Print

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

[blairl]

bobkatz wrote on Thu, 15 September 2005 14:25

just because you can hear a difference between 192K and 96K doesn't mean that 192 is automatically better. You have to find out why there is a difference.

What if you substitute 96 for 192 and 48 for 96 in that quote?

[$a1Ty]

i was thinking the other day about this

with converters if you record at 44.1 are they flat 20hz-20khz or do they roll of a bit, so if you record at 88.2(or whatever) it will be flat over 20hz-20khz

[maxdimario]

Quote:

they can't hear a difference between SACD and cd, not a profound difference and it takes a profound difference to revolutionize a media shift

maybe that's why, but I'm not buying it because I know as soon as *I* buy it there will be another format change and SACD will be obsolete.

why waste my money on an obsolete format?

it's like buying computers... unless you use them for work, it makes no sense to buy the latest model every time one comes out.

[Ronny]

jwhynot wrote on Thu, 15 September 2005 17:26

Sure microphones capture above 20k.  You can roll off 12 db/octave and still have meaningful content at 40k.

It's insignificant in the grand scheme. We know that instruments have harmonics that can be measured at 80k with ref mics, however these harmonics typically make up less than 1% of the energy. We can not dismiss the element of masking by the frequenies, that are most perceptive to the human ear and that's why air freq's or freq's above 20k, even if they can be captured and reproduced on a special system are for the most part  insignificant. Much of the claims made by people that say they hear frequency interaction above 20k affecting freq's below 20k aren't confirmed by blind testing.  

Quote:

I agree with BK's post as well. My experience with higher sampling rates is that it makes more difference to me at the D/A stage. I like the results when I up-sample from 48 to 96 on playback, when I'm mixing, about 90% as much as the difference if you start from the A/D stage.

That's why we have 128x delta-sigma oversampling. The capture at 44.1k is upsampled to something like 5600k, from my understanding, so that the intitial input is high sampled, than of course the sample rate is lowered at the decimation stage. I agree that the initial capture at the ADC is important, but once the signal is digitized, no real need to process the oversampled rate.

Quote:

But this virtually infinite debate all comes down to the merest tiniest detail of difference in the sound, all of which is absolutely insignificant relative to just about every other aspect of recording.

No doubt about it. Moving a microphone 2 inches can change the sound more than the audible differences between 44.1 and 96k.

Quote:

You want that extra bit of "quality", I suggest high SRs are the last resort - after the qualities inherent in the instruments/voices themselves, the arrangement, the honesty of the music, the magic of the moment, mike placement, signal path, processing, take selection, editing, and reason for the music existing in the first place. Get all that right (and the 100 or so other factors I haven't listed) and then start worrying about conversion and sample rates.

I couldn't agree more.

Quote:

JW PS the move to 24 bit is a much bigger enhancement to my ears.

Without a doubt with some converters, but factors include the quality of the analog side of the ADC, a high end 16 bit converter will sound better than a low quality 24 bit converter. Bottom line is that not all AD converters are created equal. Just because 24 bit has more resolution, does not mean that all 24 bit converters are going to sound better than all 16 bit converters. There is more to the picture than bit depth, when it comes to the quality of the capture and reproduction. The dynamic range of 16 bit is -15dB lower than the self noise of the best Neumann mics and the noise floor of a 24 bit converter is theoretical below -127dB (Lavry converter for example).  

[maxdimario]

Quote:

The late conductor of the Philadelphia orchestra used to evaluate test LP pressings on a $39.95 Victrola phonograph, yet he could tell if the timbres were wrong. Pay attention to the critical listeners.

I can hear the effect of different gear and timbre on the phone, much to the surprise of my friends, but I don't listen to music on the phone, and I believe that the ear will pick up, or learn to pick up as much information as it is used to process.

I like my listening system to bring out the reality and performance element in music. I don't really care about distortion and freq. response, as I find that most records are over-eq'ed anyway, and speakers have ridiculous THD and frequency specs.

given that, an amplifier that is not flat from 20 to 50 Khz has got some serious design flaws and probably won't be musical.
While I don't mind having some kinds of analog distortion mixed in with the sound, digital kinds of distortion are annoying and cd's are lacking the here and now element which makes music that much more captivating, hence my cd player only gets 15-45 min. of playing time every once in a while, and it does not leave a lasting impression of the music.

