Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Johnny B]

Thanks for the responses....

Nyquist is often used to justify using slower rates....the question is not simply a matter if the theory can be proven with math...that's almost a big "so-what?"

Why? Because it's one thing to get something to work on paper and quite another to get it to work in the real world. As we've seen, there is no such thing as a perfect filter, and since you have no perfect filters then this explains in large measure the push toward higher sample rates...that, and the debate over what the truly "important" part of the frequency spectrum is vital for the human being's system to experience.

Given the real world barriers to fully implementing Nyquist's theory...then that leaves plenty of room for improvement. Moreover, let's also not forget that there may be other possible methods to do the sampling process that may, in fact, end up working better in the real world.

To keep a closed mind, to think that *everything* is now solved, is to deny the very real problems which now exist in digital sound.

To me, the right approach is to seek out constant advancement...constant improvement...

If your don't look, if you do not challenge, then no technology will ever move forward...

It'll just be stuck in the past...

Believe me, I'm not saying the past has no value whatsoever, I'm just saying we should not stay stuck there...

Hell, we would have never flown or gone to the Moon if we simply stayed stuck in the past. And BTW, to make that happen, some old belief systems had to be challenged and had to be set aside as "just not working" under the circumstances presented.

"Constant Innovation" means a willingness to constantly challenge the past.

[Jon Hodgson]

Johnny B wrote on Mon, 03 October 2005 17:34

To keep a closed mind, to think that *everything* is now solved, is to deny the very real problems which now exist in digital sound.

You keep repeating this, as if the people who challenge you when you repeat 'higher sample rates, more  bits' ad nauseum have ever said that present digital audio SYSTEMS are perfect.

Want to point out where anyone has said that?

What we're telling you is not that present systems (and by systems I mean the analogue stages, the quantization, the clocking, the digital processing etc) are perfect and cannot be improved, but rather that you need to know what you're talking about before you start preaching to others about how to improve it, and posting bullshit thread titles like "a challenge to Nyquist' and 'A Challenge to Fourier'... to those of us who understand the subject you just look like a fool, and to those who don't you just muddy the waters and spread disinformation which makes advancement of the subject, and improvements in sound quality LESS likely, not more.

Right now you're like someone who believes the earth is flat ranting at everyone else that they're luddites for not devising better navigation methods.

[Ronny]

Jon Hodgson wrote on Mon, 03 October 2005 13:03

Johnny B wrote on Mon, 03 October 2005 17:34 To keep a closed mind, to think that *everything* is now solved, is to deny the very real problems which now exist in digital sound. You keep repeating this, as if the people who challenge you when you repeat 'higher sample rates, more  bits' ad nauseum have ever said that present digital audio SYSTEMS are perfect. Want to point out where anyone has said that? What we're telling you is not that present systems (and by systems I mean the analogue stages, the quantization, the clocking, the digital processing etc) are perfect and cannot be improved, but rather that you need to know what you're talking about before you start preaching to others about how to improve it, and posting bullshit thread titles like "a challenge to Nyquist' and 'A Challenge to Fourier'... to those of us who understand the subject you just look like a fool, and to those who don't you just muddy the waters and spread disinformation which makes advancement of the subject, and improvements in sound quality LESS likely, not more. Right now you're like someone who believes the earth is flat ranting at everyone else that they're luddites for not devising better navigation methods.

It'a all Nyquist's fault. He never got digital to sound like analog.  


Johnny B. rather than rewrite the book on A/D conversion, I suggest that you start with tracking mic manufacturers. Get them to standardize a microphone that picks up above the human hearing range. To take advantage of 384k, you'll have to get them to stop making mics that only go to 20k and get them to design mics that capture 192k. Than get all of the speaker manufacturers to create a speaker that will playback content sampled above 48k. Beware that the cost of tracking mics will go up higher than most ref mics and the playback systems can theoretically playback frequencies that you can't hear, but will cost 10 times the price that a typical set of monitors cost. Until you change the devices that capture and playback, 384k isn't going to mean diddly as 192k is never captured nor ever played back.  

Or, simply start performing blind tests on your friends, between sample rates. 44.1, 48, 88.2 and 96k. You'll find that what all of the engineers on Dan's, Brad's and George's forums have been unanimously telling you is correct in the real world and not just on paper. You'll open your mind to what is really happening when you start testing in the blind and you'll have a better understanding of where to improve the audio signal chain, than trying to do it on speculation based on misinformation that you read. If it's inaudible than it is certainly inconsequential and not worth your waste of time, or anyone elses for that matter.    

[Johnny B]

Ronny wrote on Mon, 03 October 2005 18:41

more. It'a all Nyquist's fault. He never got digital to sound like analog.

I may not go that far, after all, Nyquist's ice cold, dead body was buried long ago. Well before the current digital "implementations" by others.

As for the reasons for digital not sounding as good as analogue, there are many heated debates about that. And heated debates on how best to solve all the problems in digital.

Are there better alternatives to doing it the old way? I dunno, maybe.

