The No.1 Website for Pro Audio
The Optimal Sample Rate for Quality Audio
Old 11th May 2012
  #1
Gear Head
 
🎧 10 years
The Optimal Sample Rate for Quality Audio

Interested in the facts?

One of the world’s top converter designers Dan Lavry has written a new paper in simple language to demystify the subject.

http://www.lavryengineering.com/pdfs...lity_audio.pdf

Find out why “more” is not always “better!”
Old 12th May 2012
  #2
Gear Addict
 
NotchontheRocks's Avatar
 
🎧 10 years
Good article. Love how he uses language and terms that beginners and veterans alike can understand.

Sent from my DROID BIONIC using Gearslutz App
Old 12th May 2012
  #3
Lives for gear
 
MASSIVE Master's Avatar
 
Verified Member
🎧 15 years
Linked from my blog -- Thanks!
Old 12th May 2012
  #4
Audio Alchemist
 
Lagerfeldt's Avatar
 
Verified Member
3 Reviews written
🎧 15 years
Great! An expansion on this thread by Dan Lavry, I've been linking to several times:

https://gearspace.com/board/1234224-post72.html
Old 12th May 2012
  #5
Gear Addict
 
acorneau's Avatar
 
Verified Member
1 Review written
🎧 15 years
Very nice white paper.

I also linked to the paper from my facebook page.
Old 12th May 2012
  #6
Gear Maniac
 
Sonicologico's Avatar
 
🎧 15 years
Beautiful document, many thanks
Old 12th May 2012
  #7
Lives for gear
 
Alexey Lukin's Avatar
 
Verified Member
🎧 10 years
I like the paper, but it looks a bit speculative. I wish it had some measurements actually showing these harmful nonlinear distortions at ultrasonic frequencies or degradation of the conversion accuracy because of too fast sampling. There's also a contradiction: Dan is saying that the ideal A/D and D/A chain is like a wire, but later he shifts his ideal to the wire with ultrasonics filtered out?
Old 14th May 2012 | Show parent
  #8
Gear Guru
 
Verified Member
🎧 15 years
Quote:
Originally Posted by Alexey Lukin ➡️
I like the paper, but it looks a bit speculative. I wish it had some measurements actually showing these harmful nonlinear distortions at ultrasonic frequencies or degradation of the conversion accuracy because of too fast sampling. There's also a contradiction: Dan is saying that the ideal A/D and D/A chain is like a wire, but later he shifts his ideal to the wire with ultrasonics filtered out?
Another issue I have with this theory is that most modern converters sample in the MHz range and only reduce the rate in the digital domain (which isn't subject to speed issues in electronics). They also use the same modulator rate for all sample rates that are multiples of 44.1Khz (5,6 Mhz) and the same for 48 Khz multiples (6.1 Mhz).

Also Dan writes:
Quote:
Originally Posted by Dan Lavry
"Good conversion requires attention to capturing and reproducing the range we hear while filtering and keeping out energy in the frequency range outside of our hearing. At 44.1 KHz sampling the flatness response may be an issue. If each of the elements (microphone, AD, DA and speaker) limit the audio bandwidth to 20 KHz (each causing a 3dB loss at 20 KHz), the combined impact is -12dB at 20 KHz."
I've checked the specs of a few flagship converter chips and they are all pretty much flat to 20Khz. (Here are some examples: Burr Brown: -1 dB @ 21.3Khz, Asahi Kasei: +- 0.1 dB @ 20Khz, Cirrus Logic: -0.1 dB @ 20727 Hz, Wolfson: +- 0,005 dB @ 20021 Hz). So for these converters at least there is no 3 dB loss at 20Khz.

Most monitoring speakers are flat to 20 Khz (and anyway, this doesn't affect the recorded signal).

Alistair
Old 14th May 2012 | Show parent
  #9
Gear Head
 
🎧 10 years
Quote:
Originally Posted by Alexey Lukin ➡️
I like the paper, but it looks a bit speculative. I wish it had some measurements actually showing these harmful nonlinear distortions at ultrasonic frequencies or degradation of the conversion accuracy because of too fast sampling. There's also a contradiction: Dan is saying that the ideal A/D and D/A chain is like a wire, but later he shifts his ideal to the wire with ultrasonics filtered out?
This may appear to be inconsistent except that the subject of the paper is "Quality Audio" and by definition; ultrasonic frequencies are not considered to be "audio."
Old 15th May 2012 | Show parent
  #10
Lives for gear
 
Verified Member
🎧 15 years
Quote:
Originally Posted by UnderTow ➡️

I've checked the specs of a few flagship converter chips and they are all pretty much flat to 20Khz. (Here are some examples: Burr Brown: -1 dB @ 21.3Khz, Asahi Kasei: +- 0.1 dB @ 20Khz, Cirrus Logic: -0.1 dB @ 20727 Hz, Wolfson: +- 0,005 dB @ 20021 Hz). So for these converters at least there is no 3 dB loss at 20Khz.
When I was in college, the accepted definition for bandwidth was the point at which the output falls by 3dB. I would guess that Dan is using the same definition here.

James.
Old 15th May 2012
  #11
Gear Guru
 
SWAN808's Avatar
 
1 Review written
🎧 10 years
cool. not sure you can state it as 'fact' tho - more like Dans researched/proposed opinion. Each expert seems to have a slightly different outlook as far as I can tell.

