how it compares to linear interpolation, sinc interpolation and polyphase filtering. Any comments would be appreciated.
http://recforums.prosoundweb.com/index.php/mv/msg/7575/98207 /11771/?SQ=618d52af60703c03c3a6efa2d73a4ba1
Thanks for the incredible info you've shared so far on these forums.
Regards,
JL
Post by: Jon Hodgson on October 18, 2005, 04:51:27 AM
That all looks interesting, I'll have a good read later, right now I have some DSP programming to do.
cheers
Jon
Post by: Johnny B on October 18, 2005, 10:32:24 AM
"Do you feel minimum phase filters, as used by Wolfson in new D/A products, have merit over conventional FIR implementations?
DC"
Looks like an improvement, but perhaps more in the nature of an "incremental improvement" rather than an "earth-shaking breakthrough." At this point, any and all improvements are good.
Post by: Johnny B on October 18, 2005, 10:43:17 AM
Thanks for the link to Stanford's CCRMA (pronounced "Karma") Lab. I'll bet some of you may not know that work done on early Yammy synths resulted in an income stream for the lab which, as I understand it, continues to this day. Such income streams can help with things like new earthquake resistant buildings which are much safer places for the staff and students to work
Post by: kraster on October 18, 2005, 11:45:36 AM
I used to be an analog purist but I, personally, put that down to nostalgia more than anything else. The Hiss of tape or the Rumble of a turntable is what I associated with good music from a young age and it took me some time to recognize this bias (excuse the pun!) in myself.
I like Digital as a recording medium now because, when properly implemented, it's more accurate, less prone to noise, more cost effective, easier to manipulate and easier to store than analog tape. Oh it sounds good too.
So for once and for all, without any sense of facetiousness, I ask you what, in quantifiable terms, are the problems with digital and the benefits of analog as a recording medium as you perceive them?
All the best,
Karl.
Post by: Johnny B on October 18, 2005, 01:10:56 PM
I think I know where you are coming from, however, if you get some great analogue gear and do some listening comparisons, you may decide to change your opinion.
Post by: C-J on October 18, 2005, 03:18:11 PM
| Johnny B wrote on Tue, 18 October 2005 20:10 |
Karl, I think I know where you are coming from, however, if you get some great analogue gear and do some listening comparisons, you may decide to change your opinion. |
Yeah, Johnny,
But we all have repeatedly tried to ask about your technical opinion. Not just the "because it sounds better, y'all!" statements. No luck, so far...
Judging from your posts, it seems like you have analog gear that supports 100kHz mics better than any existing digital system...?
C.J., Finland
Post by: minister on October 18, 2005, 03:20:58 PM
| Johnny B wrote on Tue, 18 October 2005 12:10 |
I think I know where you are coming from, however, if you get some great analogue gear and do some listening comparisons, you may decide to change your opinion. |
i would like to second the nomination to award Jon Hodgson's efforts!! they were stout-hearted, patient and informative. if i want my head to spin, i come here and give a read. i am learning about the theories and design implementation of digital, so, while a lot of this is over my head, i have come to a greater understanding by reading his prodigious posts -- of course, also those of the other notables here. so, they were not in vain. they taught me something. i have more reading and understanding to do. but, really, THANKS JH!! i am sure that once i read the basic books on the topic, Dan Lavry's papers, i will come back to those posts to see if more lights come on.
Post by: Malcolm Boyce on October 18, 2005, 03:31:29 PM
| minister wrote on Tue, 18 October 2005 16:20 |
if i want my head to spin, i come here and give a read. i am learning about the theories and design implementation of digital, so, while a lot of this is over my head, i have come to a greater understanding by reading his prodigious posts -- of course, also those of the other notables here. so, they were not in vain. they taught me something. i have more reading and understanding to do. but, really, THANKS JH!! i am sure that once i read the basic books on the topic, Dan Lavry's papers, i will come back to those posts to see if more lights come on. |
Ditto.
Post by: Jon Hodgson on October 18, 2005, 04:02:06 PM
I don't know what you're reading, and it's a while since I took a look at it (my copy's in another country right now), but I always found John Watkinson's "The Art of Digital Audio" to be very readable.
Post by: minister on October 18, 2005, 04:27:39 PM
Post by: kraster on October 18, 2005, 11:26:49 PM
| Johnny B wrote on Tue, 18 October 2005 18:10 |
Karl, I think I know where you are coming from, however, if you get some great analogue gear and do some listening comparisons, you may decide to change your opinion. |
Johnny B,
With your evasiveness you should be a politician.
If you read my post I said that I used to be an analog purist. This would imply that I have listened to a lot of analog gear. I grew up with it. My father used to cut up my plastic lunch boxes so he could make colour coded knobs for the console he was building. He also built an eight track tape machine and a load of other stuff.
But my question to you was: What specifically is the problem with Digital in your opinion? No half baked pseudo-science side issues.
All I can ascertain from you is that you have an irrational dislike of Digital Audio and the mathematicians who helped create it (including a pronounced abhorrance for poor old Nyquist).
I just can't figure out why this is the case in light of all the expert opinion that's been offered up here...???
Don't forget my question.
Post by: Johnny B on October 19, 2005, 03:26:47 AM
Your father sounds like great man...it also sounds as if he spent quite a bit of time with you and that you learned some valuable lessons. IMHO, that's great father. Sounds like you were lucky to have him.
As for me, it's not that I hate math or Nyquist or technology or anything like that, but that I'm always willing to challenge all past belief systems, all old theories, everything that has ever come before when looking at any problem. I throw out the baby, the bath water, and the entire plumbing system in an effort to gain some form of new insight. This does not always work for me, sometimes I have to bring back the baby, the bath water, and all the old plumbing system and just admit my defeat or find an entirely different approach to attack the problem needing a solution.
IOW, it's just a technique I use to varying decrees of success or failure. Some of the people around the high-tech biz used to have a phrase: "Fail, Forward, Faster," and they would construct entire seminars around this concept. I suppose the idea can be be reduced to "The more failures you have, the faster you move forward."
The classic example is the Thomas Edison "Light Bulb Story." There, after 9999 attempts, an observer asked old Tom, "Mr. Edison, you have tried to make this thing work 9999 times, aren't you tired of all those failures?" And old Tom is supposed to have replied, "Why my good man, I have not failed. I now know 9999 ways not to make a light bulb."
Around the 10,000th try, he finally got it to work.
One could look at the challenges presented by digital in the same way, many years of failures, not quite getting it right, getting it to work some, but still not quite right...all such efforts could be viewed as failures which constantly move the tech forward.
With digital, I think it's clear that some things are not quite going according to plan...after 25 years of "efforting" by a wide variety of people and companies...I'm just not happy with digital sound quality...I think digital can be made to sound far better than what exists today. But with 25 years of "efforting," an element of frustration can sink in and increase one's willingness to challenge all the underlying tech. Sometimes a fresh approach works best, not always, but it works well enough in so many situations that it can become a viable approach worthy of consideration.
I think more than a few people are not happy, I think they hear problems in the "time domain." People have advanced all sorts of theories of why they do not like the sound of digital. Even after all the so-called "models" and "emulators" have been become popular, a significant group of people still does not *really* like the sound quality of digital.
That means there are still some serious problems in digital, and those problems require some new solutions. I think a good step in the right direction would be to allow the old 16-bit CD format to die. If I had all the other answers required to finally crack the digital nut, I'd be running to the Patent Office and calling around to enter into licensing deals so I could collect royalties and move to an island paradise. I half expect a couple of bright young minds working out of the proverbial "garage" to win the prize and wind up on that lovely island.
Post by: Ronny on October 19, 2005, 08:35:40 AM
Name me one high profile engineer that doesn't use digital somewhere in their chain and I'll name you 50 that do. Name me 10 high profile engineers that say that digital sucks and annie is better for all applications. I'll name you 100 that use both, utilizing the optimal format to suit the given situation.
My last advice to you. Learn frequency reaction and your M-F curve from the test that I spend 30 minutes typing up for you, instead of ignoring people that are tying to help you. Quit dodging everyone's questions when they pin you down to explain your cockamamie logic and follow the laws of physics with your explanations without any voodoo or countering information that was proven years ago. Start testing scientifically in the blind so that you'll realize that you can't tell analog from digital when they are both done right. There isn't any all analog music anymore, wake up, smell the roses, adapt to the modern age. Quit trying to save the world from bad recording formats based on your psuedo evaluations. You can't tell the difference between 24 and 16 bit unless you can hear below -96dB, so don't give us anymore golden earred bullcrap. You think you are open minded and have new ideas, but your knowledge of audio puts you at a very narrow minded disadvantage to intelligently discuss audio with anyone on the 3 forums that I've heard you sprout your very misinformed analogies on. Treat your elders and the folks that write the book on digital audio design with more respect and than you may just start learning something.
Post by: Johnny B on October 19, 2005, 11:23:58 AM
If you think digital is absolutely great just the way it is, that's ok with me. If you believe, however wrongfully, that digital has reached the pinnacle and the only place for it to go is downward, that's fine with me.
I do not *really like* the sound of digital.
I prefer the sound of analogue, I feel it sounds superior.
Post by: kraster on October 19, 2005, 11:24:10 AM
| Johnny B wrote on Wed, 19 October 2005 08:26 |
Karl, Your father sounds like great man...it also sounds as if he spent quite a bit of time with you and that you learned some valuable lessons. IMHO, that's great father. Sounds like you were lucky to have him. As for me, it's not that I hate math or Nyquist or technology or anything like that, but that I'm always willing to challenge all past belief systems, all old theories, everything that has ever come before when looking at any problem. I throw out the baby, the bath water, and the entire plumbing system in an effort to gain some form of new insight. This does not always work for me, sometimes I have to bring back the baby, the bath water, and all the old plumbing system and just admit my defeat or find an entirely different approach to attack the problem needing a solution. IOW, it's just a technique I use to varying decrees of success or failure. Some of the people around the high-tech biz used to have a phrase: "Fail, Forward, Faster," and they would construct entire seminars around this concept. I suppose the idea can be be reduced to "The more failures you have, the faster you move forward." The classic example is the Thomas Edison "Light Bulb Story." There, after 9999 attempts, an observer asked old Tom, "Mr. Edison, you have tried to make this thing work 9999 times, aren't you tired of all those failures?" And old Tom is supposed to have replied, "Why my good man, I have not failed. I now know 9999 ways not to make a light bulb." Around the 10,000th try, he finally got it to work. One could look at the challenges presented by digital in the same way, many years of failures, not quite getting it right, getting it to work some, but still not quite right...all such efforts could be viewed as failures which constantly move the tech forward. With digital, I think it's clear that some things are not quite going according to plan...after 25 years of "efforting" by a wide variety of people and companies...I'm just not happy with digital sound quality...I think digital can be made to sound far better than what exists today. But with 25 years of "efforting," an element of frustration can sink in and increase one's willingness to challenge all the underlying tech. Sometimes a fresh approach works best, not always, but it works well enough in so many situations that it can become a viable approach worthy of consideration. I think more than a few people are not happy, I think they hear problems in the "time domain." People have advanced all sorts of theories of why they do not like the sound of digital. Even after all the so-called "models" and "emulators" have been become popular, a significant group of people still does not *really* like the sound quality of digital. That means there are still some serious problems in digital, and those problems require some new solutions. I think a good step in the right direction would be to allow the old 16-bit CD format to die. If I had all the other answers required to finally crack the digital nut, I'd be running to the Patent Office and calling around to enter into licensing deals so I could collect royalties and move to an island paradise. I half expect a couple of bright young minds working out of the proverbial "garage" to win the prize and wind up on that lovely island. |
Johnny B,
Thanks for your post.
I still don't know why you don't like Digital other than you don't like it!!
You mention that some people hear problems in the time domain. Who are these people? Are you one of them? If so what are the problems you hear?
A lot of people don't like digital because it doesn't sound like analog but that doesn't mean analog is a better format. Some people like tape saturation, distortion and wow & flutter because they perceive it as being more musical. Horses for courses....
Johnny, in order to improve on something you must be able to quantify what it is you're trying to improve on. Wild stabs in the dark until you find the desired outcome is not a prudent course of action. I'm sure even Edison himself was not trying to make lightbulbs from catguts and balloons and his repeated efforts were more to do with the minutae of the problem he was facing rather than flailing around at random trying to find the solution.
I'm 100% sure that the practise of trial and error goes on in science and math everyday. It is the nature of the beast but I doubt many scientists and mathematicians are ignoring the vast scientific knowledgebase and start afresh each day with 1+1=2.
The reason a lot of people are frustrated with you is that you persistently assume that they could not be bothered with improving their systems. This is simply not the case. Dan Lavry has gone to great lengths to lay out his argument against higher sampling rates and this paper is available to the whole world. Good scientists know when to quit because they've already done their research and experiments and have conclusive proof that the selected course they're on is a blind alley. It's insulting to these people to suggest they haven't done this.
Personally, I think that in the early days of digital recording music suffered greatly because the implementation of the sampling theorem was far from perfect. Things have come a long way since then and ringy filters and all the other anomalies associated with digital in those days are now a thing of the past. Why? Because people quantified these problems and sought out solutions .
PS. I think it's an honour to be able to hear the advice of the learned people on this forum. I think it prudent not to abuse this privilege.
Post by: Johnny B on October 19, 2005, 12:53:49 PM
If it were easy to "quantify" these problems they could be fixed more easily.
If Edison had attempted to quantify his problems with the filaments, he may not have ever made the light bulb work. It took him 10,000 tries to get it right.
Math is a great tool, but it's an abstraction which only exists in people's heads, it's not reality, it's just a tool. It can be an extremely helpful tool in solving a problem, but sometimes, depending upon how it's applied, it can lead people in the wrong direction.
Some people seem to think of math as a "religion," rather than as an attempt to "get close" to describing what is reality which can never be fully accomplished by math. Math can be made to lie, and some liars can use math.