Vinyl on the other hand is noisy and imperfect, but on *MY* system it does have more of the performance element than cd, and every record is different.

there are some cd's that sound better than others, because they were taken from a live to two-track which is the best for picking up performance, but the vinyl wins for human factor.


In my experience the  sound of an audio system is either likeable, or not. You can twist it around, eq it, play with it and improve it, but the sound is still there.

the only true test for these kinds of issues would be to have regular meetings of audio experts from all fields and do blind tests on every kind of format.

after a repeated series of these group tests, maybe we could get somewhere with what *is* audible and what is not, and perhaps why.

[Ronny]

$a1Ty wrote on Fri, 16 September 2005 00:10

i was thinking the other day about this with converters if you record at 44.1 are they flat 20hz-20khz or do they roll of a bit, so if you record at 88.2(or whatever) it will be flat over 20hz-20khz

You are still limited to the capability of the mic, besides you can't hear 20k unless you have a system that can play it back and even so, to hear 20k freq's and above in real audio, you'd have to boost the gain by a tremendous amount to keep the lower and more perceptive to human frequencies from masking the content, and without destroying the tweeters. The highest fundamental note in a 100 piece orchestra is typically the highest note on a piano or a piccolo, that fundamental is only 4.2k. The reason why FM radio doesn't sound absolutely terrible is because most of the energy is contained below the max limit of FM which is around 15k. BTW, an SM 57 is only near flat to 15.5k, but that's all you need and that's why the 57 is still being used by folks that have a very high end mic cabinet. There are much more important issues than frequencies that are above Nyquist.

To get a better understanding of what I'm trying to relate. Play a 2k sine tone at -10dB, than without lowering gain, switch it to 15k. Although the gain is really the same, your perception will just barely allow you to hear the 15k tone while the 2k tone can be piercingly loud. Switch it to 18k and tell me if you can hear anything at all. Listen at 8k, 10k, 11k, 12k, 13k, 14k, 15k, 18k and 20k all at -10dB, notice how the perceptive volume goes down exponentially when you get above 8k. Now relate that rate of drop off of perception to frequencies above 20k, you'll find that when it comes to the human ear that these frequenies are insignificant with normal audio that has most of the perceptive energy between 1 and 4k, because our most perceptive frequencies mask the air frequencies and the higher the air freq, the more they are masked by the most perceptive frequencies.  

[Shane Ervin]

Everything on earth is flawed if you want to get right down to it.

On the other hand, lots of folks will speak the words: "Close enough for Rock'n'Roll".

Choose your own ground and be happy.

Where paying clients of a recording studio are concerned, the landscape changes a bit.  Make a list of all the things that you think are important to a paying client.  Next, consider those things that reward your sense of professional gratification (e.g., 24/96).  Then hang out the shingle, hoping that at least some clients will recognize your value proposition - and the fact that you're fun to be around because (amongst other reasons) you've got this infectious spark of interest in something both parties feel is central to making a memorable recording.

Assuming the remaining aspects of a recording project are done well, an MDM with 16 bit words at a 44,1 kSa/s rate will satisfy many folks' needs, including your artist clients and those listeners who later buy their album.

OTOH, it's a wonderful thing that modern digital elex are priced affordably enough to put higher resolution in our hands.

Life is good, no?

[Ethan Winer]

Bob,

> What is wrong with giving the judgment of timbre to musicians? <

Nothing at all! But this is not the point I was addressing. People who lack scientific training often do not understand the limitations of non-blind testing, or even understand that recording something one day at 44.1 and something else another day at 96 cannot be compared. I've seen many comparisons of gear and sample rates etc where different performances were captured and then compared. This is the kind of stuff I meant with my comment.

--Ethan

[bobkatz]

blairl wrote on Thu, 15 September 2005 23:50

bobkatz wrote on Thu, 15 September 2005 14:25 just because you can hear a difference between 192K and 96K doesn't mean that 192 is automatically better. You have to find out why there is a difference. What if you substitute 96 for 192 and 48 for 96 in that quote?

ALMOST...