Good article tho.
Old 15th May 2012 | Show parent
  #12
Gear Head
 
🎧 10 years
Quote:
Originally Posted by UnderTow ➡️
Another issue I have with this theory is that most modern converters sample in the MHz range and only reduce the rate in the digital domain (which isn't subject to speed issues in electronics). They also use the same modulator rate for all sample rates that are multiples of 44.1Khz (5,6 Mhz) and the same for 48 Khz multiples (6.1 Mhz).

Also Dan writes:

I've checked the specs of a few flagship converter chips and they are all pretty much flat to 20Khz. (Here are some examples: Burr Brown: -1 dB @ 21.3Khz, Asahi Kasei: +- 0.1 dB @ 20Khz, Cirrus Logic: -0.1 dB @ 20727 Hz, Wolfson: +- 0,005 dB @ 20021 Hz). So for these converters at least there is no 3 dB loss at 20Khz.

Most monitoring speakers are flat to 20 Khz (and anyway, this doesn't affect the recorded signal).

Alistair
Although most contemporary audio converters are a multi-bit sigma-delta based design; the fact that the modulator operates at much higher sample rates than the output sample rate does not mean that analog issues are not part of the reason why there is an optimal sample rate. And the fact that the modulator output undergoes digital processing does not mean there is no compromise at that stage of conversion.

Here is Dan Lavry’s response:
“The sigma delta modulator speed is determined by various technology tradeoffs. For example, the size of the many capacitors embedded in the IC is very limited. In fact, the modulator is optimized for the available technology. There are 3 main parameters – modulator speed, number of modulator bits and filter order. A given combination offers the ability to design a noise shaping structure.

Say you have a modulator operating at 6MHz, with 5 bits, and a fifth order filter. If you wanted to end up with 3MHz usable signal bandwidth, there is no room for noise shaping because 3MHz is already at the Nyquist frequency of the 6MHz front end. So the outcome is 5 bits quality (with 3MHz bandwidth). If the modulator were 1 bit, you have a 1 bit converter (to 3MHz). Neither of these examples is good for audio.

Now say you wanted to reduce the usable bandwidth to 1MHz. That enables some noise shaping – moving noise from the 0-1MHz range to the 1-3MHz range. The noise shaping will yield somewhat better performance for signals under 1MHz, while making the range over 1MHz not usable (that is where you send the noise to using noise shaping). Now let’s accommodate a 30KHz range. Clearly there is a lot more room for the noise shaping (30KHz to 3MHz) so the region under 30KHz can be made much better quality (more bits).

So once again, the end result for lower bandwidth ends up with better performance.

The tradeoff between speed and accuracy is alive and well inside the modulator, and the examples I used (charging caps and settling opamps) applies very well to those circuits. The slew rates inside the modulator are very fast and settling times are critical. The modulator speed is, as well as the combination of the rest of the modulator parameters are optimized for a given technology, and the modulator is a PART of the process. The end result is still the same; more speed (signal bandwidth) means less accuracy.”

The fact that some converters are designed to optimized the frequency response near 20kHz when operating at 44.1 kHz does not mean that there is not some trade-off involved to the audio quality.

Dan Lavry’s response regarding the frequency response of DA converters:
“I agree that many IC makers try and push the response to be very flat at 20KHz, and that is also an issue for quality audio. Trying to pass 20KHz to a very flat .1dB while rejecting 22.05KHz is very demanding for the digital filter (for both AD and DA). A couple of KHz transition from passing to total rejection of audio has the potential to cause problems. By contrast; with a 96Khz sample rate, the transition is around an order of magnitude lower.”
Old 15th May 2012 | Show parent
  #13
Gear Guru
 
Verified Member
🎧 15 years
Hi Dan (and whomever is passing on his responses),

First, thanks for the extensive response!

Quote:
Originally Posted by Lavrytech ➡️
Although most contemporary audio converters are a multi-bit sigma-delta based design; the fact that the modulator operates at much higher sample rates than the output sample rate does not mean that analog issues are not part of the reason why there is an optimal sample rate. And the fact that the modulator output undergoes digital processing does not mean there is no compromise at that stage of conversion.

Here is Dan Lavry’s response:
“The sigma delta modulator speed is determined by various technology tradeoffs. For example, the size of the many capacitors embedded in the IC is very limited. In fact, the modulator is optimized for the available technology. There are 3 main parameters – modulator speed, number of modulator bits and filter order. A given combination offers the ability to design a noise shaping structure.

Say you have a modulator operating at 6MHz, with 5 bits, and a fifth order filter. If you wanted to end up with 3MHz usable signal bandwidth, there is no room for noise shaping because 3MHz is already at the Nyquist frequency of the 6MHz front end. So the outcome is 5 bits quality (with 3MHz bandwidth). If the modulator were 1 bit, you have a 1 bit converter (to 3MHz). Neither of these examples is good for audio.

Now say you wanted to reduce the usable bandwidth to 1MHz. That enables some noise shaping – moving noise from the 0-1MHz range to the 1-3MHz range. The noise shaping will yield somewhat better performance for signals under 1MHz, while making the range over 1MHz not usable (that is where you send the noise to using noise shaping). Now let’s accommodate a 30KHz range. Clearly there is a lot more room for the noise shaping (30KHz to 3MHz) so the region under 30KHz can be made much better quality (more bits).

So once again, the end result for lower bandwidth ends up with better performance.