Now by pointing out that math has some inherent weakness, does not mean I'm trying to insult anyone, nor does it mean that I'd take math out of anyone 's toolbox.
But here, I think the "ear," or more accurately, the entire "ear-body-brain interaction" of the human organism is extremely complex, and that it requires a multidisciplinary approach by an entire team of qualified people to get a handle on it, let alone accurately describe and "quantify" it. At this point, I rather doubt if we have anything even approaching a good math model of the important human elements that make up the entire human experience when subjected to sound.
In any event, I feel that the "ear", or that "human experience," trumps math and must be elevated in importance and given a far higher rating on the scale. So in my very limited view, until those multidisciplinary factors are better understood, the "ears" and the entire "human experience" rule.
Maybe this will put it into some context, I dunno, but I'll try.
Would you rather have sex by way of math, models, and emulations when you know they are not quite real? Or would you rather have sex with a live, breathing human being?
And when you answer the question, can you "quantify" it with math, or do emotional factors and "feelings" account for part of your choice? Do those "emotional factors" and "feelings" have anything at all to do with your brain chemistry?
Sure you could try to apply some multidisciplinary science here, but trying to "prove it" or "quantify it" with math alone would be rather difficult.
So in this sense, it's really hard to say exactly why people do not like digital sound quality. It's also hard to quantify with math exactly why most guitar players prefer tube amplifiers. We do "know" they do, because they tell us so.
But just because it's difficult to quantify, to accurately describe, does not mean that something does not exist. There are problems in digital, people hear the problems, they experience the problems, they attempt to describe what they hear and experience, and then, because they have difficulty trying to describe or "quantify" what they are hearing and experiencing, some people call them stupid or crazy. I do not think that is the right approach, nor, do I think it is fair.
I hope this helps.
Post by: Jon Hodgson on October 19, 2005, 01:18:30 PM
| Johnny B wrote on Wed, 19 October 2005 17:53 |
C'mon Karl. If it were easy to "quantify" these problems they could be fixed more easily. If Edison had attempted to quantify his problems with the filaments, he may not have ever made the light bulb work. It took him 10,000 tries to get it right. |
Not content with trying to spread myths about digital audio, you are now completely misrepresenting/misunderstanding Edidon's work on the lightbulb!
I've read numbers of 2000 and 6000 tries, never as many as 10000, but the actual number is irrelevant.
What is relevant is what those tries consisted of.
You see Edison didn't invent the lightbulb, 10 years earlier a guy called Swan demonstrated and patented a carbon filament lightbulb in the UK, and Edison appears to have read that work.
What Edison did was REFINE the design and create a durable lightbulb, in fact the majority of those attempts were trying different plant filaments to find one which would last a long time. With each attempt he didn't throw away his and other people's work, he used the knowledge gained to shape the direction he would try next.
Of his failed attempts at the light bulb, he said, "They taught something that I didn't know. They taught me what direction to move in."
In addition later a guy joined Edison at Menlo park, Frank J. Sprague, and one of his major contributions was bringing in mathematics which allowed them to hone in more directly on optimum solutions which in the past Edison had only approached using extensive trial and error.
So basically Edison DID quantify the problems he encountered in order to achieve his goals, and he also later benefited from the use of mathematics to refine his approach.
Post by: Johnny B on October 19, 2005, 01:39:40 PM
Would you also like to correct the record about Seattle Ram and MS DOS? Or, tell the Digital Research and IBM PC DOS story? How about the Xerox PARC story and the Apple Mac? Or maybe the one about the LISA and the Apple Mac pirate group? Maybe you'd like to tell the story of Larry Ellison and Oracle? How about some DEC PDP stories? Or some whatever happened to WANG and Data General Stories? Maybe you'd like to go in the direction of an Apollo story or tell one about HP? Big Blue and the 360? Oh, I know, some Cray Supercomputer stories would be fun. Maybe a Burr-Brown story would be more on point. Since you are such an expert in history, in math, in chip layout and design, maybe you can just solve all the world's problems, not just the problems with digital sound quality.
Sorry for my bad tone, I'm having a bad day.
Post by: Jon Hodgson on October 19, 2005, 01:51:02 PM
| Johnny B wrote on Wed, 19 October 2005 18:39 |
Thanks for the Edison Story corrections, it was merely an attempt to demonstrate the "Fail, Forward, Faster" concept. Would you also like to correct the record about Seattle Ram and MS DOS? Or, tell the Digital Research and IBM PC DOS story? How about the Xerox PARC story and the Apple Mac? Or maybe the one about the LISA and the Apple Mac pirate group? Maybe you'd like to tell the story of Larry Ellison and Oracle? How about some DEC PDP stories? Or some whatever happened to WANG and Data General Stories? Maybe you'd like to go in the direction of an Apollo story or tell one about HP? Big Blue and the 360? Oh, I know, some Cray Supercomputer stories would be fun. Since you are such an expert in history, in math, in chip layout and design, maybe you can just solve all the world's problems, not just the problems with digital sound quality. Sorry for my bad tone, I'm having a bad day. |
Well I could tell you a couple of Seymour Cray anecdotes, but it's probably engineering humour and you wouldn't find them funny. I also suspect he told them much better than I could.
If you're going to present a story as the justification for your actions and attitude, then you're in no position to whine when somebody points out that your story is wrong and thus does not support your stance in any way.
Sorry to hear you're having a bad day, but you're not the only one.
Post by: kraster on October 19, 2005, 01:56:16 PM
| Johnny B wrote on Wed, 19 October 2005 17:53 |
C'mon Karl. If it were easy to "quantify" these problems they could be fixed more easily. If Edison had attempted to quantify his problems with the filaments, he may not have ever made the light bulb work. It took him 10,000 tries to get it right. Math is a great tool, but it's an abstraction which only exists in people's heads, it's not reality, it's just a tool. It can be an extremely helpful tool in solving a problem, but sometimes, depending upon how it's applied, it can lead people in the wrong direction. Some people seem to think of math as a "religion," rather than as an attempt to "get close" to describing what is reality which can never be fully accomplished by math. Math can be made to lie, and some liars can use math. Now by pointing out that math has some inherent weakness, does not mean I'm trying to insult anyone, nor does it mean that I'd take math out of anyone 's toolbox. But here, I think the "ear," or more accurately, the entire "ear-body-brain interaction" of the human organism is extremely complex, and that it requires a multidisciplinary approach by an entire team of qualified people to get a handle on it, let alone accurately describe and "quantify" it. At this point, I rather doubt if we have anything even approaching a good math model of the important human elements that make up the entire human experience when subjected to sound. In any event, I feel that the "ear", or that "human experience," trumps math and must be elevated in importance and given a far higher rating on the scale. So in my very limited view, until those multidisciplinary factors are better understood, the "ears" and the entire "human experience" rule. Maybe this will put it into some context, I dunno, but I'll try. Would you rather have sex by way of math, models, and emulations when you know they are not quite real? Or would you rather have sex with a live, breathing human being? And when you answer the question, can you "quantify" it with math, or do emotional factors and "feelings" account for part of your choice? Do those "emotional factors" and "feelings" have anything at all to do with your brain chemistry? Sure you could try to apply some multidisciplinary science here, but trying to "prove it" or "quantify it" with math alone would be rather difficult. So in this sense, it's really hard to say exactly why people do not like digital sound quality. It's also hard to quantify with math exactly why most guitar players prefer tube amplifiers. We do "know" they do, because they tell us so. But just because it's difficult to quantify, to accurately describe, does not mean that something does not exist. There are problems in digital, people hear the problems, they experience the problems, they attempt to describe what they hear and experience, and then, because they have difficulty trying to describe or "quantify" what they are hearing and experiencing, some people call them stupid or crazy. I do not think that is the right approach, nor, do I think it is fair. I hope this helps. |
But you are talking exclusively about emotional Bias here. The main reason I see most people trump for certain gear is because they've been emotionally tweaked by either personal experience or marketing forces. Buying an exact replica of Jimi Hendrix's wah-wah pedal won't make you play Voodoo Chile like him but some people believe that it will bring them closer to it.
Math is not reality, you are right, but it a device used to describe and quantify it. With great success. If I put an item in front of you and then place another on beside it we could now describe this mathematically as two items. I don't mean at all to be patronising but this is where it starts. Math can then go on to describe what these items are etc. etc.
Science and math allow us to transmute reality into some kind of medium that can be reproduced. It ISN'T reality. It's a representation of it. Ren
Post by: Johnny B on October 19, 2005, 03:48:54 PM
I've also argued for "expanding the frequency spectrum" that digital captures and reproduces in such a way to better accommodate the measurements taken by Cal Tech's Professor Boyk and the views of people like David Blackmer. And that means the sample rates will have to be adjusted higher. This is not rocket science, it's not that difficult to understand.
How much higher? I dunno.
How much of the frequency spectrum is truly important to the human being?
I dunno that either, no one does.
And please, spare me the diatribe about old "ear" studies. It's a far more complex problem than that.
And, I've argued for the abandonment of the old 16-bit CD format. Again, this is not that hard to fathom.
Do people not believe we can have a better format than 16-bit?
Post by: Jon Hodgson on October 19, 2005, 03:57:57 PM
| Johnny B wrote on Wed, 19 October 2005 20:48 |
No, I never said analogue was a more "accurate" representation than digital, although I'm sure some may want to argue this point, what I said is that "I still like the sound quality of analogue better." I'M STILL NOT HAPPY WITH THE SOUND OF DIGITAL! There. I shouted it. I've also argued for "expanding the frequency spectrum" that digital captures and reproduces in such a way to better accommodate the measurements taken by Cal Tech's Professor Boyk and the views of people like David Blackmer. And that means the sample rates will have to be adjusted higher. This is not rocket science, it's not that difficult to understand. How much higher? I dunno. How much of the frequency spectrum is truly important to the human being? I dunno that either, no one does. And please, spare me the diatribe about old "ear" studies. It's a far more complex problem than that. And, I've argued for the abandonment of the old 16-bit CD format. Again, this is not that hard to fathom. Do people not believe we can have a better format than 16-bit? |
What is it you want?
To have a digital recording system that sound more like analogue tape?
OR
to have a more accurate digital recording system?
You cannot have both. Digital recording is already considerably more accurate, in terms of replicating the signal coming out of a microphone, than analogue tape.
Post by: dcollins on October 19, 2005, 04:02:04 PM
| Johnny B wrote on Wed, 19 October 2005 12:48 |
the measurements taken by Cal Tech's Professor Boyk and the views of people like David Blackmer. |
Boyk makes no mention of audibility. You don't need a FFT spectrum analyzer to expect that a Harmon mute makes harmonics out to channel 3....
DC
Post by: Johnny B on October 19, 2005, 04:10:36 PM
| Jon Hodgson wrote on Wed, 19 October 2005 20:57 |
What is it you want? To have a digital recording system that sound OR to have a more accurate digital recording system? |
Well, if these are the options, then I would want both. Gotta have both.
Call me "a greedy bastard" if you like...
To paraphrase a guy who went through some different kinds of doors,
"I want the world, and I want it now."
Post by: Johnny B on October 19, 2005, 04:23:39 PM
| dcollins wrote on Wed, 19 October 2005 21:02 | ||
Boyk makes no mention of audibility. You don't need a FFT spectrum analyzer to expect that a Harmon mute makes harmonics out to channel 3.... DC |
Dave, I just want to give it a better chance and greater opportunity to prove itself or not, just a "Fair Trial," so-to-speak.
If it all turns out to sound like crap, well then, it's just another idea in the world that did not work out.
And c'mon, be honest, a lot of great things have come from the "Trial and Error" process.
Have ya ever had the experience of doing something that some theory said was all wrong and it all turned out OK or better than you expected? By-the-book is great for some thngs, but sometimes it's OK to throw the book out. Know what I mean?
Be honest, I'll bet something like this has happened to you more than once in your life. You don't do everything by the book, do you?
Post by: Jon Hodgson on October 19, 2005, 04:30:19 PM
| Johnny B wrote on Wed, 19 October 2005 21:10 | ||
Well, if these are the options, then I would want both. Gotta have both. Call me "a greedy bastard" if you like... To paraphrase a guy who went through some different kinds of doors, "I want the world, and I want it now." |
So you want a lower bandwidth, higher noise, higher distortion system with frequency modulation of the signal... which is more accurate?
Post by: blueboy on October 19, 2005, 05:18:49 PM
When I accidentally stumbled onto the PSW REP forum I was blown away at the professionalism of the participants, and wealth of knowledge that was "freely" available. I had never posted or even bothered to read other forums previously as the majority were filled with useless information and immature banter.
This forum, more than any other on PSW is the place to discuss how the science of audio works, and based on that scientific data, how it can possibly be improved.
I'm not discounting the fact that someday someone may accidentally discover a whole new way of representing audio in another medium that will revolutionize the way we record and playback sound. But that type of discussion belongs in a "Thinking outside the audio box" thread. (or perhaps even - "baseless ideas and wishful thinking on ways to make digital sound like analog".)
Many "lurkers" who may have had something intelligent to contribute to this thread have probably been scared off (or too bored to continue reading) about 9 pages ago.
DC's post regarding the use of Minimum Phase filters in D/A conversion sounds interesting. From the little bit of reading I have done, it looks like there may be some merit to this providing an incremental improvement in some cases.
Maybe there is a way to "dynamically" switch between linear and minimum phase filters depending on content. I don't know the math involved, or even if that is a stupid suggestion. But it sounds to me like it has the potential for an intelligent discussion, or at least it may be a great opportunity to "listen and learn".
JL
Post by: danlavry on October 19, 2005, 05:30:50 PM
| Johnny B wrote on Wed, 19 October 2005 21:10 | ||
Well, if these are the options, then I would want both. Gotta have both. Call me "a greedy bastard" if you like... To paraphrase a guy who went through some different kinds of doors, "I want the world, and I want it now." |
Johnny,
You can not have both, no matter how "greedy" you want to be. You simply can not be more accurate and less accurate at the same time. You can not have 2 heads either...