[bobkatz]

Johnny B wrote on Mon, 03 October 2005 12:34

Thanks for the responses.... Nyquist is often used to justify using slower rates....

Where do you get that?  Give me one citation where someone correctly and in context tries to use Nyquist as such a "justification".

Nyquist is simply the reason for why a particular sample rate is all that is necessary assuming that you have defined a certain bandwidth. That is all. It's not a justification for the choice of the bandwidth in the first place.

BK

[Johnny B]

C'mon Bob,

You know that everybody and their brother bought in to the idea that 16/44 bit CDs fully complied with Nyquist...and then people started to use 24/96 and then 24/192 and then who knows what the next thing will be....

On every single occasion, on every single format change...they all say they fully comply with Nyquist....

Trouble is, none of them comply with Nyquist because they can't fully comply with Nyquist...

The filters are not perfect...so no one fully complies with Nyqust because they can't...

[crm0922]

Johnny B wrote on Tue, 04 October 2005 00:03

C'mon On every single occasion, on every single format change...they all say they fully comply with Nyquist.... Trouble is, none of them comply with Nyquist because they can't fully comply with Nyquyist... The filters are not perfect...so no one fully complies with Nyqust because they can't...

What is your point, Mr B?  The analog tape machine does not function ideally either.  There are plenty of errors being added to a signal as it goes down to tape.  And not all of it is "wonderfully musical" sounding to every ear.  

Many engineers prefer the consistency and accuracy of a good digital system.  You do not, and neither do I, so why keep arguing that the technology is flawed?

Chris

[Johnny B]

Emmm, cuz it is flawed...but there may be the potential to make it better...or at least sound better...Just think, if digital sounded exactly like analogue...you probably would not hear a whimper out of people...but as it is...you hear some people roaring on about digital sound quality...

Here, I'll *try* to change the discussion somewhat and *try* to place it in a more positive direction.

Ok, we know that certain trends are expected to continue in the general digital technology, right?  64-bit CPUs and 64-bit OS etc...OK?

I've already seen some of Dan's thinking on what it would take to improve digital sound, so we need not repeat that here.

What I'd like to see posted here are other people's ideas for improving digital sound quality...

And maybe at this point we could all agree to temporarily suspend judgement about whether or not a particular idea is foolish or outlandish. Sometimes out of such a process new useful thinking can emerge...

So I'd like to see people just start throwing out ideas of how they think, suspect, or simply feel digital sound quality might be be improved...

We can always come back later to sort out the really good ideas from those which don't make the cut

How's that sound, if nothing else...it could be fun.











 

[Jon Hodgson]

Johnny B wrote on Tue, 04 October 2005 09:37

Emmm, cuz it is flawed...but there may be the potential to make it better...or at least sound better...Just think, if digital sounded exactly like analogue...you probably would not hear a whimper out of people...but as it is...you hear some people roaring on about digital sound quality...

Haven't you been paying attention?

Analogue is a really poor recording system, full of distortions and pitch inconsistencies and bandwidth limitations and high noise floors and phase shifts and general screwing up of the signal.

The fact that those errors are mostly not too unpleasant, and in many cases are pleasant, do not make emulating this in its entirity an admirable goal.

Sure digital systems can improve, but the benchmark should be to record the sound of the instrument in the room as if you were there, not to emulate an inherently flawed system such as your beloved analogue tape.

Analogue recording systems are like two units together, one is a  recording system which records and plays back the actual signal, the other is a processing unit which gives saturation effects, wow, flutter, filtering etc. The problem is that you can't turn off the processing unit, it's always there and always turned on, the only thing you can do is vary your signal to get more or less processing.

Digital offers us the opportunity to seperate the two, so that you can have a recording which DOES NOT HAVE A SOUND, and then process it as much or as little as you wish.

[Johnny B]

I dunno Jon.

Sure there are many instances when you want what is in the room to be exactly what comes out of the recording. That's often the desire.

OTOH, there are many instances where you want what comes out to sound better than what what in.

For example, I'm not one to use Autotune, but I've been told by others who do use it that it works better at the high sample rates. What exactly is going on with the FFT and the algos to make Autotune work better at high sample rates may be open to some debate.

I don't know how many people go to tape just because they like what it does to the sound, but it's enough for Rupert Neve to bring out a box just for that purpose, so there must be a sigificant number of people demanding that application.

And of course it would be nice if someday digital allowed both kinds of approaches, such as a transparent signal and an analogue colored signal to come out on the record. It'd best if you could mix and match to your heart's desire, but unfortunately we do not currently have that with digital.

The question I'm posing is a rather broad one, "Given that most computer systems are going in the 64-bit direction, what will be the best ways to improve digital sound quality?"

Sometimes technology gets improved "incrementally" and sometimes something "revolutionary" comes along. We are likely to see a bit of both.

Because we are likely to see such vast changes in the future, it opens up all sorts of interesting possibilities.  That's what I'd like to explore here.  

