Author Topic: what's the point of recording at 24/192? (Read 8976 times)

JasonSobel · « **on:** January 11, 2005, 08:27:42 AM »

Hey folks, this is sort of a theoretical question...

let me start by saying, I'm looking forward to the next wave of 24 bit recorders as much as anyone else. it'll be great to finally use the V3 at 24 bit, to stop buying DAT tapes, and be recording at 24/96... and as far as 24 bit recordings go, a few months ago, I bought myself a DVD burner, discwelder Bronze, and Denon DVD player to play DVD-audio discs. my playback consists of DVD player > digi out > grace 901 > grado rs2 headphones.

I have about 30 shows in 24 bit format right about now. most are at 24/48, but several are also at 24/96. now, I don't have a direct, head-to-head comparison to base this on, but it is hard for me to discern the differences between 24/48 and 24/96. the difference between 16 bit and 24 bit is huge for me. so I'm thinking, am I really going to be able to tell the difference between 24/96 and 24/192

or is this question more along the lines of the DAT question, why record at 16/48 instead of 16/44.1? just because you can? and in theory, it's a higher resolution recording? but the differences in media space between 16/44.1 and 16/48 was minimal. a 24/192 recording will be twice as large (file size) as a 24/96 recording. is it really worth it? maybe it doesn't seem like a big deal to me because the grace 901 only goes up to 24/96? but even with 24/192, I'm sure my Denon isn't too shabby at D/A, and could output an analog signal to the 901. I just don't think the benefits gained from 24/192 over 24/96 outweigh the double the file size...

just sort of thinking aloud here. I'd like to hear everyone elses thoughts on this.

- Jason

MattD · « **Reply #1 on:** January 11, 2005, 08:59:04 AM »

I will step up and say that I bet I couldn't pass an ABX test with 24/96 vs. 24/192 on any playback system. Around that point, the weak link becomes my ears, rather than the system. I think the extra file size (and editing time) won't be worth the benefit of a higher samplerate.

BC · « **Reply #2 on:** January 11, 2005, 11:50:19 AM »

Yeah, we are probably pushing the limits of hearing as well as reaching a point of quickly diminishing returns past 24/96. I would say that the best argument so far for it would be just because you can, and that you can always throw away resolution but can never get it back. I would think maybe 24/192 would make sense for high quality acoustic recording in ideal environments but for taping PA systems probably a waste of time and storage.

Ben

wbrisette · « **Reply #3 on:** January 11, 2005, 12:26:13 PM »

Quote from: BC on January 11, 2005, 11:50:19 AM

Yeah, we are probably pushing the limits of hearing as well as reaching a point of quickly diminishing returns past 24/96.

Anybody who really thinks that 24/48 or 24/96 is inferior to 24/192 needs to study Nyquist theorem. There are reasons you can't hear a difference between 96 and 192. In fact, odds are good that 48 to 96 won't be heard either, the ever so slight increase that can theoretically be heard usually isn't.

Wayne

muj · « **Reply #4 on:** January 11, 2005, 12:37:59 PM »

dan lavry proves here that 24/192 is a waste of time
http://www.lavryengineering.com/documents/Sampling_Theory.pdf

scb · « **Reply #5 on:** January 12, 2005, 11:10:22 AM »

this is an interesting read

http://www.earthworksaudio.com/f_wpapers/beyond20khz.html

relaxing · « **Reply #6 on:** January 12, 2005, 03:12:10 PM »

So you can detect the presence of bats flying overhead at outdoor venues?

Lil Kim Jong-Il · « **Reply #7 on:** January 12, 2005, 03:39:05 PM »

Reposted from my post in the FR2 thread:

Nyquist gets misused all the time for justifying a lower sampling rate. The maximum reproduced frequency is only part of the equation when reproducing multi channel sound. quoted from http://www.24bitfaq.org/ :

0.3.5 Which is said to provide greater benefit to the listener over conventional 16-bit 44.1/48kHz recordings: Increased sampling rate or increased word length?