Regards
Dan Lavry
www.lavrengineering.com
Post by: Malcolm Boyce on October 19, 2005, 08:13:08 PM
Maybe he's right...
Post by: dcollins on October 19, 2005, 09:18:54 PM
| Malcolm Boyce wrote on Wed, 19 October 2005 17:13 |
http://www.steelydan.com/lesindex.html Maybe he's right... |
Rupert Neve has connected sampled data systems with increased inner-city violence. Since CD's are, as he puts it, "clicking fourty-four thousand times a second."
Simple enough, eh?
DC
Post by: Jon Hodgson on October 19, 2005, 09:22:31 PM
| dcollins wrote on Thu, 20 October 2005 02:18 | ||
Rupert Neve has connected sampled data systems with increased inner-city violence. Since CD's are, as he puts it, "clicking fourty-four thousand times a second." Simple enough, eh? DC |
Strange,
personally I usually find that what tempts me towards violence is analogue "purists", and worse still "audiophiles" spouting pseudoscience!
Post by: dcollins on October 19, 2005, 09:36:21 PM
| Jon Hodgson wrote on Wed, 19 October 2005 18:22 |
personally I usually find that what tempts me towards violence is analogue "purists", and worse still "audiophiles" spouting pseudoscience! |
I think the audiophile has slowed progress, and caused much confusion in the pro audio world.
Who needs em?
DC
Post by: blueboy on October 19, 2005, 09:50:49 PM
| dcollins wrote on Wed, 19 October 2005 18:36 | ||
I think the audiophile has slowed progress, and caused much confusion in the pro audio world. Who needs em? DC |
Is it just me, or has anyone else noticed that DC says everything with a straight face?
JL
Post by: Jon Hodgson on October 19, 2005, 09:56:11 PM
| blueboy wrote on Thu, 20 October 2005 02:50 |
Is it just me, or has anyone else noticed that DC says everything with a straight face? JL |
What do you expect from a man who mastered a Spinal Tap album?
Post by: Ronny on October 20, 2005, 01:19:43 AM
| dcollins wrote on Wed, 19 October 2005 21:36 | ||
I think the audiophile has slowed progress, and caused much confusion in the pro audio world. Who needs em? DC |
Speaking of audiophiles, Dave, do you still have the link to those pictures of all the gizmos that were taken at an audiophile convention? I got a huge kick out of that and would like to see them again.
Post by: crm0922 on October 20, 2005, 02:07:06 AM
And Jon H. as well. Truly brave folks.
As Dan said, Johnny. You cannot have your cake and eat it too. Otherwise, the cake would be eaten, and be, you know, like, gone.
Chris
Post by: dcollins on October 20, 2005, 03:02:03 AM
| Jon Hodgson wrote on Wed, 19 October 2005 18:56 | ||
What do you expect from a man who mastered a Spinal Tap album? |
Hopefully not some lame "this one goes to 0b1011 joke?
Har!
DC
Post by: dcollins on October 20, 2005, 03:27:16 AM
| Johnny B wrote on Wed, 19 October 2005 13:23 |
You don't do everything by the book, do you? |
No, but you should give them a try.
DC
Post by: Johnny B on October 20, 2005, 10:48:53 AM
| Jon Hodgson wrote on Wed, 19 October 2005 21:30 |
So you want a lower bandwidth, higher noise, higher distortion system with frequency modulation of the signal... which is more accurate? |
Jon,
I'll talk like some mangers. They say things like "I just want 'whatever it takes' to make digital sound better. You guys take care of the details."
I know that may not be a sufficient or satisfying answer for some, but when you have that kind of pressure placed upon you and are under the added pressure to "do something" by some "date certain," you are often required to expand your viewpoint and even consider attempting things that were once considered impossible. If you have not read it, you may find the book entitled "The Soul of A New Machine" entertaining. It's the story of how an entirely new 32-bit minicoupter was built from the ground up in just one year! It's avaliable at Amazon.com.
To get anything effective accomplished, one sometimes has to go into a "green light mode" of creativity where all negative thoughts are banished and all views are considered valid and possible.
IOW, at this point, I see no harm in exploring alternative views, no matter how outlandish they may seem, if they somehow lead to "doing something," and that "doing something" eventually leads to efficient and effective results. Obviously, the goal is to "improve digital sound quality." I suppose one could say there are different methods of "keeping your eye on that prize." But that's the goal: Improve Digital Sound Quality.
Post by: Jon Hodgson on October 20, 2005, 11:06:30 AM
| Johnny B wrote on Thu, 20 October 2005 15:48 |
IOW, at this point, I see no harm in exploring alternative views, no matter how outlandish they may seem, if they somehow lead to "doing something," and that "doing something" eventually leads to efficient and effective results. Obviously, the goal is to "improve digital sound quality." I suppose one could say there are different methods of "keeping your eye on that prize." But that's the goal: Improve Digital Sound Quality. |
But this statement appears completely untrue, because the ONLY alternative you appear to have any interest, your apparant solution to every single difference to your beloved analogue (I won't say problem, because when you're using such a flawed system as your reference point it is unreasonable to call every difference a problem or a fault), actual or imagined, is higher sample rates, wider data paths and wider bandwidth transducers.
You claim to be open to ideas, but you are not, you are simply fixated on one idea. The fact that your idea is based on provably incorrect data and flawed deduction is completely secondary to your apparant dishonesty in not practising what you preach.
Post by: Johnny B on October 20, 2005, 11:32:28 AM
What a minute, while it's true that analogue is the "Gold Standard," that does not mean that any so-called flaws in analogue ought to be used as an excuse to hold back advancing the digital sound technology. There has to be some point of departure...
And yet, it's also sort of like the Buddhist concept of "Two, but not two," in that, the two are inextricably woven together.
I am by no means suggesting there are any off-the-shelf easy solutions. Clearly, arriving at the goal of improving digital sound quality is no trivial task, it's going to take substantial effort by a lot of people.
That is, unless some kids working out of the proverbial "garage" come up with something earthshaking.
However, the incentives for getting some new earthshaking "must have" digital sound technology patents will be that those patents will undoubtedly be worth billions.
Post by: Jon Hodgson on October 20, 2005, 11:36:51 AM
| Johnny B wrote on Thu, 20 October 2005 16:32 |
Jon, What a minute, while it's true that analogue is the "Gold Standard," |
WHAT analogue is your "Gold Standard"?
No two mics sound the same
No two preamps sound the same
No two desks sound the same
No two multitracks sound the same
No two 2 track masters sound the same
No two tape formulations sound the same
No two speaker models sound the same
And that's not even counting your delivery medium, which if you're going to make it vinyl, means you've got a huge bunch of variables there.
SO WHAT IS YOUR STANDARD?
Post by: Johnny B on October 20, 2005, 04:02:04 PM
Look, let's not drift off again, ok?
Either you like analogue sound quality or you don't. If you do, as I suspect you might, then it's the Gold Standard, it's the "Brass Ring," it's the "Holy Grail" that everyone has been chasing with all the plugs, models, and emulators and so on.
C'mon, you cannot be seriously trying to dispute this point.
Are we now trying to pit analogue defects vs. digital defects?
Is that a productive pursuit?
I think the question is, or ought to be:
"How do you go about improving the sound quality of digital?"
And in your answer, please include a discussion of how you would improve the formats.
For our purposes, you can leave out the "protection and anti-copying schemes."
Post by: Jon Hodgson on October 20, 2005, 04:06:19 PM
Don't ask me what type of soup, even though there are 16 million different soup recipes, and for each of those recipes you could use a thousand variations on the ingredients, it is quite obviously soup.
You either like soup or you don't, and if you like soup then you'll have to agree that it's the gold standard by which all food should be judged.
Post by: Johnny B on October 20, 2005, 04:32:09 PM
Just because I suffer from a learning disability does not mean you have Carte Blanche to make fun of a one with a handicapped brain.
Nevertheless, I would hardly call analogue sound quality "soup."
Ok, so you do not like analogue sound quality...just say that and be done with it.
Post by: stuntbutt on October 20, 2005, 04:40:39 PM
---------------------------------------------
John Katsafanas
Post by: jimmyjazz on October 20, 2005, 04:42:42 PM
| Johnny B wrote on Thu, 20 October 2005 16:32 |
Ok, so you do not like analogue sound quality...just say that and be done with it. |
DAMN but you're the insufferable type.
Why can't he like both "analog" and "digital", or at least the best each has to offer? I do. Nothing does "fat" like tape at 15 IPS. Nothing does "detail" like quality digital.
Why aren't you demanding that anyone work on improving analog? Do you think it has no deficiences?
Post by: Jon Hodgson on October 20, 2005, 04:46:25 PM
| Johnny B wrote on Thu, 20 October 2005 21:32 |
Jon, Just because I suffer from a learning disability does not mean you have Carte Blanche to make fun of a one with a handicapped brain. |
Oh, so now you're going for the moral high ground?
I'm afraid it doesn't wash, you can't act like a know-it-all, and then claim a learning disability when your knowledge and logic are called into question.
You choose to enter the kitchen, in fact you're the one who turned on the stove, don't complain about the heat.
| Johnny B wrote on Thu, 20 October 2005 21:32 |
Nevertheless, I would hardly call analogue sound quality "soup." |
I wasn't comparing analogue to soup, I was using that whole statement as a metaphor for yours.
Basically it was bullshit, it was a statement that made no real sense and gave no real information, just like yours.
You cannot have a "golden standard" unless you have a STANDARD.
When Rupert Neve designs a circuit he doesn't say "I'm going to make it sound like analogue", he has values he measures and an ideal standard he tries to approach with those measurements, for noise, bandwidth, linearity. He uses his wealth of experience and excellent brain to design and refine circuits to achieve those standards.
Saying "analogue" is the standard is as logical as saying "soup" is the standard, because analogue can be made to have any of a million different sounds, good, bad and indifferent.
| Johnny B wrote on Thu, 20 October 2005 21:32 |
Ok, so you do not like analogue sound quality...just say that and be done with it. |
I like nice sounds, whatever makes them. Since I happen to know and understand what sound is, and I also have training and experience in the design and implementation of analogue and digital electronic audio systtems, I also know that when it comes down to it, there is no "analogue" or "digital" sound.
Post by: Jon Hodgson on October 20, 2005, 05:45:57 PM
| stuntbutt wrote on Thu, 20 October 2005 21:40 |
For me, the "Gold Standard" of a recording medium is the source without the medium; In other words, listening to the mic live in the control room, not through the 2", without AD/DA. --------------------------------------------- John Katsafanas |
Now THAT is a reasonable answer.
In fact the specifics aren't important, what is important is that there ARE specifics.
I personally agree with this reference, because I don't believe that an ideal recording system would have a sound of its own, but it's also perfectly reaonable for someone to say that the reference should be a Revox B77 running at 7.5ips with Quantegy 456 tape, if that's the sound they're trying to achieve.
Hell, they can say the reference should be a dictaphone with a broken speaker for all I care, everyone has a right to their own taste.
The whole point of references is that it must be possible to REFER to them. If you want to make something more (or less) like something else, you have to be able to identify and measure the differences.
And yes "ear" people come into this, they're vital in fact, they are the ones who help engineers identify what differences are important, and what ranges are acceptable.
Post by: bobkatz on October 21, 2005, 12:08:43 AM
| dcollins wrote on Tue, 18 October 2005 05:40 |
Do you feel minimum phase filters, as used by Wolfson in new D/A products, have merit over conventional FIR implementations? DC |
They will definitely sound different! They will not be subject to the same pre-echo situation. I would be VERY interested in auditioning a 96 kHz D/A with a gentle minimum phase fiilter that's 0 dB at 20 kHz. I don't care whether it's flat above that...
BK
Post by: Johnny B on October 21, 2005, 12:08:59 AM
| Jon Hodgson wrote on Thu, 20 October 2005 22:45 |
And yes "ear" people come into this, they're vital in fact, they are the ones who help engineers identify what differences are important, and what ranges are acceptable. |
Ah, thank you for that one.
However, Jon, you first try to turn the challenge into an "analogue defect vs. a digital defect" battle...then you say there is no difference...it's the sound in the room that counts...You now have me thoroughly confused with your stand...
It's my understanding that quite a few "ear people" are not *really* happy with digital sound quality and THEY DO hear a difference between analogue and digital.
Tell me that's really news to you.
My spies may have brought me some good news. They tell me that Wolfson may be on to something...and others are working quietly to bring about some really big changes...We shall see if this turns out to be good intel or was something that was worked over in a cherry-picked fashion and spun dry in the Bush Admin. style. Ugh!
It might be a good idea to keep a close eye on the Patent Office for new filings tho'.
///
Post by: Johnny B on October 21, 2005, 12:21:50 AM
The paper gives 26 references of which about 20 refer to AES papers etc. that point to the FIR "Time Smear" problems.
Cheers.
///
Post by: Johnny B on October 21, 2005, 12:47:58 AM
| Jon Hodgson wrote on Thu, 20 October 2005 22:45 |
Now THAT is a reasonable answer. |
Yes indeed! My last post was a very reasonable answer.
Cheers
Post by: kraster on October 21, 2005, 08:06:33 AM
| Johnny B wrote on Fri, 21 October 2005 05:47 | ||
Yes indeed! My last post was a very reasonable answer. Cheers |
Now you are resorting to quantifiable phenomena just beacause it suits you. The Filter ringing issues have been well documented and the subject of many a debate.
It would seem that Wolfson have offered a possible improvement with minimum phase filters but I doubt they were stabbing in the dark when they discovered it.
Johnny, out of interest, what, sonically, is your favourite recording?