The tradeoff between speed and accuracy is alive and well inside the modulator, and the examples I used (charging caps and settling opamps) applies very well to those circuits. The slew rates inside the modulator are very fast and settling times are critical. The modulator speed is, as well as the combination of the rest of the modulator parameters are optimized for a given technology, and the modulator is a PART of the process. The end result is still the same; more speed (signal bandwidth) means less accuracy.”
I think I didn't make myself clear: I understand the points in your article about the analogue stages but I don't see how that applies to the practical world of modern multi-bit sigma-delta modulator converter designs. My point/question wasn't so much about the much higher rate of the modulator or the lack of any compromises in the digital conversion stages but rather about the fact that the modulator rate is constant for all target sample rates subdivisions of the modulator rate. If the target rate is 44.1 Khz, 88.2 Khz or 176.4 Khz, the modulator is always running at 5.6 Mhz. If the target rate is 48 Khz, 96 Khz or 192 Khz, the modulator rate is always 6.1 Mhz.

The modulator is delivering the exact same signal whether the target rate is 44.1, 88.2 or 176.4 Khz (or higher). In light of this, how do the issues you mention with the analogue electronic components affect the different target sample rates differently? If your point was about a difference between 44.1/88.2/176.4 and 48/96/192 I could understand as they have different modulator rates but I do not understand how this would work between say 44.1 Khz and 88.2 Khz.

I also don't see how software algorithms are affected by the speed of the analogue stages? I don't believe they are. Either the software (and the computing hardware it runs on) runs properly or it breaks don't catastrophically. Unlike analogue, there is no graceful degradation of performance. It works or it doesn't. A computer CPU running at 5 Mhz is no more accurate than one running at 3Ghz. The analogue speeds of the components have no bearing on the digital software accuracy. (Of course there are other compromises on the software/digital side but those do not pertain to the points made in the article being discussed as far as I am aware).

Quote:
The fact that some converters are designed to optimized the frequency response near 20kHz when operating at 44.1 kHz does not mean that there is not some trade-off involved to the audio quality.
Of course. My comment was specifically about that alleged -3 dB point at 20 Khz (and thus a cumulative loss of 12 dB at 20Khz in the example given) mentioned in the article. That doesn't seem to exist in the real world of modern converters and studio monitor speakers. (At least the ones I am aware of). If there are other compromises being made I would be interested in hearing any more information about them but I don't think it helps the cause against ever higher sampling rates to present arguments that do not apply in the real world of modern converters.

Quote:
Dan Lavry’s response regarding the frequency response of DA converters:
“I agree that many IC makers try and push the response to be very flat at 20KHz, and that is also an issue for quality audio. Trying to pass 20KHz to a very flat .1dB while rejecting 22.05KHz is very demanding for the digital filter (for both AD and DA). A couple of KHz transition from passing to total rejection of audio has the potential to cause problems. By contrast; with a 96Khz sample rate, the transition is around an order of magnitude lower.”
I'm not sure what "very demanding" means technically. To me the only question is: Are the designs achieving what they set out to achieve or not? Based on what I know it seems that they do.

I fully understand what you are saying in theory but it seems that in practise, according to any of the more extensive listening experiments I have taken part in or read about, with good converters people can't actually distinguish between a 44.1 Khz ADC -> DAC loop and a straight wire. In light of that, how serious are these audio problems? Isn't, if not 44.1 Khz, at least 48 Khz more than sufficient for all our conversion needs[1]?

Also, purely on a technical point, the ADC chips I have checked use the same filter slope for 44.1/48 Khz, 88.2/96 Khz and 176.4/192 Khz. [2] I don't mean the same cut-off point in absolute values but the same cut-off point as a fraction of FS. Of course if FS is doubled (or quadrupled) any issues with filtering would be moved even further outside the audible band. I am just trying to figure out what you mean by "very demanding". Surely if there was any overt technical issues with the filter slope/processing load on the systems (like overheating or such), the chip designers would have relaxed the filters slopes at higher rates. Or am I missing something?

Also, again purely on a technical point, it seems the flagship chips I have checked have a transition band on their filters of around 4 to 5 Khz (for 44.1 Khz operation). This means there will be a bit of aliasing of frequencies between 22,05 Khz and whatever the stop-band is (up to 25.6 Khz on the Cirrus Logic CS5381 which was the highest of the chips I checked). Do you see this as a problem or would you consider it a good compromise?

Thanks in advance,

Alistair

[1] With the obvious exceptions of things like recordings for scientific purposes or recordings intended for, for instance, pitch shifting or other such processing for sound design purposes.

[2] With the exception of the Wolfson WM8786 chip which uses a different filter slope for quad rate conversion. (But the same at single and double rate speeds).
Old 15th May 2012
  #14
Gear Guru
 
Verified Member
🎧 15 years
Quote:
Originally Posted by japanmandu ➡️
"Lavrytech"? Are you actually representing Lavry Engineering, or what?

If you're here ostensibly representing a company / selling stuff, aren't you supposed to identify yourself?
I would contend that the name lavrytech does exactly that. I suppose a signature with more details might be appropriate.

Quote:
With all due respect, you've cross-posted your thread to other Gearslutz forums, and you're putting forth the veiled accusation that other companies' products are somehow inferior, and that they are in a conspiracy to defraud everyone about sample rate. That's a little ridiculous if you think about it.
Is it? Why?