Good 48 sounds extremely good. 96 is not proportionally that much better than 48 as 48 is bettter than 44. Let me rephrase that, 48 is significantly better sounding than 44. 96 is slightly better-sounding than 48.

However, ALL OF US are making these evaluations on converters. I now have found evidence to believe that many of the converters themselves prejudice 96K over 44 and 48K. A superior DAC design that I now own has made 44 and 48 much closer in sound to 96 than ever before.

Evidence like this makes me think twice, three times about the future and the practical aspects. Let's find out WHY (and IF) before making a headlong rush to a VERY expensive option.

[Ashermusic]

bobkatz wrote on Fri, 16 September 2005 21:09

blairl wrote on Thu, 15 September 2005 23:50 bobkatz wrote on Thu, 15 September 2005 14:25 just because you can hear a difference between 192K and 96K doesn't mean that 192 is automatically better. You have to find out why there is a difference. What if you substitute 96 for 192 and 48 for 96 in that quote? ALMOST... Good 48 sounds extremely good. 96 is not proportionally that much better than 48 as 48 is bettter than 44. Let me rephrase that, 48 is significantly better sounding than 44. 96 is slightly better-sounding than 48. However, ALL OF US are making these evaluations on converters. I now have found evidence to believe that many of the converters themselves prejudice 96K over 44 and 48K. A superior DAC design that I now own has made 44 and 48 much closer in sound to 96 than ever before. Evidence like this makes me think twice, three times about the future and the practical aspects. Let's find out WHY (and IF) before making a headlong rush to a VERY expensive option.

Fascinating! I have never been able to hear much difference on a CD from a 48 file that has been dithered down to 44.1 and a 44.1 file. So maybe I don't need better ears, just better D/A convertors?

[Bob Olhsson]

Ethan Winer wrote on Fri, 16 September 2005 13:25

...People who lack scientific training often do not understand the limitations of non-blind testing,..

Frankly I find just as many people who don't understand the limitations of blind testing.

[Ronny]

Bob Olhsson wrote on Sat, 17 September 2005 09:03

Ethan Winer wrote on Fri, 16 September 2005 13:25 ...People who lack scientific training often do not understand the limitations of non-blind testing,.. Frankly I find just as many people who don't understand the limitations of blind testing.

Bottom line is the consumer is always listening in the blind. What is more important, what we think or what they perceive? This reminds me of the old annie versus digi debate. It's not that one is better than the other, it's that both are available to use for their individual benefits. More data, the more valid the results. ITR, blind testing/non blind testing, short term/long term, are best used in conjunction with each other, just like annie and digi together, for many apps, are often better than either one separate of the other.

[JamSync]

bobkatz wrote on Fri, 16 September 2005 21:09

I now have found evidence to believe that many of the converters themselves prejudice 96K over 44 and 48K.

And that's precisely the problem when using a set of the same converters running at different rates to record the same performance to various digital formats. While not as questionably valid as using different converters, it's definitely part of the same continuum of discrepancy.

[wireline]

So, if I may state the (to me) obvious:

A set of higher end 48 or 44.1 convertors will capture many things (including timbre, which is always a subjective call) with a better/more accurate representation than low-mid convertors operating at 96 or higher (Bob, that IS what you said, wasn't it?)

The whole thing about conversion rates has become a circular debate, with every aspect of the debate being questioned...timbre on some instruments while not others, sample rates, blind-test methodology....

So, in short - what IS the answer to the questions?  Does 16/44.1 capture most timbres, with acceptable losses, or do we as "the industry" have a compelling need to go higher?  Do the benefits of higher rates offset current production 'techniques' resulting in better music?

Just really curious...

[Bob Olhsson]

Ronny wrote on Sat, 17 September 2005 10:54

...Bottom line is the consumer is always listening in the blind. What is more important, what we think or what they perceive?...

A blind test is not the same thing as being an ordinary listener who is (hopefully always) unaware of the technology being employed. Audio artifacts remain a fatiguing annoyance to people who aren't knowledgeable enough to identify them specifically. A blind test where the listener can't easily be trained in precisely what to listen for is meaningless because it will always produce random results until such a time as the listener has stumbled across how to identify the artifact.

I've almost always found people's experiences with sound to be legitimate even when their ideas about the technology behind what they hear is off the wall.