Karl
Post by: Jon Hodgson on October 21, 2005, 08:14:23 AM
| kraster wrote on Fri, 21 October 2005 13:06 | ||||
Now you are resorting to quantifiable phenomena just beacause it suits you. The Filter ringing issues have been well documented and the subject of many a debate. It would seem that Wolfson have offered a possible improvement with minimum phase filters but I doubt they were stabbing in the dark when they discovered it. Johnny, out of interest, what, sonically, is your favourite recording? Karl |
He's also using his common trick of quoting out of context, though normally he's not as blatant about it, so I guess that here he meant it as a joke.
However I don't think I've ever encountered anyone who claims English as a first language who has as much trouble actually understanding it as he does... hell most of my colleagues aren't even native English speakers and they have less trouble than him.
Personally I'm just fed up with him claiming that I and others have said things that we not only didn't say, but that cannot be deduced or extrapolated from what we said by anyone with a basic understanding of grammar and a modicum of common sense.
Post by: Johnny B on October 21, 2005, 12:49:26 PM
| kraster wrote on Fri, 21 October 2005 13:06 | ||||
...The Filter ringing issues have been well documented and the subject of many a debate. It would seem that Wolfson have offered a possible improvement with minimum phase filters but I doubt they were stabbing in the dark when they discovered it. Karl |
Not everything is discovered with a clearly mapped out plan based on some math calcs or models. C'mon, be honest, many scientific discoveries were stumbled onto due to blind luck, sometimes they find new things when they were looking for something else entirely.
Math often comes in to try to explain what they already discovered, often, as more of an afterthought or a rationalisation.
Whether they were "stabbing in the dark," doing "trial and error" or simply following up on a "hunch" makes no difference, if, at the end of the day, they got some good results.
Post by: Bill_Urick on October 23, 2005, 09:05:47 PM
| danlavry wrote on Sat, 21 May 2005 18:50 |
I would only consider looking at audio through the eyes of the very limited Fourier series, when music becomes an infinite repetitive identical cycles.... Regards Dan Lavry |
Dan,
Sorry to butt in so late in the thread, but have you listened to the radio lately?
Post by: Jon Hodgson on October 24, 2005, 04:19:03 AM
| Bill Urick wrote on Mon, 24 October 2005 02:05 | ||
Dan, Sorry to butt in so late in the thread, but have you listened to the radio lately? |
Blimey, you're right!
Maybe you've stumbled on that new branch of mathematics that Johnny B is expecting... Radio Playlist Integrals, now there must be a commercial application for those!
Post by: Andy Simpson on October 24, 2005, 07:37:39 AM
| danlavry wrote on Wed, 19 October 2005 22:30 | ||||
Johnny, You can not have both, no matter how "greedy" you want to be. You simply can not be more accurate and less accurate at the same time. You can not have 2 heads either... Regards Dan Lavry www.lavrengineering.com |
I don't want to step in here at this stage and hurl more dust into everyone's eyes, but this is not a simple question of accuracy.
Digital might be more accurate to what the mic 'hears'.
Tape might be more 'accurate' at translating this to what sound is like in reality.
A microphone IS NOT AN EAR.
There is no point struggling to accurately transmit this non-reality, as if the perfect transmission of it would make it sound natural. It won't.
Distortion is certainly required to make this any kind of representation of reality - you could take HRTF as distortion of what the mic would have heard, and a very very significant one.
The ear is a very complicated spatial encoder. Think about the shape & reflections. There is as much information encoded here as there is shape in the ear itself.
The ear is non-linear in it's 'distortions', like tape.
Perhaps tape encodes or emulates some aspects of the 'distortions' of the ear, which makes the sound more natural by distortion of the microphone signal?
Since we know that simply passing along whatever the microphone hears is guaranteed to be unnatural, at least tape/analogue impart some distortions which appear to make the sound more natural.
Perhaps this distortion is better than no attempt at the 'ear-like distortions' that we get from digital.
What Johnny said does actually make sense.
More accurate distortions please.
Love you guys.
Andy
Post by: CCC on October 24, 2005, 08:28:31 AM
| andy_simpson wrote on Mon, 24 October 2005 12:37 |
What Johnny said does actually make sense. |
why do i get the feeling that you and he are the same person?
Post by: Jon Hodgson on October 24, 2005, 10:34:48 AM
| andy_simpson wrote on Mon, 24 October 2005 12:37 | ||||||
I don't want to step in here at this stage and hurl more dust into everyone's eyes, but this is not a simple question of accuracy. Digital might be more accurate to what the mic 'hears'. Tape might be more 'accurate' at translating this to what sound is like in reality. A microphone IS NOT AN EAR. There is no point struggling to accurately transmit this non-reality, as if the perfect transmission of it would make it sound natural. It won't. Distortion is certainly required to make this any kind of representation of reality - you could take HRTF as distortion of what the mic would have heard, and a very very significant one. The ear is a very complicated spatial encoder. Think about the shape & reflections. There is as much information encoded here as there is shape in the ear itself. The ear is non-linear in it's 'distortions', like tape. Perhaps tape encodes or emulates some aspects of the 'distortions' of the ear, which makes the sound more natural by distortion of the microphone signal? Since we know that simply passing along whatever the microphone hears is guaranteed to be unnatural, at least tape/analogue impart some distortions which appear to make the sound more natural. Perhaps this distortion is better than no attempt at the 'ear-like distortions' that we get from digital. What Johnny said does actually make sense. More accurate distortions please. Love you guys. Andy |
Andy,
I'll start with the last bit first,
What Johnny said does not make sense.
Now I'll explain why...
The fact that the microphone is not an ear IS NOT A PROBLEM. You don't want a microphone to react like an ear, nor do you want a recording system or a speaker to act like an ear.
Why not? Because you'll be listening to the final sound through your ears, so if something in your recording and playback chain also acts like an ear, it would be like listening to the signal through two sets of ears.
The only way tape could be more "accurate" would be if its inaccuracies directly nulled with the inaccuracies in microphones or speakers... they don't.
If we're talking about recording as a way to reproduce the original performance in the original room (as opposed to a creative process in its own right) then the aim must be to reproduce the same air movements as accurately as possible in the required range, and then your ears will do the rest. The best way to do this is with a perfectly linear system at every stage.
Now if people are going to be listening on headphones, especially in-ear ones, then the response of the ear to these is different than to when they are open to the air, so there is justification in systems which equalize and otherwise process the sound in order to counter this difference and thus give the illusion of listening in real space... but the ideal time to do that is at playback, and either way it should be part of processing and not recording.
Post by: Johnny B on October 24, 2005, 11:47:47 AM
1. The first place to start is to accurately define exactly what portions of the frequency spectrum are important to the human being. Which parts are generally offensive? And which parts of the frequency spectrum are associated with the brain's pleasure centre?
Perhaps as a preliminary step we should obtain extremely accurate frequency range measurements of at least all the acoustic musical instruments existing in the world. This could be used to set some boundaries as to the frequency ranges of interest for the multidisciplinary testing on humans. And while CalTech Professor James Boyk did some frequency measurements of cymbals and muted trumpets, it would be worthwhile to expand upon that research and have accurate frequency knowledge of the entire family of musical instruments, and we should also make sure we have extremely accurate information on the lows. (Ha Ha) IOW, this baseline data would attempt to provide an accurate representation of what frequencies actually exist in nature.
2. Once you have sound science supporting what frequencies are actually produced naturally by the instruments, and have sound science supporting how much of those frequencies are truly important to the human's pleasure centre, then you can begin to address any of the remaining issues.
3. All the remaining issues have been discussed ad nauseum, formats, bit depths, truncation, dither, latency, thruput, comm. issues, sample rates, filters, IM, aliasing, time smear, and so on, however, these are design and engineering challenges which can be solved over time and adjusted to fit the accurate frequency measurements of the musical instruments and the accurate measurements of the human beings responses to sound.
4. Assuming we now have progressed to the point of building digital systems based upon the sound science outlined above, you should also be able to include the desired and beloved nonlinear distortions thru hardware and software, and thus, end up with a system that allows the producer and engineer to make adjustments based upon artistic choices. This is how one *might* be able to be greedy and have both.
People are certainly free to disagree or agree with the suggested approach, they are certainly free to modify or expand it in a variety of ways to improve upon it, but I think it is necessary to develop some form of consensus for a "Road Map" toward improvement.
I wish this were a small and trivial task, but I'm afraid it's much more complex than that and will require a wide variety of extremely talented experts to solve. But we have to begin somewhere, and we will need more accurate underlying data on the human being and the musical instruments' frequencies to define the "road map" toward improvement.
Perhaps others, who are far more creative, intelligent, and talented than I am, can make better suggestions toward this end. I dunno.
Post by: Jon Hodgson on October 24, 2005, 12:18:53 PM
| Johnny B wrote on Mon, 24 October 2005 16:47 |
Jon, you can call me crazy if you like, but here is the approach I would think about and see if some adjustments might be made after more discussion and more testing. 1. The first place to start is to accurately define exactly what portions of the frequency spectrum are important to the human being. Which parts are generally offensive? And which parts of the frequency spectrum are associated with the brain's pleasure centre? Perhaps as a preliminary step we should obtain extremely accurate frequency range measurements of at least all the acoustic musical instruments existing in the world. This could be used to set some boundaries as to the frequency ranges of interest for the multidisciplinary testing on humans. And while CalTech Professor James Boyk did some frequency measurements of cymbals and muted trumpets, it would be worthwhile to expand upon that research and have accurate frequency knowledge of the entire family of musical instruments, and we should also make sure we have extremely accurate information on the lows. (Ha Ha) IOW, this baseline data would attempt to provide an accurate representation of what frequencies actually exist in nature. 2. Once you have sound science supporting what frequencies are actually produced naturally by the instruments, and have sound science supporting how much of those frequencies are truly important to the human's pleasure centre, then you can begin to address any of the remaining issues. 3. All the remaining issues have been discussed ad nauseum, formats, bit depths, truncation, dither, latency, thruput, comm. issues, sample rates, filters, IM, aliasing, time smear, and so on, however these are design and engineering challenges which can be solved over time and adjusted to fit the accurate frequency measurements of the musical instruments and the accurate measurements of the human beings responses to sound. 4. Assuming we now have progressed to the point of building digital systems based upon the sound science outlined above, you should also be able to include the desired and beloved nonlinear distortions thru hardware and software, and thus, end up with a system that allows the producer and engineer to make adjustments based upon artistic choices. This is how one *might* be able to be greedy and have both. People are certainly free to disagree or agree with the suggested approach, they are certainly free to modify or expand it in a variety of ways to improve upon it, but I think it is necessary to develop some form of consensus for a "Road Map" toward improvement. I wish this were a small and trivial task, but I'm afraid it's much more complex than that and will require a wide variety of extremely talented experts to solve. But we have to begin somewhere, and we will need more accurate underlying data on the human being and the musical instruments' frequencies to define the "road map" toward improvement. Perhaps others, who are far more creative, intelligent, and talented than I am, can make better suggestions toward this end. I dunno. |
Congratulations, you've basically described what we've been doing for decades. I don't know why you're so obsessed with Boyk's work, I for one already new that instruments have harmonics that go way beyond the 20k boundary, nobody has ever claimed they don't. I'm sure that most people you argue with on here also knew that... in fact you're the only one who seems to find it news. Many objects radiate in the ultravlolet and infrared regions, but nobody takes that to mean that cameras need to capture that radiation.
You've also shown why digital audio has an inherent advantage over recording to analogue formats, because it is pretty well impossible to ever get them to the point where they are distortion free to the limits of human hearing.
Post by: Johnny B on October 24, 2005, 12:36:32 PM
Yeah, I've heard the "light" example thrown around before, I do not think it's a good analogy.
You are not seriously arguing that we already know *everything* we need to know, are you? I do not think we have really good data for the "square one" baseline yet.
Some of the researchers findings have only been published as recently as the beginning of the new century.
As a result, we now know that HFC's outside of the so-called audible range affect the brain's pleasure centre.
However, to get even more accurate data, will will need the application of a multidisciplinary approach. IIRC, you more or less agreed to the necessity of using of a wide variety of experts to do more study and so on.
And are you now suggesting that having accurate measurements of the frequencies produced naturally by musical instruments is not extremely useful information?
I thought you liked detail, definition, specificity, precision, and accurate underlying data, am I wrong?
Post by: C-J on October 24, 2005, 01:34:20 PM
| Johnny B wrote on Mon, 24 October 2005 19:36 |
As a result, we now know that HFC's outside of the so-called audible range affect the brain's pleasure centre. |
Johnny,
Could you please explain how this logically connects to "better sounding" analog tape, which cannot reach up to "that HFC's outside of the so-called audible range"?
C.J.
Post by: Jon Hodgson on October 24, 2005, 01:45:27 PM
| Johnny B wrote on Mon, 24 October 2005 17:36 |
Jon, Yeah, I've heard the "light" example thrown around before, I do not think it's a good analogy. |
And why is it not a good analogy to illustrate that for reproduction of the experience there is no need to record what is outside of a human ability to detect?
You yourself said it was important to know what humans can detect.. why is this? Because then you know what to record, the fact that instruments output frequencies outside those ranges is not important (whether that range of human perception is up to 10k or 1000k)
| Johnny B wrote on Mon, 24 October 2005 17:36 |
You are not seriously arguing that we already know *everything* we need to know, are you? I do not think we have really good data for the "square one" baseline yet. Some of the researchers findings have only been published as recently as the beginning of the new century. |
Hmmm, and once again you're making claims that I'm arguing something I haven't. Learning more about the mechanisms of hearing is always useful, even if it doesn't lead a wider range of required frequencies to capture (and the more I look into it the more I feel that any effects induced by frequencies beyond the recognized audible range, should they even exist, could be replicated more simply, and more controlably in manners other than attempting to record them.) it can certainly lead to better compression methods, and new and interesting processing and synthesis methods.
| Johnny B wrote on Mon, 24 October 2005 17:36 |
As a result, we now know that HFC's outside of the so-called audible range affect the brain's pleasure centre. |
I've seen one paper, I have not seen the papers where the experiments have been replicated and critiqued by others. Until the research has been verified and repeated, and the experiments scrutinized for any possible reason the results might have occured outside of the hypothesis, then we don't KNOW anything, we simply have reason to suspect.
| Johnny B wrote on Mon, 24 October 2005 17:36 |
However, to get even more accurate data, will will need the application of a multidisciplinary approach. IIRC, you more or less agreed to the necessity of using of a wide variety of experts to do more study and so on. |
What is it with you and your multidisciplinary approach???