Quote:
A smart guy I know [Bruno Putzeys] once explained to me that a designers "advice" is really only reliable when he or she is referring to his or her own designs.
With all due respect to Bruno, although I understand the sentiment, the way it is written is I think a little simplistic. It would mean that no engineer could ever comment or criticise any other's design. I think that is a bit silly.

Alistair
Old 16th May 2012 | Show parent
  #15
Gear Head
 
🎧 10 years
Quote:
Originally Posted by UnderTow ➡️
Hi Dan (and whomever is passing on his responses),

First, thanks for the extensive response!



I think I didn't make myself clear: I understand the points in your article about the analogue stages but I don't see how that applies to the practical world of modern multi-bit sigma-delta modulator converter designs. My point/question wasn't so much about the much higher rate of the modulator or the lack of any compromises in the digital conversion stages but rather about the fact that the modulator rate is constant for all target sample rates subdivisions of the modulator rate. If the target rate is 44.1 Khz, 88.2 Khz or 176.4 Khz, the modulator is always running at 5.6 Mhz. If the target rate is 48 Khz, 96 Khz or 192 Khz, the modulator rate is always 6.1 Mhz.

The modulator is delivering the exact same signal whether the target rate is 44.1, 88.2 or 176.4 Khz (or higher). In light of this, how do the issues you mention with the analogue electronic components affect the different target sample rates differently? If your point was about a difference between 44.1/88.2/176.4 and 48/96/192 I could understand as they have different modulator rates but I do not understand how this would work between say 44.1 Khz and 88.2 Khz.

I also don't see how software algorithms are affected by the speed of the analogue stages? I don't believe they are. Either the software (and the computing hardware it runs on) runs properly or it breaks don't catastrophically. Unlike analogue, there is no graceful degradation of performance. It works or it doesn't. A computer CPU running at 5 Mhz is no more accurate than one running at 3Ghz. The analogue speeds of the components have no bearing on the digital software accuracy. (Of course there are other compromises on the software/digital side but those do not pertain to the points made in the article being discussed as far as I am aware).



Of course. My comment was specifically about that alleged -3 dB point at 20 Khz (and thus a cumulative loss of 12 dB at 20Khz in the example given) mentioned in the article. That doesn't seem to exist in the real world of modern converters and studio monitor speakers. (At least the ones I am aware of). If there are other compromises being made I would be interested in hearing any more information about them but I don't think it helps the cause against ever higher sampling rates to present arguments that do not apply in the real world of modern converters.



I'm not sure what "very demanding" means technically. To me the only question is: Are the designs achieving what they set out to achieve or not? Based on what I know it seems that they do.

I fully understand what you are saying in theory but it seems that in practise, according to any of the more extensive listening experiments I have taken part in or read about, with good converters people can't actually distinguish between a 44.1 Khz ADC -> DAC loop and a straight wire. In light of that, how serious are these audio problems? Isn't, if not 44.1 Khz, at least 48 Khz more than sufficient for all our conversion needs[1]?

Also, purely on a technical point, the ADC chips I have checked use the same filter slope for 44.1/48 Khz, 88.2/96 Khz and 176.4/192 Khz. [2] I don't mean the same cut-off point in absolute values but the same cut-off point as a fraction of FS. Of course if FS is doubled (or quadrupled) any issues with filtering would be moved even further outside the audible band. I am just trying to figure out what you mean by "very demanding". Surely if there was any overt technical issues with the filter slope/processing load on the systems (like overheating or such), the chip designers would have relaxed the filters slopes at higher rates. Or am I missing something?

Also, again purely on a technical point, it seems the flagship chips I have checked have a transition band on their filters of around 4 to 5 Khz (for 44.1 Khz operation). This means there will be a bit of aliasing of frequencies between 22,05 Khz and whatever the stop-band is (up to 25.6 Khz on the Cirrus Logic CS5381 which was the highest of the chips I checked). Do you see this as a problem or would you consider it a good compromise?

Thanks in advance,

Alistair

[1] With the obvious exceptions of things like recordings for scientific purposes or recordings intended for, for instance, pitch shifting or other such processing for sound design purposes.

[2] With the exception of the Wolfson WM8786 chip which uses a different filter slope for quad rate conversion. (But the same at single and double rate speeds).
Here is Dan's response (Alistair’s words in quotes):
“I'm not sure what "very demanding" means technically. To me the only question is: Are the designs achieving what they set out to achieve or not? Based on what I know it seems that they do.”
)

A very sharp linear phase filter requires very long delay (FIR), and has pre-ringing issue. The shape of the impulse for sharp filter causes higher side lobes amplitude. That only becomes an issue relatively rarely (music dependent) and for people with ability to hear very high frequencies. One can do the digital filtering with IIR, but doing that and keeping linear phase is a problem when the filter slope is steep. There are various schemes to help reduce those filter limitations, and some offer some improvements, but the performance of the stiff filter for 44.1KHz is not up to par with that of 88.2-96KHz filter.

“I fully understand what you are saying in theory but it seems that in practice, according to any of the more extensive listening experiments I have taken part in or read about, with good converters people can't actually distinguish between a 44.1 KHz ADC -> DAC loop and a straight wire. In light of that, how serious are these audio problems? Isn't, if not 44.1 KHz, at least 48 KHz more than sufficient for all our conversion needs[1]?”