A: Increased word length. From Ken Pohlmann’s “Principles of Digital Audio,” 4th edition, p390-391:

“The use of high sampling rates such as 96 and 192kHz may seem unnecessary. In rare cases, a person may be able to hear frequencies of 24 or 26kHz, far below the cutoff frequencies of 48 and 96kHz [for 96 and 192kHz sampling rates respectively]. In most cases, high frequency hearing response is below 20kHz. Thus, for steady-state tones, the higher frequency response may not be useful. However, it can be argued that high sampling frequencies improve binaural time response, leading to improved imaging. For example, if short pulses are applied to each ear, a 15 microseconds difference between the pulses can be heard, and that time difference is shorter than the time between two samples at 48kHz. Some people can hear a 5 microseconds difference, and that corresponds to the time difference between two samples at 192kHz. In theory, this high sampling rate may improve spatial imaging. Thus, it may take two ears to distinguish between a recording at 48kHz, and one at 192kHz.”

wbrisette · « **Reply #8 on:** January 12, 2005, 03:47:54 PM »

Quote from: Maynard G. Beatnik on January 12, 2005, 03:39:05 PM

Reposted from my post in the FR2 thread:

Nyquist gets misused all the time for justifying a lower sampling rate.

For our uses, I would challenge anybody to tell the difference in blind hearing test (using identical equipment) to tell the difference between 24/96 and 24/196. My guess is that there maybe one or two people who can tell the 24/48 vs. 24/96. Personally for me, it's not worth the extra drive space (and I have space to spare) for the very, very minute differences that might be heard.

I can record at 24/96, but choose 24/48. In my listening tests, I haven't been able to tell the difference with the type of material I've been recording and in the venues that I have been recording.

This is what I urge (and continue to urge people to do) folks to do. Make their own recordings, listen to them in a critical listening environment, then make their own judgements. I think that really is the best test equipment that can be used. Not some lab gear, but the gear God gave us that is between our two ears. Listen, figure out for yourself if it's worth it, then make your choice based on what you hear.

Wayne

jpschust · « **Reply #9 on:** January 12, 2005, 03:55:44 PM »

this is a question and not meant to be an assertation of truth, however wouldn't conversion from 192 to 44.1 theoretically be a more accurate representation of the original sound than 96 to 44.1 since it would contain more data from which to draw its conversion?

Lil Kim Jong-Il · « **Reply #10 on:** January 12, 2005, 04:09:38 PM »

I agree that very many people will not be able to perceive a difference either because they do not have adequate playback equipment or they do not have adequate hearing. But that would not stop me from using 192kH sampling rate. If I were to argue against using 192kHz, I would argue that we are unlikely to encounter a concert PA source that presents a sufficiently detailed image to justify that much resolusion in the recording. I thik that the reason most people don't hear a huge difference between 48kHz and 96kHz in their PA recordings is simply because the spatial information in those settings is not that detailed.

However, I think that an orchestral setting would present sufficient spatial detail such that 192kHz would be worthwhile. Capturing that resolution would allow more precise placement of the individual instruments in an accoustic recording. Whether you or I could hear it given our ears is immaterial to me. If the spatial information is there and can be captured, I would think that it is valid to record at those higher rates.

My point was that the application of Nyquist Theorm as is common in these threads is not a fully proper justification for stereo channel recording sampling rates. When I read comments like recording of bats or that no microphone can pick up 64kHz sounds, I have to make this point to remind people that there is a more complicated model at work here than if we were recording a point source into a single channel.

wbrisette · « **Reply #11 on:** January 12, 2005, 04:17:57 PM »

Quote from: Maynard G. Beatnik on January 12, 2005, 04:09:38 PM

However, I think that an orchestral setting would present sufficient spatial detail such that 192kHz would be worthwhile. Capturing that resolution would allow more precise placement of the individual instruments in an accoustic recording. Whether you or I could hear it given our ears is immaterial to me. If the spatial information is there and can be captured, I would think that it is valid to record at those higher rates.

Well, let me tell about one individual up in the SF area of California. He is doing orchestral recordings for a living on a Deva V. He and I have had some pretty interesting conversations because we are two of the only people using these things outside the film industry. The Deva V can record at 24/192. However, he only does 24/96. Even with top dollar gear and full access to the orchestra, he told me that 24/96 is about all he will use because he doesn't hear, nor do his clients any difference. Now, I understand the arguments for and against it. If you want to record at 24/192, knock yourself out. Personally I'll stick with what I'm doing because I don't hear the difference (and I'm using QTC-1 mics in my setup 4 Hz - 40 KHz) and that end the end is what it is all about.

Wayne

Lil Kim Jong-Il · « **Reply #12 on:** January 12, 2005, 04:28:00 PM »

Fair enough. There is a big difference between a decision based on emperical results and arguing theory.