Do you think you invented it or something?
Who do you think came up with the 20-20k range in the first place? Do you think it was electronic engineers? Do you think it was "Digital Math Scoundrels"?? Do you think that this research has not been tested thousands of times by experts who's interests were purely the response of the ear and brain who had nothing to do with audio recording?
Think again.
For an example of what has been going on for a few decades, in fact probably back to the start of recorded audio, take a paper by a former colleague of mine, Dr Alastair Sibbald
http://www.sensaura.co.uk/whitepapers/pdfs/dev005.pdf
In it you'll find acoustics, physics, mathematics, physiology, neurology, it discusses the transmission of sound through various mediums, the acoustics resulting from the shape of the head, the position of the ears, the shape of the ears, the way the pinnae in the ear work, how our brains decode this information... is that multidisciplinary enough for you?
Congratulations, you've now discovered what most of us have known for years, that working out how to make systems which interact with humans requires more than one discipline, now drop the bold, it's just irritating and condescending.
| Johnny B wrote on Mon, 24 October 2005 17:36 |
And are you now suggesting that having accurate measurements of the frequencies produced naturally by musical instruments is not extremely useful information? I thought you liked detail, definition, specificity, precision, and accurate underlying data, am I wrong? |
I do like detail, but
1) I didn't say that the information was neccessarily useless, I just said that IT'S NOT NEWS, just because you didn't know that instruments created those sorts of harmonics before reading Boyk's paper DOESN'T MEAN WE'RE ALL AS IGNORANT.
2) If your interest is in a recording medium then knowing the range of frequencies that something produces is only useful if that range falls within the boundaries of hearing, and you can then take advantage of this to capture a smaller range. Hardly any use in a generic system, but perhaps if you're trying to compress. However this doesn't look very likely, in fact it looks nigh on impossible.
So the information is interesting, but seems of limited use in devising the parameters of a recording system.
Post by: Johnny B on October 24, 2005, 03:13:06 PM
From there, you can begin to construct test suites on the human response and how those specific frequencies, which, are now known, because at that future point we will have measured them with precision and not guess work, affect things like heart rate, breathing, and the pleasure centres of the brain.
I remember talking to one researcher who said that there was some evidence that if you push 5 to 10Hz at people you can change their heart rate, this of course, may account, in part, for some people seemingly liking bass-heavy content.
But the point is, having accurate measurements of the frequencies produced naturally by musical instruments can provide a basis for further tests on humans and how those specific frequencies affect the the pleasure centres in the brain.
I see nothing at all wrong with having more accurate data and more accurate measurements of the frequencies produced by, at a minimum, the acoustic musical instruments of the world. This is why the work of CalTech's Professor James Boyk is so important, it opens the door for more accurate measurements of acoustic instruments, more knowledge, and more accurate underlying data.
Post by: danlavry on October 24, 2005, 03:20:58 PM
| Johnny B wrote on Mon, 24 October 2005 17:36 |
Jon, Yeah, I've heard the "light" example thrown around before, I do not think it's a good analogy. |
Yes, I stated that analogy. I did not "throw it around" and your use of such phrases does not reduce the validity of that analogy.
In fact, it is very inappropriate to "just state" that you do not think the analogy is good. You must explain your position as to why it is not a good analogy. Anything less is just a waste of space, because (you see I am explaining my position) we are having a discussion based on facts and knowledge, thus there is no room for a "stand alone" statement such as "I do not think it is good" without a follow up.
Regards
Dan Lavry
www.lavryengineering.com
Post by: jimmyjazz on October 24, 2005, 03:27:21 PM
Post by: Jon Hodgson on October 24, 2005, 03:49:01 PM
| jimmyjazz wrote on Mon, 24 October 2005 20:27 |
AGAIN, what makes you think much of the work you are "proposing" has not already been carried out? |
Well obviously if it had been done then it would overwhelmingly support Johnny B's preconceptions that sampling requires massively high sample rates and longer word lengths.
</sarcasm>
Post by: minister on October 24, 2005, 03:57:32 PM
| Johnny B wrote on Mon, 24 October 2005 14:13 |
I think you have to start at square one and have extremely accurate measurements of the musical instruments natural frequencies. |
| Johnny B wrote on Mon, 24 October 2005 14:13 |
From there, you can begin to construct test suites on the human response and how those specific frequencies which are now known, because at that future point we will have measured them with precision and not guess work, affect things like heart rate, breathing, and the pleasure centres of the brain. |
| Johnny B wrote on Mon, 24 October 2005 14:13 |
But the point is, having accurate measurements of the frequencies produced naturally by musical instruments can provide a basis for further tests on humans and how those specific frequencies affect the the pleasure centres in the brain. |
| Johnny B wrote on Mon, 24 October 2005 14:13 |
I see nothing at all wrong with having more accurate data and more accurate measurements of the frequencies produced by, at a minimum, the acoustic musical instruments of the world. |
Post by: CCC on October 24, 2005, 04:10:06 PM
| jimmyjazz wrote on Mon, 24 October 2005 20:27 |
AGAIN, what makes you think much of the work you are "proposing" has not already been carried out? |
In the unlikely event that your question is not purely rhetorical allow me to toss out an answer; the thing(s) that make him think that this work has not been carried out are ignorance, confusion, misinformation, wilful blindness - some of the above, or all of the above. I think y'all need your heads examined for continuing to attempt to render thoughtful answers, frankly.
Post by: Johnny B on October 24, 2005, 05:05:52 PM
If people do not think we should gather more precise information and then perform some more human testing, well, there's nothing I can do to change their minds.
I do, however, find the position against gathering more precise facts and doing more precise testing on human beings somewhat puzzling???
Post by: CCC on October 24, 2005, 05:21:27 PM
| Johnny B wrote on Mon, 24 October 2005 22:05 |
No, it's not that some work has not be done in the past, it's that the body of knowledge is not complete. I'm simply arguing for gathering more precise facts, obtaining more precise knowledge, and then doing some more precise testing on humans. If people do not think we should gather more precise information and then perform some more human testing, well, there's nothing I can do to change their minds. I do, however, find the position against gathering more precise facts and doing more precise testing on human beings somewhat puzzling??? |
There are many, many people in this field who do considerable research into the nature of musical instruments, acoustics, the psychology and physiology of hearing, electronics, digital audio, loudspeaker design... and so on. You name it, there's a busload of white-coat-with-pocket-protector-type-people who have dedicated decades of their lives to researching it. Nobody doubts that we need more precise information on everything pertaining to the physical reality of acoustical spaces and instruments, more precise information on the nature of hearing, more development in all forms of recording technology. This is an ongoing endeavour for a great many people. A few, such as the esteemed Dan Lavry and some of the posters on this forum, have considerable expertise in some of the aforementioned categories.
The problem, as I see it, is that some people round these parts don't like taking XYZ for an answer...and instead prefer posting endless cycles of tail-chasing generalities and platitudes.
Post by: Johnny B on October 24, 2005, 10:41:47 PM
As far as the precise measurements a la Professor Boyk of all the acoustic musical instruments in the world and their precisely measured frequency range is concerned, I'll ask people to please post a link where one can find all this information and underlying data. I'm asking for complete and well-measured research findings here, so if anyone does have a link to such precise frequency measurements, please be kind enough to post that link. Thanks.
Post by: Augustine Leudar on October 24, 2005, 11:10:21 PM
Post by: Augustine Leudar on October 24, 2005, 11:14:21 PM
| Johnny B wrote on Thu, 20 October 2005 21:02 |
Either you like analogue sound quality or you don't. If you do, as I suspect you might, then it's the Gold Standard, it's the "Brass Ring," it's the "Holy Grail" that everyone has been chasing with all the plugs, models, and emulators and so on. |
My favourite software is programs which do not try to immitate analogue effects but rather explore new sound processing techniques that analogue is completely incable of , such as granular synthesis and spectral morphing.
Post by: dcollins on October 24, 2005, 11:21:44 PM
| Johnny B wrote on Mon, 24 October 2005 19:41 |
data. I'm asking for complete and well-measured research findings here, so if anyone does have a link to such precise frequency measurements, please be kind enough to post that link. Thanks. |
http://asa.aip.org/
You are a member of the ASA, aren't you?
DC
Post by: CCC on October 24, 2005, 11:37:20 PM
| dcollins wrote on Tue, 25 October 2005 04:21 | ||
http://asa.aip.org/ You are a member of the ASA, aren't you? DC |
Damn, you beat me to it.
May I also suggest http://www.aes.org/journal/
Also, Johnny, here's a couple of points for you to ignore;
- 'measurements of all the musical instruments in the world'. This is rather broad. I think we already know that the notional limits of 20-20k are the purported limits of human perception, not to be confused with the spectral limits of any physical system. You cite authorities for the proposition. Do you really need to see the frequency response of a dulcimer, taiko drum, accordion, and thumb piano to be totally sure that musical instruments do produce substantial harmonic content above 20k?
- I'm no expert on research, but I'm currently working on a second graduate degree part-time. Recently a professor made a very obvious point - a point that needs no repeating (I think); he said don't go 'round thinking that if you Google it, it will come. Many if not most authorities are either not available in e-format or they are available only from password protected sites. If your litmus test for in depth research is whether you can find data with a search engine, then you just aren't doing the work you need to be doing.
Post by: jimmyjazz on October 25, 2005, 10:38:13 AM
Your appeal to authority is transparent. Why are you so reluctant to give credence to differing opinions, especially when there are dramatically more "authorities" who vehemently disagree with [said experts]? No doubt the course of science is periodically shaped by the renegade who defies conventional wisdom, but far more often than not, the majority rules for a reason.
Post by: Johnny B on October 25, 2005, 02:31:16 PM
John, yeah I want it all...I want the "full frequency spectrum" for all the instruments...I want to know what frequencies are naturally produced...all of them...and I want it all measured as accurately and precisely as possible.
And let us consider this, having an accurate and comprehensive database such I'm suggesting can be useful for other endeavors...For example, it could be useful for plug writers, synth developers, and simply tweaking eq's as well, and of course, looking more closely to see exactly how the naturally produced frequencies interact with one another and how they affect, either negatively or positively, the brain's pleasure centers. Is it hard to accurately measure all the world's acoustic instruments in a similar manner that was utilised by CalTech's Professor James Boyk? Yes!...It will probably take a substantial grant to fund this research and develop an accurate underlying data base.
Post by: Johnny B on October 25, 2005, 02:37:28 PM
One must have a sense if humour in these things...right? Post by: jimmyjazz on October 25, 2005, 03:25:41 PM
Post by: Johnny B on October 25, 2005, 03:40:07 PM
Experts at what?...negative thinking...nay saying...practitioners of the "It can't be done religion"..."We really don't really wanna know 'cuz it might upset our applecart cult" or maybe it's the old "NIH" negative thinking at work, as in, the old "Not Invented Here" syndrome...
Look, there is ZERO REASON to hold back on gathering more precise data and performing more research.
None, Nada, Nix....
I'm just arguing for the acquisition of some more accurate data and placing it in a freely accessible repository ...and people are upset?...That makes no sense...
"Yes, it's very illogical Captain Kirk"...
"Yes Spock, you are quite right. It *is* illogical to argue against gathering more accurate data."
Post by: CCC on October 25, 2005, 04:08:31 PM
| Johnny B wrote on Tue, 25 October 2005 20:40 |
Look, there is ZERO REASON to hold back on gathering more precise data and performing more research. None, Nada, Nix.... I'm just arguing for the acquisition of some more accurate data and placing it in a freely accessible repository ...and people are upset?...That makes no sense... |
There is zero reason to keep harping on this. It makes no sense.
Post by: Ronny on October 25, 2005, 06:52:35 PM
Captain Kirk, it is illogical to repeat seeking data that is already known and understood.
Post by: Andy Simpson on October 25, 2005, 07:34:33 PM
| Jon Hodgson wrote on Mon, 24 October 2005 15:34 |
Andy, Now I'll explain why... The fact that the microphone is not an ear IS NOT A PROBLEM. You don't want a microphone to react like an ear, nor do you want a recording system or a speaker to act like an ear. Why not? Because you'll be listening to the final sound through your ears, so if something in your recording and playback chain also acts like an ear, it would be like listening to the signal through two sets of ears. The only way tape could be more "accurate" would be if its inaccuracies directly nulled with the inaccuracies in microphones or speakers... they don't. If we're talking about recording as a way to reproduce the original performance in the original room (as opposed to a creative process in its own right) then the aim must be to reproduce the same air movements as accurately as possible in the required range, and then your ears will do the rest. The best way to do this is with a perfectly linear system at every stage. Now if people are going to be listening on headphones, especially in-ear ones, then the response of the ear to these is different than to when they are open to the air, so there is justification in systems which equalize and otherwise process the sound in order to counter this difference and thus give the illusion of listening in real space... but the ideal time to do that is at playback, and either way it should be part of processing and not recording. |
Ask yourself this question:
What if the ear encodes information from natural sound waves that the microphone cannot?
And, if the microphone does not encode sound like an ear, then by definition information IS LOST.
This information CANNOT be put back or created by the speaker.
Whether we think this information is necessary or relevant is not the question, it IS LOST.
So perhaps the non-linear distortions of tape in some way mimick this natural ear encoding process in such a way as to make the sound appear more natural after the tape distortion process.
Like hand-coloured cells of a black & white film - it's not 'real' but it looks more 'natural' than black & white.
Current research is a LONG way from fully understanding the ear/brain.
So also are we a LONG way from being able to dismiss the distortions of tape as being less natural sounding than no distortions of a microphone signal.