For my old ears, 44.1KHz is great. I am trying to accommodate all ears, and there are reports of few people that can actually hear slightly above 20KHz. I do think that 48KHz is pretty good compromise, but 88.2 or 96KHz yields some additional margin. Some audio people don’t agree that 44.1KHz yields a wire-like result.

“Also, purely on a technical point, the ADC chips I have checked use the same filter slope for 44.1/48 KHz, 88.2/96 KHz and 176.4/192 KHz. [2] I don't mean the same cut-off point in absolute values but the same cut-off point as a fraction of FS. Of course if FS is doubled (or quadrupled) any issues with filtering would be moved even further outside the audible band. I am just trying to figure out what you mean by "very demanding". Surely if there was any overt technical issues with the filter slope/processing load on the systems (like overheating or such), the chip designers would have relaxed the filters slopes at higher rates. Or am I missing something?”

It is easy to keep the same filter on an IC and just move the filter cutoff when changing the rate. It is also easier to make a 192KHz IC and add a X2 decimator to come up with 96KHz, and another X2 for 48Khz (or 44.1KHz derived from 176.4KHz). But it is not the optimum. A converter optimized for say 48KHz sampling (24KHz “audio” bandwidth in theory) will not be able to accommodate 96KHz (the noise shaping of 48KHz automatically makes frequencies over 24KHz “full of noise”). A converter that can do 192KHz has a built in compromise- the noise shaper is optimized to accommodate 96KHz bandwidth, not 20-30KHz audio the ear needs.

“Also, again purely on a technical point, it seems the flagship chips I have checked have a transition band on their filters of around 4 to 5 KHz (for 44.1 KHz operation). This means there will be a bit of aliasing of frequencies between 22,05 KHz and whatever the stop-band is (up to 25.6 KHz on the Cirrus Logic CS5381 which was the highest of the chips I checked). Do you see this as a problem or would you consider it a good compromise?”

I think it is a compromise to allow aliasing to the hearing range. This is one of the reasons why flagship converters are not a reference for the optimal quality. Once manufacturers agreed to accommodate 192KHz, they agreed to make compromises. All the 192KHz IC’s also offer 174.6, 96, 48 and 44.1KHz output sample rates; but they are built with a noise shaper for up to 96KHz of signal bandwidth. If they used the same parameters to make the best say 30KHz signal bandwidth (60-70KHz sampling), the results would be better optimized for audio. We don’t need 90KHz signals in audio…


And one more point: I brought the examples of cap charging and settling speeds as 2 examples. These are only 2 examples of many. I don’t have the time and space to go through a detailed analysis of sigma delta. I pointed out that the modulator speed is a decision based on available technology. In theory, faster modulator is very beneficial – it multiplies the effectiveness of noise shaping by a heck of a lot. The modulator rates have been increasing. The old DSD was 64fs and now some IC’s go at 512fs, even higher. So why not 4096fs or more? Because of the drawbacks of speed/accuracy tradeoffs due to the technology limitations (including the 2 analog examples I used and much more).

My statement about the speed/accuracy tradeoff is wide reaching. I am not ready to claim that the principle covers all aspects of life; although usually taking one’s time to do something offers the ability to accomplish a better job. But in electronics, the speed/accuracy tradeoff is solid. Alistair seems to be very pragmatic - interested in end results so he can appreciate what I said- that there is an optimum rate, and that faster does not always result in better accuracy. My main point is that there is an OPTIMAL RATE, not that faster is better. Alistair’s “liking” 44.1KHz actually supports my argument that too fast is ridicules. In this day and age where mp3 is so widely used, I would be pleased to see the CD format hold its ground. And for people that want the highest quality, I want to have some reasonable margin to optimize audio for the most sensitive and critical ears.

Dan
Old 16th May 2012 | Show parent
  #16
Gear Head
 
🎧 10 years
Originally Posted by japanmandu
"Lavrytech"? Are you actually representing Lavry Engineering, or what?

If you're here ostensibly representing a company / selling stuff, aren't you supposed to identify yourself?


Quote:
Originally Posted by UnderTow ➡️
I would contend that the name lavrytech does exactly that. I suppose a signature with more details might be appropriate.

With all due respect, you've cross-posted your thread to other Gearslutz forums, and you're putting forth the veiled accusation that other companies' products are somehow inferior, and that they are in a conspiracy to defraud everyone about sample rate. That's a little ridiculous if you think about it.


Is it? Why?

A smart guy I know [Bruno Putzeys] once explained to me that a designers "advice" is really only reliable when he or she is referring to his or her own designs.

With all due respect to Bruno, although I understand the sentiment, the way it is written is I think a little simplistic. It would mean that no engineer could ever comment or criticise any other's design. I think that is a bit silly.

Alistair
Thanks for your balanced approach, Alistair.

LavryTech represents the writing of more than one person at Lavry Engineering. It seemed rather obvious to us too, and we were not trying to hide anything. We are not trying to "sell" anything; except knowledge of the facts.
Old 16th May 2012 | Show parent
  #17
Lives for gear
 
DSD_Mastering's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by Lavrytech ➡️
[COLOR="Blue"]We are not trying to "sell" anything; except knowledge of the facts.
And here all along I though Dan sold converters. Who'da thunk?