[C-J]

Johnny B wrote on Tue, 04 October 2005 21:45

And of course it would be nice if someday digital allowed both kinds of approaches, such as a transparent signal and an analogue colored signal to come out on the record. It'd best if you could mix and match to your heart's desire, but unfortunately we do not currently have that with digital.

Hmm? So you're looking for a transparent, colored signal?
Reminds me of Bob's "Oxymoron of the month" thread.

C.J.

[blueboy]

After reading through Dan's arguments regarding the benefit (or lack thereof) of higher sampling rates I'm starting to think it is better to look at other areas for improving the "sound" of digital. Although I can't claim to understand all of the aspects involved, it makes sense that anything over the optimal rate of 60khz is sufficient, and that there really aren't any issues with micro-timing.

24/96 makes sense then as it should be more than adequate, but 24/192 is just a waste of processing power and storage space.

Most people want to hear the the distortions that are inherent in analog technology, but current digital "imitations" just don't seem to work very well.

If we don't fully understand what it is that our ears are reacting to to give us that analog sound, we can't create algorithms to emulate it.

On the other hand, something like convolution processing captures exactly what is going to a particular frequency at a particular amplitude (whether we understand what is happening or not). Just like most digital reverb algorithms don't sound that "real", but their convolution equivalents sound much more realitic. Not perfect but much closer.

Is it not possible then to take multiple convolution impulse responses from an analog source (covering and extremely wide dynamic range to cover ever possible variation), and then create a dynamic convolution processor that can morph between these impulses (depending on the input level) to create a much more natural sounding analog emulation?

Is anybody doing "dynamic convolution" yet? If so, how does it sound?

JL

[Jon Hodgson]

blueboy wrote on Tue, 04 October 2005 20:17

Is it not possible then to take multiple convolution impulse responses from an analog source (covering and extremely wide dynamic range to cover ever possible variation), and then create a dynamic convolution processor that can morph between these impulses (depending on the input level) to create a much more natural sounding analog emulation? Is anybody doing "dynamic convolution" yet? If so, how does it sound? JL

Focusrite have combined a dynamic convolution processor with a variable input impedance mic preamp to create a unit that is supposed to emulate a whole range of mic preamps and compressors.

I can't comment personally on how well it works, but I can say it has had positive reviews in the press for what that's worth.

[blueboy]

Jon Hodgson wrote on Tue, 04 October 2005 14:07

blueboy wrote on Tue, 04 October 2005 20:17 Is it not possible then to take multiple convolution impulse responses from an analog source (covering and extremely wide dynamic range to cover ever possible variation), and then create a dynamic convolution processor that can morph between these impulses (depending on the input level) to create a much more natural sounding analog emulation? Is anybody doing "dynamic convolution" yet? If so, how does it sound? JL Focusrite have combined a dynamic convolution processor with a variable input impedance mic preamp to create a unit that is supposed to emulate a whole range of mic preamps and compressors. I can't comment personally on how well it works, but I can say it has had positive reviews in the press for what that's worth.

Looks very cool...

  http://www.focusrite.com/productdetails.asp?id=37&iRange =1

Has anybody heard this yet?

And Sintefex has already trademarked the term "Dynamic Convolution" so it's obviously not a new concept...

Damn I'm always too late with these good ideas.

Here's some technical documents for anyone interested:

http://www.sintefex.com/docs/appnotes/dynaconv.PDF
http://www.sintefex.com/docs/appnotes/compsim.pdf

JL

[PookyNMR]

Johnny B wrote on Tue, 04 October 2005 02:37

Emmm, cuz it is flawed...but there may be the potential to make it better...or at least sound better...

So, what are you doing to make it better?  Seriously!

Do you honestly think that all those competing companies out there are not trying to make it better (marketing BS and sales gimicks aside)?  Do you think they don't care about sound quality?  Do you not think that the brilliant engineers at Lavry, Mytek, EMM labs, etc, are trying their hardest to make improvements and come out with the best possble products?

Johnny B wrote on Tue, 04 October 2005 02:37

.Just think, if digital sounded exactly like analogue...you probably would not hear a whimper out of people...

Here I think you're delusional.  If digital had a noise floor, phase shift, distortions, compressions, limited bandwidth, wow and flutter, etc, etc like analog you think people would be happier?  

2 kinds of people complain about digital.  #1 are highly skilled people who have actually logged considerable hours with state of the art analog and digital gear and can hear specific differences.  #2 are those who still lack some basic engineering skill and blame their bad sound on their gear and their misuse of it.

In my occasionaly humble opinion learning and refining your engineering skills will make 10x10^99 times more of a difference to the quality of your sound than whether or not your record at 44.1 or 192 or 2" tape.

For some like the heavy hitters on this board, I can see why they could want some progress (though I gaurantee you they all work with a digital rig (or 10) every single day).  But do you ever notice how it's the low end pee ons like you and me that are the ones that do most of the complaining?  I think we'd get much farther if we actually learn what the heck we're doing before we start to argue about 2" vs 192k.

Peace.