Andy
Post by: Jon Hodgson on October 25, 2005, 07:47:10 PM
| andy_simpson wrote on Wed, 26 October 2005 00:34 | ||
Ask yourself this question: What if the ear encodes information from natural sound waves that the microphone cannot? And, if the microphone does not encode sound like an ear, then by definition information IS LOST. This information CANNOT be put back or created by the speaker. Whether we think this information is necessary or relevant is not the question, it IS LOST. So perhaps the non-linear distortions of tape in some way mimick this natural ear encoding process in such a way as to make the sound appear more natural after the tape distortion process. Like hand-coloured cells of a black & white film - it's not 'real' but it looks more 'natural' than black & white. Current research is a LONG way from fully understanding the ear/brain. So also are we a LONG way from being able to dismiss the distortions of tape as being less natural sounding than no distortions of a microphone signal. Andy |
Hi Andy,
Having worked in the world of 3d sound, I know a little about it, and the workings of the ear and brain in this respect. As an electronics engineer I also know something of the distortions that go on in tape.
From my knowledge of both these factors, I tend to find your theory extremely unlikely. The problem is that although the ear is a complex directional device, it's basically a directional EQ system, which makes it largely linear, if you get into distortion you're probably damaging something.
To somehow recreate the effects of the ear, tape distortion or other effects occuring in the analogue recording system would somehow have to affect sound coming from different directions in different ways, but since we've already converted this multidimensional dimensional signal to a two dimensional one (level and time), this information does not exist for the tape to magically do this trick.
Basically if the microphone has failed to capture information that the ear would use, tape cannot create it.
Post by: CCC on October 25, 2005, 07:51:22 PM
| andy_simpson wrote on Wed, 26 October 2005 00:34 |
Ask yourself this question: What if the ear encodes information from natural sound waves that the microphone cannot? |
You already know this. The ear, the human head, the torso, all interfere in the soundfield. Actually, so does the body of a microphone. But, of course, the interference is different. The effect on frequency response of the pinnae is well documented. And, of course, certain manufacturers publish polar patterns at various frequencies that gives us some idea about how the body of the mic and the size of the diaphragm affects pickup.
| andy_simpson wrote on Wed, 26 October 2005 00:34 |
And, if the microphone does not encode sound like an ear, then by definition information IS LOST. |
Yes.
| andy_simpson wrote on Wed, 26 October 2005 00:34 |
This information CANNOT be put back or created by the speaker. |
Of course.
| andy_simpson wrote on Wed, 26 October 2005 00:34 |
Whether we think this information is necessary or relevant is not the question, it IS LOST. |
Yep. It's a point that's worth making twice.
| andy_simpson wrote on Wed, 26 October 2005 00:34 |
So perhaps the non-linear distortions of tape in some way mimick this natural ear encoding process in such a way as to make the sound appear more natural after the tape distortion process. |
"Perhaps", but so very, very, very highly unlikely as to be easily dismissed. However, if you have some kind of empirical evidence that the the transfer function of analog tape is somehow comparable to head related transfer functions I'd like to see it. Can you do this?
| andy_simpson wrote on Wed, 26 October 2005 00:34 |
Current research is a LONG way from fully understanding the ear/brain. |
That's for sure.
| andy_simpson wrote on Wed, 26 October 2005 00:34 |
So also are we a LONG way from being able to dismiss the distortions of tape as being less natural sounding than no distortions of a microphone signal. |
Could you rephrase this in the form of a sentence?
Post by: Augustine Leudar on October 26, 2005, 06:47:05 AM
Have you ever heard of Binaural stereo microphones ? They fit in the ears a bit like walkman headphones. That means they capture the directional information for the sound as it bounces off the ear lobe. Humans positions sound by how it bounces off the earlobe (which may also make some frequency adjustments).
Johnny B :
You claim that digital plugins are just trying to imitate analogue ones yet you didnt reply to my comment about digital sound processing which analogue is completely incable of such as Granular synthesis, Spectral morphing or even Convolution reverb .
Perhaps Ive got the wrong end of the stick but would be curious as to what you make of these completely revolutionary soundprocessing techniques (well revolutionary for me anyway).
Post by: CCC on October 26, 2005, 07:37:23 AM
| Augustine Leudar wrote on Wed, 26 October 2005 11:47 |
Ummm hope im not butting into the more knowledgable members conversations here..... Have you ever heard of Binaural stereo microphones ? They fit in the ears a bit like walkman headphones. That means they capture the directional information for the sound as it bounces off the ear lobe. Humans positions sound by how it bounces off the earlobe (which may also make some frequency adjustments). |
yes, the subject has come up a few times
Post by: AndreasN on October 26, 2005, 10:36:00 AM
If ultrasonics are so important.. how come people on stage don't object to reinforced music? 'hey! there's some frequencies I can't hear missing in the overheads!"
Don't you think someone would have noticed if 20K plus really was important to the feeling of the instruments?
Sort of odd that an ultrasonic missionaire also literally screams out his love for annie! You got two quests going on here, one proclaiming tape is infinitely better than digital, another one claiming that digital need further extension into the high frequencies. How come? What's your link between these two opposing ideas?
Post by: Andy Simpson on October 26, 2005, 12:39:15 PM
| John Sorensen wrote on Wed, 26 October 2005 00:51 | ||
"Perhaps", but so very, very, very highly unlikely as to be easily dismissed. However, if you have some kind of empirical evidence that the the transfer function of analog tape is somehow comparable to head related transfer functions I'd like to see it. Can you do this? |
The owness is on you to prove that it ISN'T related - since you wish to dismiss it, but to do that we'd have to understand the ear completely.
Otherwise, the only relevant fact we have is that we DO like tape.
Anyway, yes I am looking at the link between the transfer function of tape & HRTF but it is only indirectly related to my work and I won't be publishing the actual work until early next year.
At present I can say that I'm on my way to (indirectly) proving the actual link.
I can prove that there is a link right now, but I can't prove that it's relevant to hearing just yet!
Sorry to be vague......
Andy
Post by: Jon Hodgson on October 26, 2005, 12:48:10 PM
| andy_simpson wrote on Wed, 26 October 2005 17:39 | ||||
The owness is on you to prove that it ISN'T related - since you wish to dismiss it, but to do that we'd have to understand the ear completely. Otherwise, the only relevant fact we have is that we DO like tape. Anyway, yes I am looking at the link between the transfer function of tape & HRTF but it is only indirectly related to my work and I won't be publishing the actual work until early next year. At present I can say that I'm on my way to (indirectly) proving the actual link. I can prove that there is a link right now, but I can't prove that it's relevant to hearing just yet! Sorry to be vague...... Andy |
Erm...
How can a single channel recording have a link to a Head Related Transfer Function, which by definition is highly directional, an to work properly requires two outputs, for the left and right ears?
Post by: CCC on October 26, 2005, 02:21:26 PM
| andy_simpson wrote on Wed, 26 October 2005 17:39 | ||||
The owness is on you to prove that it ISN'T related - since you wish to dismiss it, but to do that we'd have to understand the ear completely. |
The word is 'onus'.
As far as understanding the ear completely goes, you should know that this is a silly trollish statement. Complete understanding of physics and psychophysics is surely well beyond our capabilities. The scientists that I know don't ever really claim that they have 'complete' understandings of things. I do believe that at best they talk about having working models, insight, theories, wisdom, etc....but complete understanding is right out. Just because that is the case does not mean that we can or should just pull implausible ideas out of our proverbial asses and pretend that they're 'good' ideas just because the body of knowledge is imcomplete.
The ear is a 'directional equalizer'. Depending on the angle of incidence it imparts different, complex filtering on sound waves en route to the ear drum, which assists the ear brain mechanism in determining the origin of a sound in 360 degree space.
When we calibrate an analog machine we are setting nominal reference level and frequency response (and bias, of course). The means for doing this is a bit crude - a low frequency, a mid frequency, a high frequency. If you have ever swept an oscillator while looking at the meters on an analog machine you will see that they are rarely, if ever, flat across the audible spectrum.
What do we like about tape? Various things, but key among the attributes is tape compression. When you record hot the response of the tape becomes non-linear. The non-linear relationship between input and output first becomes pronounced in the high end. The precise details differ with brand of tape, and ips.
This is a quick and basic overview of the role of the pinnae and the frequency response characteristics of tape. It is not reasonable to speculate about a relationship.
| andy_simpson wrote on Wed, 26 October 2005 17:39 |
Otherwise, the only relevant fact we have is that we DO like tape. |
uh, ok.
| andy_simpson wrote on Wed, 26 October 2005 17:39 |
Anyway, yes I am looking at the link between the transfer function of tape & HRTF but it is only indirectly related to my work and I won't be publishing the actual work until early next year. |
I can't wait.
| andy_simpson wrote on Wed, 26 October 2005 17:39 |
At present I can say that I'm on my way to (indirectly) proving the actual link. I can prove that there is a link right now, but I can't prove that it's relevant to hearing just yet! |
Ok, well that doesn't make sense - oh well, whatcha gonna do...
Post by: Johnny B on October 26, 2005, 03:06:10 PM
| John Sorensen wrote on Wed, 26 October 2005 19:21 |
Complete understanding of physics and psychophysics is surely well beyond our capabilities. The scientists that I know don't ever really claim that they have 'complete' understandings of things. |
Yes!
And to get a more accurate database, which should support further advances in digital sound, will require more advanced research using an appropriate multidisciplinary approach bringing together the collaborative efforts of many people having a wide variety of expertise in many different scientific disciplines. No one person can "know it all."
As far as the reasons why some people like analogue sound better than digital sound, the answers *may* involve some of analogue's so-called defects, but if the goal is to improve digital sound quality, then the defects in digital may be far more important to focus on correcting. But beyond that, the closer you get to making digital more accurate and precise, more compatible and accommodating to any new research learned from the multidisciplinary team's research, the closer you should be to making everything more realistic including plugs, models, and emulations because you are now closer to human reality, perception, and experiences. This should occur because you've now corrected your underlying database with more precision and more accurate data.
It's far different to be at the symphony hall listening to the orchestra play than to hear the same performance on record, tape, or CD. With the kinds of speaker arrays that Bill (aka "Level') is suggesting, the kinds of multiple spherical mic set-ups, then eventually this too should result in a more realistic recording and playback system.
Still, no matter how one wants to view the future, the AD/DA has to be adjusted and the technology brought forward to accommodate new research findings on the human's total response to sound. After all, recorded music is all about trying to get appropriate emotional responses and touching those pleasure centres in the brain. The researchers scientific findings regarding HFC's out of the so-called audible range having a positive effect upon the brain's pleasure centres must be taken into account and cannot be summarily dismissed.
Post by: jimmyjazz on October 26, 2005, 03:15:43 PM
| Johnny B wrote on Wed, 26 October 2005 15:06 |
And to get a more accurate database, which should support further advances in digital sound, will require more advanced research using an appropriate multidisciplinary approach bringing together the collaborative efforts of many people having a wide variety of expertise in many different scientific disciplines. |
Engineering is an exercise in tradeoffs, whether one is trading one characteristic for another or simply prioritizing the laundry list of "things to work on next".
I make those tradeoffs every day, but I'm a mechanical engineer working in another field, so I have little insight into the priorities of those working in digital audio. You clearly don't, either. Whether or not a library of instrument spectra is deemed critical by those "in the know" is beyond me . . . but I think I'll just let them make that decision. Why don't you?
You don't even know enough about the subject to be dangerous, Johnny, and that's saying something.
Post by: Johnny B on October 26, 2005, 03:59:07 PM
I suppose you are on the side of being "anti-research."
Post by: Jon Hodgson on October 26, 2005, 04:50:44 PM
| Johnny B wrote on Wed, 26 October 2005 20:59 |
I suppose you are on the side of being "anti-research." |
Do you have any idea how childish you seem when you come out with crap like this?
Post by: Augustine Leudar on October 26, 2005, 05:04:47 PM
| Johnny B wrote on Wed, 26 October 2005 20:06 |
the closer you should be to making everything more realistic including plugs, models, and emulations because you are now closer to human reality, perception, and experiences. This should occur because you've now corrected your underlying database with more precision and more accurate data. recorded music is all about trying to get appropriate emotional responses and touching those pleasure centres in the brain. |
You seem to be insistant that digital plugins etc should always imitate analogue. This is not the case. At least for creative applications.
You still havent responded to two of my posts now referring to convolution reverb, spectral morphing software, timestretching without pitch shifting and granular synthesis. They are all unique processeses that do completely new things to sound (apart from convolution reverb) and in no way imitate any analogue technique. There is no possible analogue equivelant to these processes yet they open up huge arenas for creative potential.
Post by: Johnny B on October 26, 2005, 05:26:22 PM
Those were but examples, sure you could have lots of goodies, but before we have any of it, that advanced research must move forward and the AD/DA's must also move forward and be adjusted to the new advanced research findings.
In a very real sense there is a "Pro-Advanced Research" vs. an "Anti-Any Further Research" battle going on.
I suspect some people are afraid. perhaps terrified, of what the experts from a multidisciplinary team publish in their future findings.
Underlying are the irrational, emotional, and angry outbursts and personal attacks against me, which BTW I tend to ignore as being childish, may be the ugly human motivator...it's called "Fear."
No real scientist worth their salt would ever argue against further advanced research, and yet, that seems to be the fear-based position of the "anti-further research" mob.
Post by: Jon Hodgson on October 26, 2005, 05:32:10 PM
| Johnny B wrote on Wed, 26 October 2005 22:26 |
Augustine, Those were but examples, sure you could have lots of goodies, but before we have any of it, that advanced research must move forward and the AD/DA's must also move forward and be adjusted to the new advanced research findings. |
We already have them, were you asleep when that happened?
And what do you say is wrong with AD/DA converters? And don't say "how they perform" or "how they sound"... be specific... you'd better state which units you're talking about, and what you're comparing them to whilst you're at it.