BTW... there are 2-3 more forums you haven't posted on yet.
Old 16th May 2012
  #18
Gear Guru
 
Verified Member
🎧 15 years
Quote:
Originally Posted by DSD_Mastering ➡️
And here all along I though Dan sold converters. Who'da thunk?
With a name like DSD_Mastering you clearly don't have a horse in this race riiiight...

Seriously guys, if you don't have any technical arguments or questions to present, please don't needlessly clutter up an interesting thread.

Thanks,

Alistair
Old 16th May 2012 | Show parent
  #19
Gear Addict
 
Storm Mastering's Avatar
 
🎧 10 years
Quote:
Originally Posted by UnderTow ➡️
I think I didn't make myself clear: I understand the points in your article about the analogue stages but I don't see how that applies to the practical world of modern multi-bit sigma-delta modulator converter designs. My point/question wasn't so much about the much higher rate of the modulator or the lack of any compromises in the digital conversion stages but rather about the fact that the modulator rate is constant for all target sample rates subdivisions of the modulator rate. If the target rate is 44.1 Khz, 88.2 Khz or 176.4 Khz, the modulator is always running at 5.6 Mhz. If the target rate is 48 Khz, 96 Khz or 192 Khz, the modulator rate is always 6.1 Mhz.

The modulator is delivering the exact same signal whether the target rate is 44.1, 88.2 or 176.4 Khz (or higher). In light of this, how do the issues you mention with the analogue electronic components affect the different target sample rates differently? If your point was about a difference between 44.1/88.2/176.4 and 48/96/192 I could understand as they have different modulator rates but I do not understand how this would work between say 44.1 Khz and 88.2 Khz.

I also don't see how software algorithms are affected by the speed of the analogue stages? I don't believe they are. Either the software (and the computing hardware it runs on) runs properly or it breaks don't catastrophically.
I'm absolutely not an expert in converters, but I would tend to think with this fixed modulator rate that if your target rate is smaller, you use more information per sample, and so this sample is more precise (statistically, if you mesure X times the same thing, and take the mean of those mesures, you will have a smaller imprecision).

For Digital also, just think about the FFT : for a given series of samples : you can gain precision in the time domain, at the expense of been less precise in the frequency domain, and vice versa.
Old 16th May 2012 | Show parent
  #20
Gear Head
 
🎧 10 years
Quote:
Originally Posted by DSD_Mastering ➡️
And here all along I though Dan sold converters. Who'da thunk?

BTW... there are 2-3 more forums you haven't posted on yet.
Another great example of the quality of the arguments against the assertion that there is an optimal sample rate for quality audio.

Yes, Lavry Engineering does sell converters; but nowhere in the paper did we even imply that Lavry converters are "better" than other brands because of the existence of a trade-off between speed and accuracy.

Taking things out of context is a waste of everyone's time and energy.
Old 16th May 2012
  #21
Mastering Moderator
 
Riccardo's Avatar
 
Verified Member
2 Reviews written
🎧 15 years
We all know Dan sells converters, this doesn't mean he cannot express his thoughts, opinions, views, and technical topics.
Let's all carry on in a civil manner please.
Old 18th May 2012
  #22
Gear Head
 
🎧 10 years
From Dan Lavry:

I have been making the case against higher sample rates for audio for a long time. I have encountered no credible arguments to my paper “Sampling Theory”. The same is true for my recent paper “The Optimal Sample Rate for Quality Audio”. I encounter some that want to counter the message by “shooting the messenger”. Meanwhile the facts I preset are correct and UN-challenged. I realize that reading the papers demands time and concentration. So here is a shorter description of many of the points I presented in the papers. Let’s refrain from diverting the conversation away from the topics.

1. Sampling is not intuitive. SAMPLING IS NOT ANALOGUS TO PIXELS! A more detailed picture may require more pixels, but more audio detail does NOT require more samples. There is an “electronic tool” (filter) that enables recovering ALL of the audio from a limited number of samples. It is not intuitive and requires much study. In fact it is counter-intuitive and goes against “everyday common sense.” This is the reason why the marketing of “more samples is better” is successful in convincing so many of the false notion.

2. Nyquist theorem (theorem is a PROVEN theory) tells us that recovering ALL the audio intact does require the sampling rate (frequency of sampling) to be at least twice as fast as the highest signal (audio) frequency. Theory demands a perfect “reconstruction tool” filter. In practice, real world filters require sampling a little faster than twice the audio bandwidth. For 20 KHz audio bandwidth, the theory requires at least 40 KHz sample rate. The 44.1 KHz standard provides 4.1 KHz margin. The margin for the filter (from the theoretical filter) is 100*(44.1KHz-2*20KHz)/(2*20KHz) = 10.25%

3. Some people argue that we need more than 20 KHz for audio. The decision as to how wide the audio range is should be left to the ears. Say we agree to accept a 25 KHz as the audio bandwidth. When using 88.2 KHz sampling, (and 25 KHz for the audio bandwidth) the margin is i100*(88.2KHZ-2*25KHz) /(2*25KHZ) = 76.4%.

4. At 96 KHz sampling and 25 KHz audio, the margin is 92%. At 96 KHz sampling and 30 KHz audio the margin is 60%. At 192KHz sampling and 30KHz the margin is 220%!. For anyone crazy enough to claim they hear or feel 40 KHz, when sampling at 192 KHz the margin is still 140%. At 384 KHz sampling the margin is 380%!