In all honesty, I'll probably never run higher than 96kHz but mostly because that is what I can put in an audio-DVD. I do plan to run a full suite of tests when then 722 gets here.

edit: Wayne, I'd +T you for this discussion but you must have said something I liked in another thread because I have to wait 12 hrs.

heath · « **Reply #13 on:** January 12, 2005, 04:36:30 PM »

for our purposes I agree wholeheartedly that there really is no reason to go higher than 96k. For the archival work I do, however, 192 is a demand that most of my clients make. We are in the business of archiving, however, and we are also working with studio productions. FWIW, in the field, I run 48k. At work, it's either 96k or 192k. Some of our clients also ask for 88.2k

H

wbrisette · « **Reply #14 on:** January 12, 2005, 05:13:21 PM »

Quote from: heath on January 12, 2005, 04:36:30 PM

for our purposes I agree wholeheartedly that there really is no reason to go higher than 96k. For the archival work I do, however, 192 is a demand that most of my clients make. We are in the business of archiving, however, and we are also working with studio productions. FWIW, in the field, I run 48k. At work, it's either 96k or 192k. Some of our clients also ask for 88.2k

And that's what do, you give the client what they want. 88.2, those usually are old ADAT folks who had to use multi 44.1 channels to get higher resolution. I'm running 48K because that's what I've really listened to that makes the most sense for me. Here is the test, I'll offer to folks.

Last Thursday I ran my 4 mics at Cooper's Uncle (local bluegrass band). The QTC-1 mics were on stage clamped just below their mics, thus you have the QTC-1's as the main mics. Then I ran my SR-77 mics at stage lip to pick up the other stuff. I just put this on DVD-A. This was recorded at 24/48.

This Thursday I have told them I would record them again, in the same venue. I'll setup my mics the same way again, but I'll run the DEVA at 24/96. Then burn it on DVD-A as well.

I'll mark the DVD's as 1 and 2 and ship them out to anybody who wants to do a blind test. Then they can tell me which is which... If anybody is up for this, just drop me a line and I'll get you setup.

Wayne

wbrisette · « **Reply #15 on:** January 12, 2005, 05:15:01 PM »

One more thing +T to everybody for their thoughts on this. This has been one of the more useful discussions about this that I have seen in a long time.

Wayne

zowie · « **Reply #16 on:** January 12, 2005, 06:51:51 PM »

Are there perhaps some additional benefits to higher sampling rates for minimizing degradation from post processing? There are for bit depths beyond 24.

Lil Kim Jong-Il · « **Reply #17 on:** January 12, 2005, 07:04:26 PM »

Quote from: zowie on January 12, 2005, 06:51:51 PM

Are there perhaps some additional benefits to higher sampling rates for minimizing degradation from post processing? There are for bit depths beyond 24.

Are you talking about actual quantization bit depths at greater 24-bit?

JasonSobel · « **Reply #18 on:** January 12, 2005, 07:29:04 PM »

Quote

For example, if short pulses are applied to each ear, a 15 microseconds difference between the pulses can be heard, and that time difference is shorter than the time between two samples at 48kHz. Some people can hear a 5 microseconds difference, and that corresponds to the time difference between two samples at 192kHz. In theory, this high sampling rate may improve spatial imaging. Thus, it may take two ears to distinguish between a recording at 48kHz, and one at 192kHz.

that something I hadn't thought about. +T to everyone who's posted some interesting stuff. those articles are pretty interesting.

Quote

Last Thursday I ran my 4 mics at Cooper's Uncle (local bluegrass band). The QTC-1 mics were on stage clamped just below their mics, thus you have the QTC-1's as the main mics. Then I ran my SR-77 mics at stage lip to pick up the other stuff. I just put this on DVD-A. This was recorded at 24/48.

This Thursday I have told them I would record them again, in the same venue. I'll setup my mics the same way again, but I'll run the DEVA at 24/96. Then burn it on DVD-A as well.

I'll mark the DVD's as 1 and 2 and ship them out to anybody who wants to do a blind test. Then they can tell me which is which... If anybody is up for this, just drop me a line and I'll get you setup.