Post by: Johnny B on October 26, 2005, 05:47:03 PM
Moreover, I think you will have to come up with some convincing rationale to justify the "Anti-Any-Further-Advanced Research" position. Personal attacks on me simply will not do the trick. You will need to present some hard evidence and convincing data and show your chain-of-reasoning if you can ever even hope to hold back further advanced research by the multidisciplinary teams of experts.
Post by: jimmyjazz on October 26, 2005, 05:52:33 PM
| Johnny B wrote on Wed, 26 October 2005 17:26 |
No real scientist worth their salt would ever argue against further advanced research, and yet, that seems to be the fear-based position of the "anti-further research" mob. |
Pardon my French, but what the fuck are you talking about? There is no "anti-further research mob". What a ridiculous red herring you've put forth.
Read it slowly: nobody is against "further research". The thing is, you don't seem to know
a) what research has already been performed
b) what other research might be useful
I swear, you're either a troll, 15 years old, or just not quite all there. Whatever. I'm done with you, and I hope everyone else is, too. I'm embarassed to have participated in these threads at all.
Post by: Augustine Leudar on October 26, 2005, 06:12:04 PM
| Johnny B wrote on Wed, 26 October 2005 22:26 |
Augustine, Those were but examples, sure you could have lots of goodies, but before we have any of it, that advanced research must move forward and the AD/DA's must also move forward and be adjusted to the new advanced research findings. In a very real sense there is a "Pro-Advanced Research" vs. an "Anti-Any Further Research" battle going on. I suspect some people are afraid. perhaps terrified, of what the experts from a multidisciplinary team publish in their future findings. Underlying are the irrational, emotional, and angry outbursts and personal attacks against me, which BTW I tend to ignore as being childish, may be the ugly human motivator...it's called "Fear." No real scientist worth their salt would ever argue against further advanced research, and yet, that seems to be the fear-based position of the "anti-further research" mob. |
I dont think so. I dont think anyone would complain if higher res equipment were available, its hardly job threatening (like proposing nuendo as an industry standard instead of protools hehe unrealistic I know). I cant quite work you out, you talk of the progress of science but tout Analogue over Digital, a weird contradiction.
I think people are annoyed with you not because you promote new technology but because you insulted their profesionalism and intelligence even though there are obviously some highly skilled individuals on this site.
You bring up some interesting subjects and Im sure if youd presented your opinions in a different way and for different reasons they could spark some interesting discussions instead of flame wars.
Post by: Jon Hodgson on October 26, 2005, 06:21:21 PM
| Johnny B wrote on Wed, 26 October 2005 22:47 |
No, that is not my burden here. I do not have to prove to you that I do not like the way digital sounds. YOU, OTOH, have the burden of presenting unassailable proof beyond a reasonable doubt that digital sound has advanced as far as it will ever go. I do not think you can meet that burden. |
No I don't, because I don't make that claim.
What I do suspect though, as many other people who have read your posts have already suggested, is that you have no idea what quality equipment sounds like (digital or analogue), you claim to hear anomolies and distortions that have been eliminated from decent converters for a decade, and generally, you don't know what you're talking about.
Post by: minister on October 26, 2005, 06:24:51 PM
| Johnny B wrote on Wed, 26 October 2005 16:26 |
In a very real sense there is a "Pro-Advanced Research" vs. an "Anti-Any Further Research" battle going on. |
basically, you come in here and shoot from the hip like a teenage cowboy drunk on on his first night of whiskey and then go run behind all the smoke. it is pretentious.
there was a guy like you in some philosophy classes i had. the professor called him a name that begins with an "a" and ends with an "e" and is 7 letters long.
you don't listen. you don't read what is written to you. you don't hear what people are saying to you. but you sure do shoot off your guns!
but yes, why don't we all just stop talking to him and have a decent conversation where you can all edjamacate ME! for a change.
Post by: Jon Hodgson on October 26, 2005, 06:33:55 PM
| minister wrote on Wed, 26 October 2005 23:24 |
but yes, why don't we all just stop talking to him and have a decent conversation where you can all edjamacate ME! for a change. |
You're right, and I think we should start with some vital science for both recording engineers and DSP programmers.
So, does anyone out there know...
How do you make a good pizza?
Post by: Level on October 26, 2005, 06:37:25 PM
| Quote: |
How do you make a good pizza? |
Pick up the phone is how I make them! I know a guy from Jersey that opened a shop. Yes..they are 30 bucks but they are the definitive pizzas...(and he will even put the damned fish on it if you ask!)
Post by: Jon Hodgson on October 26, 2005, 06:41:52 PM
| Level wrote on Wed, 26 October 2005 23:37 | ||
Pick up the phone is how I make them! I know a guy from Jersey that opened a shop. Yes..they are 30 bucks but they are the definitive pizzas...(and he will even put the damned fish on it if you ask!) |
Sounds great,
but does he deliver to Belgium... and will I get my money back if it's cold?
Post by: Johnny B on October 26, 2005, 06:44:57 PM
Anyway, the problems in digital apply across the board for the most part, sure you can have some special tweaks and turns on the analogue part of it, but still the digital problems are there.
Your position seems to be like the techs who come back with a "No Trouble Found" report which turns out to be all wrong upon closer examination.
Look, if you are happy with the way things are, that's fine with me. Those who are perfectly satisfied with the way things are can pitch their tent in the "Do-Nothing Camp." The "Do-Nothing Camp" is meant to apply to those who want no changes at all.
But I thought that just about everyone, at least, agreed that there was room for digial sound to improve.
Further advanced research using teams of multidisciplinary experts seems like a perfectly reasonable and logical approach to use on the road to improvement.
Post by: Jon Hodgson on October 26, 2005, 06:51:37 PM
| Johnny B wrote on Wed, 26 October 2005 23:44 |
Anyway, the problems in digital apply across the board for the most part, sure you can have some special tweaks and turns on the analogue part of it, but still the digital problems are there. |
GREAT!!
They're across the board
So you don't have to mention any specific units
You can just mention the specific problems, and what blind A/B tests you do to detect them, and what your reference would be if there is one.
And if you like, we can go with purely the digital side of it... we'll ignore the analogue side, including the converters, and just stick with PCM and processing... come on, digital processing is deterministic, so any problems can be repeated, calculated and measured.
Post by: Johnny B on October 26, 2005, 08:56:49 PM
Oh, no you don't. The problems have been posted on the net ad nauseum, I'm not gonna give you another "fix-it" list..Just do a web search, you'll find tons of specifics if that's what you are looking for...I got better things to do...but you go right ahead with your research on specific AD/DA products and YOU can come back with YOUR findings and any proposed fixes and workarounds...
Oh, and if you *do* find the perfect digital solution, I'm sure some of the posters here will be quite interested in having YOU explain it in fulsome detail...and even provide them with a complete shopping list...
If you *do* make out such a shopping list for them, I think YOU should also be kind enough to volunteer your home phone number, that way, if any problems arise, YOU can get the late night phone calls for end-user support.
Look, if you don't want more advanced research using teams of experts utilising a multidisciplinary approach and applying more complete and sound science to the problem, fine with me.
.
Post by: Jon Hodgson on October 26, 2005, 09:50:14 PM
| Johnny B wrote on Thu, 27 October 2005 01:56 |
Jon, Oh, no you don't. The problems have been posted on the net ad nauseum, I'm not gonna give you another "fix-it" list..Just do a web search, you'll find tons of specifics if that's what you are looking for...I got better things to do...but you go right ahead with your research on specific AD/DA products and YOU can come back with YOUR findings and any proposed fixes and workarounds... Oh, and if you *do* find the perfect digital solution, I'm sure some of the posters here will be quite interested in having YOU explain it in fulsome detail...and even provide them with a complete shopping list... If you *do* make out such a shopping list for them, I think YOU should also be kind enough to volunteer your home phone number, that way, if any problems arise, YOU can get the late night phone calls for end-user support. Look, if you don't want more advanced research using teams of experts utilising a multidisciplinary approach and applying more complete and sound science to the problem, fine with me. |
As I expected...
You don't actually know anything do you?
You can't actually come up with anything concrete, or are you just afraid I'll tear it apart.
And you think you can lecture others on scientific approaches?
From the rubbish you've repeatedly posted I think I can safely say that I've forgotten more about digital processing, processor architectures and probably also analogue audio circuitry, sound propogation through air, the functioning of the ear, and psychoacoustics in the past week than you'll learn in your lifetime, and there are many who know more than I do.
I used to think you were just a misguided, but well meaning analogue head, but your recurring twisting of what people say, your complete inability to make a concrete claim, let alone back it up, your evasion of any direct questions, and the pathetic childish accusations you hurl at people who don't jump to agree with you as being anti-research, or tied to old thinking... leads me to conclude that you're just a troll, and a coward to boot.
I'm done with you, I've only put up with you this long because
I hoped that others might learn from what I said, well I'm happy to keep talking with them, especially since they probably have things to teach me too, but I can't be bothered to dignify your utterly vacuous claims and accusations with a response any more, and anyway, having a battle of wits with an unarmed opponent is rather unsporting of me.
So if anyone else wants to discuss things, whether it's science or engineering, or both, then let's get to it, but remember, this is a science forum, not a phylosophy one, nor a theology one, and it certainly doesn't involve black magic... so be prepared to back up what you say and talk facts and figures.
Post by: CCC on October 26, 2005, 10:57:05 PM
| jimmyjazz wrote on Wed, 26 October 2005 22:52 | ||
Pardon my French, but what the fuck are you talking about? There is no "anti-further research mob". What a ridiculous red herring you've put forth. Read it slowly: nobody is against "further research". The thing is, you don't seem to know a) what research has already been performed b) what other research might be useful I swear, you're either a troll, 15 years old, or just not quite all there. Whatever. I'm done with you, and I hope everyone else is, too. I'm embarassed to have participated in these threads at all. |
Very nicely put indeed.
Post by: Ronny on October 27, 2005, 12:06:40 AM
| John Sorensen wrote on Wed, 26 October 2005 22:57 | ||||
Very nicely put indeed. |
Doesn't matter, nothing sinks in and you can't argue with someone that thinks they are always right and makes up crap to prove a point that they think is new. If it was me and I had 30 mo'fo's on 3 newsgroup forums all telling me the same thing, I'd have to start doubting my own information and I can be very stubborn at times. If you can admit that you "may" be wrong once in awhile, you can learn a whole lot more, but staying locked into a narrow minded thought process and arguing points that were proven long ago is lugubrious. Most of the people you can eventually reach, but one comes along once in awhile that is untouchable. I don't think Johnny B. is a troll though, just very misguided from not knowing the basics of what has been covered before and not having the background to argue his views on the same level as most of the people that grace these forums.
Post by: maxdimario on October 27, 2005, 05:13:45 AM
| Quote: |
eliminated from decent converters for a decade, |
Jon are you saying that the converters made in 1995 sound good?
I worked with prism and apogee in 2003 and although they were good, they exhibited that digital coldness just the same.
this was on a pro tools system.
I believed from the recent positive comments of 'you guys' regarding modern converters that the converters made up to a couple of years ago were noticeably worse than the ones made just recently..
I Am not chiming in anymore until I can get my hands on one of the new and superior converters like the lavry, weiss etc.
...but by making this statement... you are basically telling me that digital hasn't improved considerably in the last 4 years... which makes me doubt the whole issue of 'digital perfection' even more..
say it's not so Jon..say it's not so...
Post by: Jon Hodgson on October 27, 2005, 05:41:38 AM
| maxdimario wrote on Thu, 27 October 2005 10:13 | ||
OOps I made the mistake of reading this thread... BAD Max..BAD Max
Jon are you saying that the converters made in 1995 sound good? |
Actually they do, but admitedly todays sound better, it's a refinement process. Perhaps "eliminated" was a bit strong, though in the case of two factors people go on about the most, quantization noise and aliasing noise, I think it's probably a justified claim, it is perfectly possible to build a system where these are effectively nil (as in below perceptible levels AND way below the analogue noise floor).
| maxdimario wrote on Thu, 27 October 2005 10:13 |
I worked with prism and apogee in 2003 and although they were good, they exhibited that digital coldness just the same. |
What the hell is "digital coldness"??
I see people repeat it ad infinitum, but what is it supposed to mean? You mean it doesn't have the distortion, phase and frequency anomolies of analogue tape?
I also read people saying digital is "clinical"... of course it's bloody clinical, it's intended to replicate a given signal as perfectly as possible, it's not supposed to have character, it's not supposed to sound warm or soft or cuddly or hard or spiky or anything, it's supposed to sound like the signal that went in!
| maxdimario wrote on Thu, 27 October 2005 10:13 |
this was on a pro tools system. I believed from the recent positive comments of 'you guys' regarding modern converters that the converters made up to a couple of years ago were noticeably worse than the ones made just recently.. |
Yes, and I think Rupert Neve will probably tell you his latest work is his best ever, but nobody keeps claiming that analogue electronics is "broken" in the way that they do digital. Clocks get better, amplifier stages get better, noise shaping gets better, things keep improving.
| maxdimario wrote on Thu, 27 October 2005 10:13 |
I Am not chiming in anymore until I can get my hands on one of the new and superior converters like the lavry, weiss etc. ...but by making this statement... you are basically telling me that digital hasn't improved considerably in the last 4 years... which makes me doubt the whole issue of 'digital perfection' even more.. say it's not so Jon..say it's not so... |
They have improved, I shouldn't have made quite such a bold statement, but Johnny repeatedly talks about asking questions we already know the answer to, and have done for a long time.
However I still have some tests to devise for you which will either open your eyes or mine, because I've seen some of the things you claim and one of us has to be very very wrong.
Post by: maxdimario on October 27, 2005, 06:11:14 AM
| Quote: |
What the hell is "digital coldness"??...I also read people saying digital is "clinical".. |
the cleanest signal path is a straight wire...it is not 'cold'.
digital sounds almost as clean as a straight wire but it is 'colder' or less lively, intimate etc..
forget the artifacts of tape.
digital takes out.. tape adds.. so to speak..