5. Some argue that at 44.1 KHz the margin of 10.25% is tight, and that theoretical filters fail to provide a near perfect reconstruction. Others argue that 20 KHz audio is too small to accommodate some ears. Such arguments support some reasonable increase in sampling rate. Many argue that 44.1 KHz rate is good enough. Others disagree. But few will argue with the statement that 44.1 KHz is at least pretty close to acceptable. In order to accommodate those that want improvements, let’s increase the margin by a factor of say 2. You want more, OK, by a factor of 4. You want more audio bandwidth? OK let’s raise it to a factor of 5… And all that is more than covered by the use of 96 KHz sample rate!

6. A few manufacturers are starting to advocate 384 KHz and even 768 KHz sample rates. When audio sampled at 44.1KHz is considered as being somewhere between “not perfect” and “near perfect”, the notion of sampling 870% faster (for 384KHz) or even 1741% (for 768KHz) faster than a CD makes no sense. I expect even the least competent of designers to be able to design a filter that does not require such huge margins. I would also expect any converter designer to have enough background to know that more samples are not analogous to more pixels! I would expect converter designers to insist that their marketing department knows that, instead of closing their eyes to the crock of steering audio in the wrong direction. I also understand it is not easy when one’s job is on the line.

7. It is not wise to keep increasing the sample rate unnecessarily. The files keep growing, and faster sampling yields less accuracy. Yet the marketing of higher sample rates has no basis, other than some spreading of misinformation. The latest I saw claims that faster sampling yields better stereo location (time resolution). The argument is false. Faster sampling offers the ability to process wider bandwidth, but has no impact what so ever on stereo location!

8. Faster sampling for capturing bandwidth that we do not hear (ultrasonic) is not wise. If we did not hear it (or feel it) we don’t need it. If we did hear it (or feel it) it is not ultrasonic, it is audible bandwidth (by definition). Ultrasonic energy may cause problems by spilling over to the audible range (intermodulation distortions). At best case, ultrasonic energy adds nothing to audio while requiring faster sampling, thus larger files and slower file transfers. In reality there is another price to pay; the faster one samples, the less accurate the result.

Dan Lavry
Old 19th May 2012 | Show parent
  #23
Here for the gear
 
🎧 10 years
Here's a quote from Benchmark Media (the makers of the Benchmark DAC1) that fully corroborates what Dan is saying:


"What Converter Manufacturers Don’t Want You to Know!
An examination of converter IC data sheets will reveal that virtually all audio converter ICs deliver their peak performance near 96 kHz. The 4x (176.4 kHz and 192 kHz) mode delivers poorer performance in many respects. In most cases, noise, distortion, pass-band ripple, stop-band attenuation and other key performance measurements are significantly better in the 2X (88.2 kHz and 96 kHz) mode of operation. Every A/D and D/A conversion IC that we have tested performs better at 96 kHz than at 192 kHz. In most cases THD+N, SNR, passband ripple, and stopband attenuation are all poorer at 192 kHz than at 96 kHz. Based upon these tests, I am not surprised that there is not yet any conclusive evidence that 192 kHz is better than 96 kHz. Given the current state of the art, 192 kHz should sound poorer than 96 kHz. 192 kHz provides additional bandwidth between 48 kHz and 96 kHz but there is no real evidence that this is useful given the limitations of our microphones, speakers, and hearing. 192 kHz adds useless bandwidth while decreasing performance."

So while Lavry has the courage of their conviction and don't offer a 192k converter, Benchmark does but they apparently optimize their 192k converters for 110k as they resample everything to 110k including 192k and 176k.
Old 29th May 2012 | Show parent
  #24
Gear Addict
 
16/44.1's Avatar
 
🎧 15 years
Quote:
Originally Posted by mstan ➡️
Here's a quote from Benchmark Media (the makers of the Benchmark DAC1) that fully corroborates what Dan is saying:


"What Converter Manufacturers Don’t Want You to Know!
An examination of converter IC data sheets will reveal that virtually all audio converter ICs deliver their peak performance near 96 kHz. The 4x (176.4 kHz and 192 kHz) mode delivers poorer performance in many respects. In most cases, noise, distortion, pass-band ripple, stop-band attenuation and other key performance measurements are significantly better in the 2X (88.2 kHz and 96 kHz) mode of operation. Every A/D and D/A conversion IC that we have tested performs better at 96 kHz than at 192 kHz. In most cases THD+N, SNR, passband ripple, and stopband attenuation are all poorer at 192 kHz than at 96 kHz. Based upon these tests, I am not surprised that there is not yet any conclusive evidence that 192 kHz is better than 96 kHz. Given the current state of the art, 192 kHz should sound poorer than 96 kHz. 192 kHz provides additional bandwidth between 48 kHz and 96 kHz but there is no real evidence that this is useful given the limitations of our microphones, speakers, and hearing. 192 kHz adds useless bandwidth while decreasing performance."

So while Lavry has the courage of their conviction and don't offer a 192k converter, Benchmark does but they apparently optimize their 192k converters for 110k as they resample everything to 110k including 192k and 176k.
Good to read this all, now i know for shure.
Glad for having a DAC1.
Old 5th June 2012
  #25
Lives for gear
 
chrisdee's Avatar
 
9 Reviews written
🎧 10 years
In the first paragraph on page 3 he writes that 60 khz is close to the ideal sampling rate. He also writes that 96khz in the real world ends up with a bandwith of approximately 40 khz. This would be 20 khz below the ideal sampling rate, right ?