Wayne

Wayne - I'd be interested in this, thanks for the offer. question for you. with your 4 mics and the DEVA, I'm assuming you recorded 4 channels, yes? did you mix it to stereo? or are you burning surround DVD-audio discs? I ask, because the only thing I really have to listen to these is my headphones. (and for this to be truely blind, I'll have to use a notecard to cover up the sample rate lights on the 901. if I forget to do that, it'll tell me which is which

)

Quote

In all honesty, I'll probably never run higher than 96kHz but mostly because that is what I can put in an audio-DVD. I do plan to run a full suite of tests when then 722 gets here.

why can't you put 24/192 onto a DVD-audio disc? DVD-audio supports 2-channel 24/192, although surround sound only up to 24/96. (someone please correct me if I'm wrong)

wbrisette · « **Reply #19 on:** January 12, 2005, 07:35:32 PM »

Quote from: JasonSobel on January 12, 2005, 07:29:04 PM

Wayne - I'd be interested in this, thanks for the offer. question for you. with your 4 mics and the DEVA, I'm assuming you recorded 4 channels, yes? did you mix it to stereo? or are you burning surround DVD-audio discs? I ask, because the only thing I really have to listen to these is my headphones. (and for this to be truely blind, I'll have to use a notecard to cover up the sample rate lights on the 901. if I forget to do that, it'll tell me which is which )

I'll end up mixing them to 2 channel, primarily because that's all I use. They get mixed later in Digital Performer.

Wayne

Lil Kim Jong-Il · « **Reply #20 on:** January 12, 2005, 08:02:53 PM »

Quote from: JasonSobel on January 12, 2005, 07:29:04 PM

Quote
In all honesty, I'll probably never run higher than 96kHz but mostly because that is what I can put in an audio-DVD. I do plan to run a full suite of tests when then 722 gets here.

why can't you put 24/192 onto a DVD-audio disc? DVD-audio supports 2-channel 24/192, although surround sound only up to 24/96. (someone please correct me if I'm wrong)

I'll put them to "audio-DVD" using the DVDV format. DVDV only supports 48k and 96k LPCM. I don't want to use DVDA format unless I have to because I want discs to play in all DVD players and most of the currently deployed DVD players do not support DVDA formatted discs. It's easy to confuse the two but "audio-DVD" has become the defacto term for a DVDV disk with just lpcm audio on it.

I got a 24/96 of Ball for chistmas and it plays in all DVD players. It is reported as DVDV disc in my player, not DVDA. So apparently this is the way the studios are addressing the compatability issue too.

Tim · « **Reply #21 on:** January 12, 2005, 09:31:22 PM »

great stuff here guys, I've been following this one very closely.... +T's to the thread!

heath · « **Reply #22 on:** January 12, 2005, 09:45:09 PM »

wayne,
I would love to hear this. I'll shoot you a PM... I have a feeling this test will reinforce my reasoning for running 48 in the field.

big +t for the offer. and for that DEVA [drool]

I would love to check out your setup some time

wboswell · « **Reply #23 on:** January 13, 2005, 01:57:58 PM »

Quote from: Maynard G. Beatnik on January 12, 2005, 08:02:53 PM

Quote from: JasonSobel on January 12, 2005, 07:29:04 PM
Quote
In all honesty, I'll probably never run higher than 96kHz but mostly because that is what I can put in an audio-DVD. I do plan to run a full suite of tests when then 722 gets here.

why can't you put 24/192 onto a DVD-audio disc? DVD-audio supports 2-channel 24/192, although surround sound only up to 24/96. (someone please correct me if I'm wrong)

I'll put them to "audio-DVD" using the DVDV format. DVDV only supports 48k and 96k LPCM. I don't want to use DVDA format unless I have to because I want discs to play in all DVD players and most of the currently deployed DVD players do not support DVDA formatted discs. It's easy to confuse the two but "audio-DVD" has become the defacto term for a DVDV disk with just lpcm audio on it.

I got a 24/96 of Ball for chistmas and it plays in all DVD players. It is reported as DVDV disc in my player, not DVDA. So apparently this is the way the studios are addressing the compatability issue too.

Very interesting... I've had issues with my Denon 2900 and DVD-R discs burned as DVD-A in discwelder. The player will play the discs and output correctly if I use the 5.1 outputs, but when I try to output a digital signal, only the left main channel plays. However, the player will output a digital signal for store bought DVD-A discs such as Robert Cray's "Time Will Tell" The packaging says DVD-A but it may actually be DVDV. FWIW, this album was released by Sanctuary, the same label that WSP's Ball was release under...

That being said, what software do you use to burn to DVDV?

Thanks!

Lil Kim Jong-Il · « **Reply #24 on:** January 13, 2005, 07:06:17 PM »

Quote from: wboswell on January 13, 2005, 01:57:58 PM

Very interesting... I've had issues with my Denon 2900 and DVD-R discs burned as DVD-A in discwelder. The player will play the discs and output correctly if I use the 5.1 outputs, but when I try to output a digital signal, only the left main channel plays. However, the player will output a digital signal for store bought DVD-A discs such as Robert Cray's "Time Will Tell" The packaging says DVD-A but it may actually be DVDV. FWIW, this album was released by Sanctuary, the same label that WSP's Ball was release under...

That being said, what software do you use to burn to DVDV?

Thanks!

I don't use a windows installation for personal computing, so I am currently wasting time trying to get the commandline tools under linux to work but I keep getting periodic static within the DVDs that I have authored using that approach. I think it has to do with the A/V stream multiplexing prior to authoring step. I want to get that to work more than anything else.

I'm not in any hurry to buy SW until the 722 ships, but when that day comes I'm going to buy the audio-dvd-creator. I know someone here who is using it and he is having success. This weekend I'm going to try the demo version under wine. Look in the computer recording forum and there is a ton of info.

I'm using a 2900 also. Nice unit.

Karl · « **Reply #25 on:** January 13, 2005, 10:58:09 PM »

One thing that I find very interesting is that everybody talks about how superior analog recording can be. What bit depth and sample rate will it take to convince people that digital can be superior?

Tim · « **Reply #26 on:** January 14, 2005, 09:35:54 AM »

Quote from: Karl on January 13, 2005, 10:58:09 PM

One thing that I find very interesting is that everybody talks about how superior analog recording can be. What bit depth and sample rate will it take to convince people that digital can be superior?

from what I've read it's DSD that is finally convincing people that digital can be as good as analog

zowie · « **Reply #27 on:** January 14, 2005, 10:46:10 AM »

OK, to get back to the original topic, here are a few excerpts and paraphrases from what Bob Katz says about using high sample rates in Mastering Audio. (ALL advanced tapers should buy this book!) This is just to get you the flavor, it doesn't do justice to his 6 page discussion of the topic.

"[H]ow can 50-year-old ears detect differences between 44.1 kHz and 96 kHz and even 192 Khz sample rates, even though most of us can't hear much about 15 kHz? I believe the answer lies in the design of digital low-pass filters which are part of the requirements of digital audio. [ . . .] One type of filter has a sharp cutoff; the consequences of sharp filtering include time-smearing of the audio, possible short (millisecond) echos wich are caused by amplitude response ripples in the passband frequency response (20 Hz - 20 kHz), . . . Moving the filter cutoff frequency to 48 kHz (for 96 kHz sampling rate) relaxes the filtering requirement and makes it easier to engineer filters with less ripple in the passband and less phase shift near the upper frequency limit."

To answer my question:
"Moorer [of Sonic Solutions] pointed out that post-processing, such as filtering, equalization, and compression, will result in less distortion in the audible band [at 96 kHz] as the errors are spread over twice the bandwith -- and half of that bandwidth is above 20 kHz."

"In addition . . if after processing the destiantion is DVD-A or SACD, then the master can be left at the higher sample rate and worklength, avoiding another generation of sound-veiling 16-bit dither and yet another sharp filter at the end of the process. Thus, consumers should not scoff at DVDs which have been digitall remastered from original 16/44.1 sources. They will be getting real, audiophile-quality sonic value in their remaserters."

While there would probably be diminishing returns, I [zowie, not Katz] would imagine that the above applied equally to 192 kHz. Katz suggests that 96kHz would perhaps be a minimal sampling rate if there is a redesign in PCM converters to make themer more psychoacoutically correct than they are today.

I would think that higher sampling rate would especially benefit those of us using consumer/prosumer equipment (M-Audio, Edirol, Sony) and software that probably has relatively lower quality digital filters compared to the big dollar rack gear.

relaxing · « **Reply #28 on:** January 14, 2005, 12:21:15 PM »

Quote from: zowie on January 14, 2005, 10:46:10 AM

To answer my question:
"Moorer [of Sonic Solutions] pointed out that post-processing, such as filtering, equalization, and compression, will result in less distortion in the audible band [at 96 kHz] as the errors are spread over twice the bandwith -- and half of that bandwidth is above 20 kHz."

I think I understand what he's getting at, but I just want to be clear -- what errors is he speaking of?
Can you provide more context?

BobW · « **Reply #29 on:** January 14, 2005, 04:48:15 PM »

192kHz is big for forensics and EFX work, when slowed playback down might occur.
96kHz is definitely discernable, though ever so slightly. 192kHz is deinitely overkill to my ears.

The DAC is much more important, IMHO.

Author Topic: what's the point of recording at 24/192? (Read 8976 times)

jpschust

zowie

zowie

BobW