...this is just to clarify what some people mean by 'cold' not to 'fan the fire' further...
Subjective opinion is difficult to discuss, we should be listening to examples all together and be talking about them in real-time ideally..
Post by: Jon Hodgson on October 27, 2005, 06:21:25 AM
| maxdimario wrote on Thu, 27 October 2005 11:11 |
digital takes out.. tape adds.. so to speak.. |
Thanks for the clarification of what you mean...
but what does it take out?
It can recreate the signal with phase discrepancies, frequency response fluctuations and noise levels which are below what anyone can hear, so what is it taking out?
Post by: maxdimario on October 27, 2005, 07:11:14 AM
| Quote: |
so what is it taking out? |
I have my own ideas, which have been repeated ad-nauseam, as many of you will know.
To hear the difference get some good headphones such as hd580 hd600 or better..get a live analog source with a very good neumann or ribbon and a discrete pre and listen through headphones through a decent headphone amp, then insert the converter in the signal path doing ad-da with internal routing if you wish to minimize routing effects.
I play acoustic guitar as it sits on my lap, or get someone else to play an instrument in the other room. Don't know if you play or not.
as you compare direct with digitized listen carefully to the high-end freq. range: pick attack, or the attack of the consonants, the sss sounds etc.
listen in terms of up-front immediacy and the sense of physical proximity to the sound as you switch the converter in and out of the signal path.
this is the only quick way to understand what people mean by 'cold'I can think of.
Post by: Jon Hodgson on October 27, 2005, 07:35:09 AM
| maxdimario wrote on Thu, 27 October 2005 12:11 | ||
I have my own ideas, which have been repeated ad-nauseam, as many of you will know. To hear the difference get some good headphones such as hd580 hd600 or better..get a live analog source with a very good neumann or ribbon and a discrete pre and listen through headphones through a decent headphone amp, then insert the converter in the signal path doing ad-da with internal routing if you wish to minimize routing effects. I play acoustic guitar as it sits on my lap, or get someone else to play an instrument in the other room. Don't know if you play or not. as you compare direct with digitized listen carefully to the high-end freq. range: pick attack, or the attack of the consonants, the sss sounds etc. listen in terms of up-front immediacy and the sense of physical proximity to the sound as you switch the converter in and out of the signal path. this is the only quick way to understand what people mean by 'cold'I can think of. |
I do play acoustic, actually I'm thinking of building a new one, but that's another story,
however this test appears to have a fatal flaw. it is not blind.
Post by: maxdimario on October 27, 2005, 08:25:39 AM
You just compare the differences using your ears, and make observations.
if you have problems hearing differences, then you can just forget about the whole issue, since the difference in such a test (great mic, pre, headphones) is about as obvious as it gets.
blind tests are not necessary all of the time.
are you using hide glue BTW?
Post by: Jon Hodgson on October 27, 2005, 08:50:02 AM
| maxdimario wrote on Thu, 27 October 2005 13:25 |
it's not fatal. You just compare the differences using your ears, and make observations. if you have problems hearing differences, then you can just forget about the whole issue, since the difference in such a test (great mic, pre, headphones) is about as obvious as it gets. blind tests are not necessary all of the time. are you using hide glue BTW? |
Considering that the placebo effect can make people with chronic and painful diseases feel better, convincing yourself that you can hear a difference is very easy.
If the difference were truly as obvious as you imply, then that would mean that either
1) There is something wrong in your test setup
2) You can hear some difference as "obvious" which is below any measurements that the past century of research into the auditory system and audio electronics has deemed detectable. The signal coming out of a good DAC is so close to that going into a good ADC that differences should not be obvious by any means - not ruling out their existance, just saying that if they were that obvious they'd be a commonly understood part of sampling theory.
So yes, not having the test be blind is a fatal flaw in my opinion. But we'll see further on down the line who is right.
I've no idea what glue I would use, never built an acoustic before, but I know a guy who does courses on it, and when he showed me how to build an electric I found it very theraputic... actually I need to finish that guitar, my life went kindof crazy after that and it's been in an almost finished state for a couple of years now. Everybody else who's done one has found they play and sound great though... at least the electrics, the acoustics are an unknown to me as yet, but I have reason to trust the guy.
Post by: maxdimario on October 27, 2005, 09:08:59 AM
3) you can hear things that people have written off as being impossible (quite a common when habitually testing equipment)...perhaps because it was convenient for marketing, political etc. to make such statements.
4) the test is valid depending on the individual's ability to hear the difference and analyze it. This is due to experience in critical listening and 'ears' basically.
The thing that worries me is: your reply above denotes that you are more apt to justify or give reason to arguments based on established knowledge, but without implementing your personal experience, than to try and hear for yourself.
You will be making a minimal effort, and either way you might be able to tune-in to what other people are talking about even if it doesn't interest you personally. At least you can try.
use an acoustic guitar BTW, not electric.
Post by: Jon Hodgson on October 27, 2005, 09:27:50 AM
But just to check something with your ears and what is obvious that I can do a test for when I have a little time...
Do you still maintain that you can tell the difference between a first generation digital recording, and one that has been simply bounced 10 times, and that such a difference would be OBVIOUS?
I'm talking about starting with a 16 bit recording, interim processing at 32 bit float (what you have in Logic) and finally rendering at 16 bits (44.1 kHz, since that's what I have available).
Post by: StudioRhythm on October 27, 2005, 11:41:46 AM
| maxdimario wrote on Thu, 27 October 2005 04:11 |
To hear the difference get some good headphones such as hd580 hd600 or better..get a live analog source with a very good neumann or ribbon and a discrete pre and listen through headphones through a decent headphone amp, then insert the converter in the signal path doing ad-da with internal routing if you wish to minimize routing effects. I play acoustic guitar as it sits on my lap, or get someone else to play an instrument in the other room. Don't know if you play or not. |
Hi Max, thought I'd point out a clear problem with at least part one of this test -- latency. Every converter has some latency built-in, and if you're listening through headphones to a source in the same room, (especially if you're playing it,) you'll get a different pattern of phase cancellation with the monitored vs. acoustic source. This will subjectively favor the analog path every time, since you'd hear your headphone signal more clearly (more "predelay" if you will from the acoustic source) than you would with the extra delay of the A/D conversion.
A good experiment to see the effects of this is to play an instrument through an A/D or DAW, and slowly adjust the amount of delay you're monitoring with from a range of as close to none as possible up to maybe 10ms, which will make it pretty obvious and uncomfortable to a sensitive performer. The lowest latency will always sound better to the performer, especially with instruments like guitar and bass where the performer is receiving an amount of direct vibrations through their body.
The important thing to realize about this is that while it can be distracting to the performance in some cases, it does not influence the actual sound being recorded at all.
| Quote: |
listen in terms of up-front immediacy and the sense of physical proximity to the sound as you switch the converter in and out of the signal path. |
You have just perfectly described what I'm talking about!
Post by: maxdimario on October 28, 2005, 10:28:51 AM
| Quote: |
Do you still maintain that you can tell the difference between a first generation digital recording, and one that has been simply bounced 10 times, and that such a difference would be OBVIOUS? |
No.
It used to be obvious 3 or 4 years ago..
As I understand the system that was used was the new pro tools hd, and I didn't get if he used bounce to disk or not.
so pro tools is working pretty good.
but yeah.. that was a pleasant surprise.
it was hardly noticeable at all.
strange thing is that the MP3 encoded versions had more of a difference, so why is that..?
Post by: maxdimario on October 28, 2005, 10:36:50 AM
| Quote: |
You have just perfectly described what I'm talking about! |
Yeah I'm aware of that, which is why I stressed that maybe he'd want to rout the thing internally.
but the difference is not a phasing thing since I've tried with remote sources as well and I do notice a particular quality in the top end which is not due to phasing.
It's hard to describe this kind of thing..
it took me about 10 years of using digital and accepting it as the standard and then one day I recorded to tape for a laugh on an ampex..and I began to question the whole thing once again.
so it took ME a long time to even notice it.
just think 'lively' as you listen.
sometimes people can kill themselves over the meaning of a single word such as 'cold' or 'warm' etc..
Post by: Jon Hodgson on October 28, 2005, 10:45:10 AM
| maxdimario wrote on Fri, 28 October 2005 15:28 |
it was hardly noticeable at all. |
So you say it's improved, but still maintain it is noticable?
Post by: StudioRhythm on October 28, 2005, 05:22:17 PM
| maxdimario wrote on Fri, 28 October 2005 07:36 |
Yeah I'm aware of that, which is why I stressed that maybe he'd want to rout the thing internally. |
But even with internal routing, you're still getting a delay, probably somewhere around 1ms with a round of A/D > D/A, which will still make a difference in how you're perceiving the combination of headphones and live source. The less-delayed signal (the straight wire) will sound better just because of the delay.
| Quote: |
but the difference is not a phasing thing since I've tried with remote sources as well and I do notice a particular quality in the top end which is not due to phasing. It's hard to describe this kind of thing.. |
I just wanted to point out that any test as described where the acoustic source is in the same room with the subject doing the test is not a valid test because of the latency, for which there is no solution. In addition, if you were to have done the test where you're playing before you had someone else play from an isolated room, that would tend to set a certain expectation in your mind that might well carry over to a test with the source in another room, since the tests were not done blind.
That said, I haven't done a direct A/B test between live source and a high-end A/D > D/A, so I'll admit that there could be a perceivable difference. Next chance I get I'll check it out and see what I can or can't hear. I have to be somewhat skeptical, though.
| Quote: |
it took me about 10 years of using digital and accepting it as the standard and then one day I recorded to tape for a laugh on an ampex..and I began to question the whole thing once again. |
I just want to state again for the record that there are so many measurable and documented things going on with analog tape that alter the sound of the input (many in a pleasing way,) that it's not really a valid argument to say that "analog sounds better, therefore digital doesn't work right." Not sure if that's what you meant, but if it was, I don't agree at all.
Post by: minister on October 28, 2005, 09:07:14 PM
| maxdimario wrote on Fri, 28 October 2005 09:28 | ||
No. It used to be obvious 3 or 4 years ago.. As I understand the system that was used was the new pro tools hd, and I didn't get if he used bounce to disk or not. so pro tools is working pretty good. but yeah.. that was a pleasant surprise. it was hardly noticeable at all. strange thing is that the MP3 encoded versions had more of a difference, so why is that..? |
...as for the mp3, i can't say....
Post by: maxdimario on October 29, 2005, 05:13:02 AM
| Quote: |
so many measurable and documented things |
Well you don't need measurements to hear that tape distorts the signal more noticeably than digital.
the difference is what and how it distorts.. and it's effect on the way the ear interprets the sound.
If you are reproducing music, you need to focus on eliminating the kinds of distortion that may affect how it comes across to the listener.
sometimes I read people saying that distortion is distortion, but although that is true from a language-logic point of view, it does not apply in music reproduction.
there are many kinds of distortion: frequency, low-order harmonic, high-order, inter-modulation, phase etc. and there are not enough terms to describe it all.
some hinder the music, some don't alter it and some actually help it.
There's nothing drastically evident about digital distortion, but what is evident to many, is the overall phsycological effect of the music on the listener.
..perhaps the latest generation ADC's have got this sorted out.
Post by: StudioRhythm on October 29, 2005, 01:19:12 PM
| maxdimario wrote on Sat, 29 October 2005 02:13 |
There's nothing drastically evident about digital distortion, but what is evident to many, is the overall phsycological effect of the music on the listener. |
I agree with everything you said except this part. I'm not convinced that such digital distortion as you call it exists (unless you're talking about clipping -- we all know that exists, but that's pretty evident when it's there!) I think that when you start talking about psychological effects, you have to be really careful to make some controlled analyses, and I don't recall ever hearing about any of those being done. There are a lot of people talking in vague ways on the internet about things like this, but there are lots of people talking about all kinds of things on the internet, and a good number of them are complete B.S.
Without some kind of real controlled tests, there are way too many variables at work here, and a huge one is that a good percentage of the music out there today is compressed, hyped, and brickwalled to death. Add to that the fact that there's lots of mediocre music out there right now and the current monopolized radio system only makes it harder for good stuff to hit the airwaves, and right there you have several reasons for a negative psychological impact on the listener, but none of them need invoke some magical unexplained flaw in digital recording.
Post by: Gunnar Hellquist on October 30, 2005, 09:33:28 AM
I have done a few tests lately on mixing digital. In Samplitude there is an "analogue tape simulation" effect available. I tried putting that at a very low effect level on the tracks before mixing. It does change the sound in a somewhat positive way, making it less "cold".
So just maybe the cold sound is simply that there is very little distortion. Just maybe a bit more distortion (the right kind) can make things sound more pleasing to the ear. It feels sort of similar to dithering, where we add noise to get a better sound. My guess is that we could add this distortion in a similar way.
In about the same way mix bus compression adds something positive to the sound. Generally I put a low ratio compressor on the main mix bus when mixing digital just to get something similar.
Gunnar
Post by: minister on October 30, 2005, 11:39:48 AM
i think this is why products like the HEDD, PHOENIX plugs, NEVE TAPE emulation box, FATSO, McDSP ANALOG CHANNEL, even the UA plugs, and summing BOXES are made and sold. to SUM analog is to get a bit of pleasing harmonic distortion. many people (including myself at times) insert analog boxes (both solid state & tube) into the mix path or on the mix bus. it's also why a lot of engineers track things like drums on 2" and then dump into a DAW for editing.
time will tell how much of this "pleasing" sound was a paradigm sonic aesthetics of our ethos, and how much of it is related to the way we, as organisms interacting with our environment, hear. maybe, there will be nostalgia for the sound of 80's reverbs, or, iTunes, or MPC 2000's in 25 years... and the old guys on forums like this will HATE the new ____ technology, talk about how pleasing 2005 era digital sounded.