Does this mean that 120 khz would be the ideal sampling frequency to record at in the digital domain (because of the Nyquist–Shannon sampling theorem) ?
Old 5th June 2012
  #26
Lives for gear
 
sat159p1's Avatar
 
3 Reviews written
🎧 10 years
What will you choose: capturing the signal at 44.1 or 88.2 (and the for CD purposes, go back to 44.1)?
Old 5th June 2012 | Show parent
  #27
Lives for gear
 
Verified Member
🎧 10 years
Quote:
Originally Posted by chrisdee ➡️
In the first paragraph on page 3 he writes that 60 khz is close to the ideal sampling rate. He also writes that 96khz in the real world ends up with a bandwith of approximately 40 khz. This would be 20 khz below the ideal sampling rate, right ?

Does this mean that 120 khz would be the ideal sampling frequency to record at in the digital domain (because of the Nyquist–Shannon sampling theorem) ?
No, Dan says that 60kHz is the ideal sample rate, not signal bandwidth.

He's aiming for the minimum sample rate that he can achieve an acceptably flat response from 0-20kHz, the extra 10kHz allows for filtering that gives enough attenuation at the Nyquist frequency without excessive ripple in the pass band.
Old 5th June 2012 | Show parent
  #28
Lives for gear
 
chrisdee's Avatar
 
9 Reviews written
🎧 10 years
Quote:
Originally Posted by Jon Hodgson ➡️
No, Dan says that 60kHz is the ideal sample rate, not signal bandwidth.

He's aiming for the minimum sample rate that he can achieve an acceptably flat response from 0-20kHz, the extra 10kHz allows for filtering that gives enough attenuation at the Nyquist frequency without excessive ripple in the pass band.
Does this mean 88khz would be more ideal than 96khz ?
Im not shure I understand the difference between sample rate and bandwith.
Old 5th June 2012 | Show parent
  #29
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by chrisdee
In the first paragraph on page 3 he writes that 60 khz is close to the ideal sampling rate. He also writes that 96khz in the real world ends up with a bandwith of approximately 40 khz. This would be 20 khz below the ideal sampling rate, right ?

Does this mean that 120 khz would be the ideal sampling frequency to record at in the digital domain (because of the Nyquist–Shannon sampling theorem) ?

Quote:
Originally Posted by Jon Hodgson ➡️
No, Dan says that 60kHz is the ideal sample rate, not signal bandwidth.

He's aiming for the minimum sample rate that he can achieve an acceptably flat response from 0-20kHz, the extra 10kHz allows for filtering that gives enough attenuation at the Nyquist frequency without excessive ripple in the pass band.
IOW, "because of the Nyquist–Shannon sampling theorem", to achieve an ideal bandwidth of 30kHz, you need a sample rate of 60kHz. But that's not commonly used of course, so we need to use the next best available sample rate: 88.2kHz or 96kHz.

I'm hoping this thread might survive without someone mentioning digital stepping. Or dither. Or aliasing due to non-linear processing.

Dan is only talking about two things:

1. Recording (capture) of digital audio
2. Playback of digital audio.

It is worth noting that subsequent, typical digital processing done by audio engineers (which happens between recording and playback) is subject to additional criteria related to linearity. Although he has very useful, valid opinions on those aspects, Dan isnt talking about ITB processing here.

Sean
Old 5th June 2012 | Show parent
  #30
Lives for gear
 
Verified Member
🎧 10 years
Quote:
Originally Posted by chrisdee ➡️
Does this mean 88khz would be more ideal than 96khz ?
If Dan designed the converter, then probably. However someone else's filters might have worse passband response than his and moving the cutoff frequency up a bit might have greater benefits than the detriment of letting more noise through.

BUT, a couple of important things to consider here..

1) Unless the designer has made a ballsup we're talking miniscule differences here, ones which you'd probably be unlikely to hear in even the most favourable conditions.

2) Dan concerns himself purely with the optimum sample frequency for capturing and playing back audio, not with anything else. For example depending on what you're doing and how your plugins work, a higher sample rate might be preferable for processing. Some people also report that the lower monitoring latencies achievable with higher sample rates give better performances when tracking.


Quote:
Im not shure I understand the difference between sample rate and bandwith.
The normal response would be to say that the sample rate is double bandwidth, as per Nyquist.

But that's a little simplistic when talking real world, Shannon showed that if you have a signal bandlimited to < f, then you can capture all that information by sampling at 2f. However getting that bandlimited signal in the first place is a challenge.

In audio we generally refer to bandwidth as being the part where the frequency response is "flat" (the standard is to the frequencies with 3dB rolloff, but it's not uncommon to specify band width with less deviation than that), which is going to be some way away from the frequency at which the signal dwindles away to effectively nothing.

So if you want a sampling system with a bandwidth of 20kHz, you're going to have a signal which is bandlimited to somewhere > 20kHz, and need to sample it at double that. Dan is of the opinion that he can get a bandwidth of 20kHz by bandlimiting at 30kHz, and so states 60kHz as his optimum sample rate.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 290 views: 54773
Avatar for David Spearritt
David Spearritt 30th January 2021
replies: 64 views: 9425
Avatar for dan le
dan le 18th February 2012
replies: 2517 views: 344428
Avatar for Skamm Goodiez
Skamm Goodiez 4 days ago
Topic:
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump