HydrogenAudio

Hydrogenaudio Forum => Scientific Discussion => Topic started by: NullC on 2012-03-06 02:14:53

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: NullC on 2012-03-06 02:14:53
Of interest, but certainly not news to anyone here is Monty's educational piece regarding the lack of value of 24/192 as a distribution format:

http://people.xiph.org/~xiphmont/demo/neil-young.html (http://people.xiph.org/~xiphmont/demo/neil-young.html)

Some of the commentary on the internet has been not especially well informed, for example there are some crazy claims at https://news.ycombinator.com/item?id=3668310 (https://news.ycombinator.com/item?id=3668310).  If you're in a "someone is wrong on the internet!" correcting mood, you might want to go leave some comments in furtherance of the collective intelligence of mankind.

Cheers
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: NullC on 2012-03-06 04:36:45
It's also on slashdot now: http://news.slashdot.org/story/12/03/06/00...ds-is-pointless (http://news.slashdot.org/story/12/03/06/0048259/why-distributing-music-as-24-bit192khz-downloads-is-pointless)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: ExUser on 2012-03-06 04:52:51
(http://i.imgur.com/WNSQ1.gif)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: mudlord on 2012-03-06 05:08:16
(http://i.imgur.com/WNSQ1.gif)

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: DigitalMan on 2012-03-06 05:27:37
Very, very nicely done.  Long live TOS #8.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Carledwards on 2012-03-06 07:50:56
Excellent. Glad to see this and I hope it gets widespread attention.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Batman321 on 2012-03-06 08:00:32
Great article!   


Meanwhile in a parallel universe where science is for losers who aren't rich enough to buy insanely expensive gear:

SH Forums (http://www.stevehoffman.tv/forums/showthread.php?t=278868)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-06 08:32:29
At the risk of spinning off a digression: what is the maximum effective number of bits available in a DAC these days -- and does that number depend on the sampling frequency?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: skamp on 2012-03-06 09:27:52
^^^ http://nwavguy.blogspot.com/2012/03/odac-update.html (http://nwavguy.blogspot.com/2012/03/odac-update.html)

Quote
ENOB stands for Effective Number of Bits and is another measure of a DAC’s performance. No 24 (or 32) bit audio DAC can achieve true 24 bit performance, In fact, 20 ENOB is generally considered the “Holy Grail” of real world DAC performance. The ODAC is just under 19 ENOB and the Benchmark [DAC1], even referenced to its full 7+ volt maximum output, is 19.3 ENOB. The FiiO E10, even in 24 bit mode, is only 16.2 ENOB.


Note: the guy is touting his own design for a DAC/headphone amp combo (referenced above as ODAC), but he has a good track record of providing what looks like precise measurements, and making sense…
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: AndyH-ha on 2012-03-06 09:39:06
The answer depends on conditions, like the chem lab experiments. At standard temperature and pressure ... .
20 bits is pushing the limit for practical purposes, thermal noise is too high for anything better. For some endeavors, such as some aspects of radio astronomy, cutting edge particle physics, and military systems, cryogenic cooling is practical and common place. Liquid nitrogen, or even more extreme materials, with the low temperatures they provide, considerably reduce the intrinsic device noise levels.

Sampling frequency and bit depth are independent of each other.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: itisljar on 2012-03-06 10:25:27
Finally. Thank you.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: razer on 2012-03-06 12:16:34
Such a great read.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-06 12:30:03
Sampling frequency and bit depth are independent of each other.


So ... at least it is not the case that a DAC fed a 192/24 signal as oppsed to a 48/20 or 44.1/16, will have to operate at higher load --> more thermal noise --> lower e.n.o.b.?
(Shouldn't be, it is not that demanding?)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: DonP on 2012-03-06 12:59:27
Sampling frequency and bit depth are independent of each other.



Delta sigma?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Arnold B. Krueger on 2012-03-06 13:41:22
At the risk of spinning off a digression: what is the maximum effective number of bits available in a DAC these days -- and does that number depend on the sampling frequency?


Effective number - you mean in terms of resolution that is actually delivered to the analog domain?

Check the TI and ESS web sites. They seem to be flogging the SOTA in this regard the hardest.

I think that noise down 133 dB down is the SOTA - what about 21 bits?  If you want to be hard to please you would also demand that spurious responses (IM+THD) would also be part of the equation, in which case things seem to be about 3-6 dB worse.

I basically agree with the 20 bit number that others have put forth.

There is a strategy for improving the dynamic range of a DAC that can be expanded almost endlessly, at a increasingly high cost. Run multiple DACs that each generate statistically independent noise in parallel. Dynamic range improves by 3 dB for every doubling of DACs. I believe that ESS goes to 4x or even 8x for specing their products.

The progression is:

2 DACs in parallel, get a 3 dB improvement
4 DACs in parallel, get a 6 dB improvement
8 DACs in parallel, get a 9 dB improvement
16 DACs in parallel, get a 12 dB improvement
...

Thing is, this does nothing for spurious responses because they are always coherent, like the signal. The ratio between spurious responses and signal remains the same as you add DACs.

In the past people have run DACs in antiphase which can help even-order distortion. But you only get one iteration of that approach.

As DAC functionality gets cheaper and cheaper, the numbers race may cause these approaches to be more common.  But, a good single DAC is overkill enough!
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Arnold B. Krueger on 2012-03-06 13:45:14
Sampling frequency and bit depth are independent of each other.



Delta sigma?



After measuring about a hundred audio interfaces, the overwhelming majority being Sigma-Delta, it seems to be a general rule that the best effective performance is found at lower sample rates (e.g. 44 KHz), all other things being equal.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: dumdidum on 2012-03-06 13:52:28
Monty: "The more that pseudoscience goes unchecked in the world at large, the harder it is for truth to overcome truthiness... even if this is a small and relatively insignificant example."

Meanwhile in a parallel universe where science is for losers who aren't rich enough to buy insanely expensive gear:

SH Forums (http://www.stevehoffman.tv/forums/showthread.php?t=278868)

That SH forum thread you linked reminds me of creationism. I wouldn't be too optimistic about convincing such "audiophiles" by appealing to science and logic.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Martel on 2012-03-06 14:15:52
Sampling frequency and bit depth are independent of each other.
Theoretically maybe. But this is definitely false for integrating ADCs (commonly used in digital voltmeters) which have have a reciprocal relationship between accuracy (bit depth) and sampling rate.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Wombat on 2012-03-06 16:02:02
Some of the commentary on the internet has been not especially well informed, for example there are some crazy claims at https://news.ycombinator.com/item?id=3668310 (https://news.ycombinator.com/item?id=3668310).  If you're in a "someone is wrong on the internet!" correcting mood, you might want to go leave some comments in furtherance of the collective intelligence of mankind.

One more page to link to if asked for some explanation, many thanks for that. What saddens me is the resulting discussion on this news channel.
There the typical "I am enigineer" come in and simply claim that "Nyquist was wrong because i can hear all these harmonics!"
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: saratoga on 2012-03-06 17:26:16
Sampling frequency and bit depth are independent of each other.


So ... at least it is not the case that a DAC fed a 192/24 signal as oppsed to a 48/20 or 44.1/16, will have to operate at higher load --> more thermal noise --> lower e.n.o.b.?
(Shouldn't be, it is not that demanding?)


Usually they operate at about the same internal clock speed regardless of sampling rate, just with different oversampling ratios.  Wouldn't make sense to run a DAC made for high speed operation at a lower speed, since more oversampling means less risk of aliasing and generally less quantization noise. 

Also, in this context, thermal noise doesn't necessarily refer to the temperature of the A/D alone, but also to the temperature of everything its hooked up to. The terminating resistance in whatever the A/D is reading still contributes the same thermal noise no matter how cool you make the A/D for instance.  This is why refrigeration is not a great option like it is in optical detectors.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Satellite_6 on 2012-03-06 19:56:07
"It's hard to fake ambisonics or holographic audio, sort of like how 3D video always seems to degenerate into a gaudy gimmick that reliably makes 5% of the population motion sick."

"If it wasn't the most boring party trick ever, it was pretty close."

haha!

Very amusing and very informative. I just want 16/44.1 lossless downloads, so to whoever starts offering that on a wide scale, shut up and take my money. 

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: jensend on 2012-03-06 21:53:51
By the by, I recently added a little writeup about TOS 8 on the wiki (http://wiki.hydrogenaudio.org/index.php?title=TOS_8). Thought I'd mention it after seeing Monty referencing TOS 8.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: soulsearchingsun on 2012-03-06 22:39:19
From the article:
Quote
Audibility of a CD-Standard A/D/A Loop Inserted into High-Resolution Audio Playback [...]
Not one listener throughout the entire test was able to identify which was 16/44.1 and which was high rate, and the 16-bit signal wasn't even dithered!
Not to be nitpicky, but IMHO you only dither if you resample digitally?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: klonuo on 2012-03-06 23:04:13
By the by, I recently added a little writeup about TOS 8 on the wiki (http://wiki.hydrogenaudio.org/index.php?title=TOS_8). Thought I'd mention it after seeing Monty referencing TOS 8.

While there, someone could also fix links in pointed ABX article - all but one external links are 404
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: wakibaki on 2012-03-06 23:57:14
I'm happy to see a truly coherent argument against the necessity for higher bit depths and sample rates.

Waste of storage space is a small margin of disadvantage however. The advocates of 24/96 and higher were happy to pay the cost of increased storage space to achieve a perceived advantage, and storage space (or transmission bandwidth) is an increasingly cheap commodity.

This leaves us with degradation of fidelity as a result of processing unnecessary ultrasonics. Unfortunately the very same tests (Meyer and Moran) we used to make our case previously now mitigate against us. If a 16/44k1 bottleneck is inaudible in a system avowedly processing ultrasonics, then the ultrasonics cannot reasonably be said to have (audibly) degraded the sound.

w
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: NullC on 2012-03-07 00:34:29
This leaves us with degradation of fidelity as a result of processing unnecessary ultrasonics. Unfortunately the very same tests (Meyer and Moran) we used to make our case previously now mitigate against us. If a 16/44k1 bottleneck is inaudible in a system avowedly processing ultrasonics, then the ultrasonics cannot reasonably be said to have (audibly) degraded the sound.


There have been plenty of results that show 'audibility' of ultrasonics that result from failing to prevent the distortion (and a number of the presentation on the subject that have negative results spend a lot of time walking over all the external distortion sources they had to control for). You can probably easily generate such a result for yourself at home (at least the audio output of my laptop produces pretty impressive distortion when a 21khz signal that I can't hear is present).

Perhaps its ultimately not that big a deal— but that space is still space that could be better put to use providing surround, useful multitrack, or redundancy for error correction to make the files resistant to damage even if you do think that the space (and bandwidth) is cheap enough that it's not worth worrying about. Of course, if you're already track separated surround corruption protected files and still think the space is cheap— then by all means, give me 24/192— I can filter out the ultrasonics that would make my amp distort on my end.

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: wakibaki on 2012-03-07 00:54:24
I thought he made good points, but he quoted Meyer and Moran. I just think it's better that we're all informed as to the counter-arguments rather than ending up with egg on our faces.

I hate the way science-based electronics has been infected by this insanity, but there was a time when audio wasn't like this, and I believe the tide will turn.

w
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: pdq on 2012-03-07 02:41:23
Not to be nitpicky, but IMHO you only dither if you resample digitally?

You only dither if you go from a higher bit depth to a lower one. I don't recall if in that study the A/D was more than 16 bits.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: BearcatSandor on 2012-03-07 04:49:49
The article mentions that SACDs often sound better because they are 'mastered better'. This is something i've been wondering about.  Is it worth collecting DVD-As and SACDs because thay may be 'mastered better'? Does that mean they compare favorably with original master recording labels like Mobile Fidelity Sounds Labs, DCC and Audio Fidelity, or are they usually the same old victems of the loudness war just shoved into a bigger 24-bit evelope.  I'd think if they were higher quality (re)masters they'd advertise that on the packaging to grab the audiophile market.

This confuses me.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: BearcatSandor on 2012-03-07 06:50:56
I'm happy to see a truly coherent argument against the necessity for higher bit depths and sample rates.

Waste of storage space is a small margin of disadvantage however. The advocates of 24/96 and higher were happy to pay the cost of increased storage space to achieve a perceived advantage, and storage space (or transmission bandwidth) is an increasingly cheap commodity.

Have there been 16-bit/44.1k descrete surround sound formats that were sucessfully marketed? Woudn''t that seem like a logical thing?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: WernerO on 2012-03-07 08:04:15
Not to be nitpicky, but IMHO you only dither if you resample digitally?

You only dither if you go from a higher bit depth to a lower one.



You dither whenever resolution is reduced. This includes analogue to digital conversion as a limit case.

The analogue input of the ADC ideally contains a noise source at the required level.

In practice the front-end electronics and signal chain often provide this noise implicitly, although not
necessarily with the optimal spectral distribution.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: soulsearchingsun on 2012-03-07 10:00:52
Not to be nitpicky, but IMHO you only dither if you resample digitally?
You only dither if you go from a higher bit depth to a lower one.
Oops

You dither whenever resolution is reduced. This includes analogue to digital conversion as a limit case.

The analogue input of the ADC ideally contains a noise source at the required level.

In practice the front-end electronics and signal chain often provide this noise implicitly, although not
necessarily with the optimal spectral distribution.
Thank you, didn't know this. Sounds plausible.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: 2Bdecided on 2012-03-07 10:19:45
I think this is a fantastic article.


Without wishing to reduce it's value, or claim that the following is audible, there's one small point I'd take issue with. With wide bandwidth recordings, there are more ultrasonics to cause intermodulation distortion, it's true. BUT in most recordings of real instruments, the harmonics are just that - harmonics. Whereas with "badly" reproduced 44.1kHz audio, the high frequencies that escape through the anti-image filter aren't real harmonics at all, they're images of real harmonics.

Like it or not, with most commercial converters (A>D and D>A) the 20-24kHz range (especially the 21-23kHz range) isn't "clean" - there are easily measurable aliases and images - these are trivial to see with extreme source content.

If you filter 44.1kHz audio at about 20kHz, quite gently, ensuring little ringing, and no content above ~21kHz, then you're home and dry. The same would prevent IMD issues with wide bandwidth recordings.

Point is, if you're less careful in either case, the IMD from wide bandwidth recordings has a closer* harmonic relationship with the original music than the IMD from images+aliases+ringing.

Though on any reasonable measure, there's a lot more of it, so any theoretical advantage in terms of "benign"* spectral components is probably outweighed by their amplitude.

* - harmonics can still create some very weirdly related IMD - but aliases and images + IMD create even less well related spectral components.



But I love the article. I love the analogy with light.

Cheers,
David.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-07 11:57:32
My sixth sense (i.e., my faithful placebo) just took an hour break from its ever enduring task of convincing me how bad -V0 sounds,
just to enlighten Ye all on where the hi-res race will take us from here:

Kindly brought to you by Porcus Prophecies, Inc.
(with all reservations for mild exaggregations, a bias which my faithful placebo might sometimes be prone to)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: icstm on 2012-03-07 14:20:21
This article is a perfect answer to my FIRST (http://www.hydrogenaudio.org/forums/index.php?showtopic=93518&hl=) question        <-- me really happy


In the slashdot article there is a great comment that really starts to provide a robust defence to all those who don't want to beleive the science of it all

Quote
I also spent 4 years studying an EE degree, and although it was not especially focused on signal processing, I now work for a large pro audio company.

Some of the issues pointed to in this and other posts regarding oversampling and AA filters are not really relevant to the subject at hand, given the technology currently in use. A statement like 'oversampling at 192 kHz' shows a lack of knowledge regarding the kinds of audio converters that have been in use for a good while now. A Delta Sigma ADC running with an Fs of 48 kHz might often be oversampling at 3.072 MHz or 6.144 MHz. Anti aliasing filters that many people have mentioned are implemented digitally inside the converter (no need for external analog filters, which may well exhibit many of the problems mentioned), and actually have extremely good pass band ripple.

Look at datasheets for converters from manufacturers such as TI (burr brown) [ti.com], cirrus [cirrus.com] [page 36 here has detailed plots of 48, 96, and 192 kHz pass pand characterisitcs for the device, highlighting the fact that increasing the sampling rate does not improve pass band ripple for this device (also note the scale is 0.02 dB/div)], AKM [asahi-kasei.co.jp], Wolfson micro [wolfsonmicro.com] You will find pass band pass responses that are flat to within less than +/- 0.05 dB over the audible range, and stop band attenuation in excess of 100 dB, whether sampling at 48 kHz or 192 kHz. If you can find anything in actual converter datasheets that points to better converter performance from selecting a higher sampling rate, I would be interested to see it.

All in all, the basics of sampling theory don't really help people to understant the real world issues in designing a moden high end audio device. And in the end, surely the proof of the pudding is in the blind tests, that never seem to show that anybody can tell any difference when moving to higher rates? Even if there were a few people who could hear this difference in some perfect listening envirmonment, would it really make sense for everyone else to go out and buy 192 kHz equipment?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: xiphmont on 2012-03-09 08:12:01
(http://i.imgur.com/WNSQ1.gif)


Oooh, ow.  Guess it wasn't that good then, huh?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: xiphmont on 2012-03-09 09:12:30
I thought he made good points, but he quoted Meyer and Moran. I just think it's better that we're all informed as to the counter-arguments rather than ending up with egg on our faces.


I'm actually curious as to your specific objection/concern.  I've read the various critiques written by detractors of the BAS tests over the years, but too many of those arguments relied on willful obtuseness and eye rolling. I'd like to hear the methodology/implementation critiques from those who nevertheless agreed with the conclusions.

The point has also been made that [in the article] first I argue "ultrasonics hurt fidelity" and then cite M&M, which supposedly undermines the argument because no one could hear a difference.  In no way does M&M rebut the assertion that ultrasonics _can_ cause audible distortion.  They were using high end setups designed at expense for audiophile-grade frequency extension, and the results show they obviously weren't affected by audible IMD.  Am I missing something else?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-09 11:43:30
The point has also been made that [in the article] first I argue "ultrasonics hurt fidelity" and then cite M&M, which supposedly undermines the argument because no one could hear a difference.  In no way does M&M rebut the assertion that ultrasonics _can_ cause audible distortion.  They were using high end setups designed at expense for audiophile-grade frequency extension, and the results show they obviously weren't affected by audible IMD.  Am I missing something else?


This is kind of 'the negation of never isn't always but sometimes', right?

If some but not all hi-fi's are robust enough not to intermodulate when fed ultrasonics, while other fairly common equipment would be overloaded, degrade sound or even be damaged, then it isn't  ill-justified to dub an 88.2 or above format 'harmful'. It is a matter of degree, severity and even of opinion, but just the mere fact that some equipment are off the hook, does not invalidate the labeling.

(Remember, tobacco smoking probably kills only a [forty-something percent] minority of its users.)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: icstm on 2012-03-09 15:16:39
If some but not all hi-fi's are robust enough not to intermodulate when fed ultrasonics, while other fairly common equipment would be overloaded, degrade sound or even be damaged, then it isn't  ill-justified to dub an 88.2 or above format 'harmful'. It is a matter of degree, severity and even of opinion, but just the mere fact that some equipment are off the hook, does not invalidate the labeling.

(Remember, tobacco smoking probably kills only a [forty-something percent] minority of its users.)

how does it do damage? (ultrasonics, not smoking...)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: ExUser on 2012-03-09 15:36:58
Oooh, ow.  Guess it wasn't that good then, huh?
I've been waiting for this article for a long time, Monty. My deepest, sincerest thanks.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-03-09 15:45:53
how does it do damage? (ultrasonics, not smoking...)

I think he means to the sound, not the equipment. Otherwise you did read the presentation, right?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-09 16:05:57
how does it do damage? (ultrasonics, not smoking...)

I think he means to the sound, not the equipment. Otherwise you did read the presentation, right?


Meant sound (intermodulation distortion), cannot rule out that equipment can be damaged. First, clipping (or square/sharp pulses) are in a Fourier setting, overtones. They had better be filtered. Second, user won't turn down ultrasonics just because it's too loud for a tweeter.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-03-09 21:55:21
The article mentions that SACDs often sound better because they are 'mastered better'. This is something i've been wondering about.  Is it worth collecting DVD-As and SACDs because thay may be 'mastered better'? Does that mean they compare favorably with original master recording labels like Mobile Fidelity Sounds Labs, DCC and Audio Fidelity, or are they usually the same old victems of the loudness war just shoved into a bigger 24-bit evelope.


They might be.  Or might not be.  There is no guarantee and no way to know for sure in advance.

Quote
I'd think if they were higher quality (re)masters they'd advertise that on the packaging to grab the audiophile market.


For awhile, simply slapping 'SACD'  or 'high resolution' on the release was enough to convince gullible audiophiles that a remaster was 'better sounding'. 

(For some uneducated consumers, it still is.)

In its favor, the SACD 'Scarlet Book' format specification (unlike the CD 'Redbook' spec) discouraged loudness-war-type mastering for the DSD layer of SACDs.  But recording engineers know how to get around such minor obstacles.  At the most basic level, they can just smash the life out of the record before it's even converted for a  'hi rez' release.  That's what happened to Oasis 'What's the Story Morning Glory' -- the SACD was no more dynamic than the CD, because the master tape itself has already been leeched of dynamic range.  No degree of 'high resolution' remastering is going to increase the dynamic range of that recording. (Of course, the folks who mastered that album in the mid 1990s had no inkling there'd be an SACD version in its future.)

There are some verified cases where the DSD layer is a more 'dynamic' mastering than the Redbook layer of the same SACD disc -- Pink Floyd's 'Dark Side of the Moon' SACD is the famous example. This was a mastering choice, not a technical necessity.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-03-09 22:18:28
I thought he made good points, but he quoted Meyer and Moran. I just think it's better that we're all informed as to the counter-arguments rather than ending up with egg on our faces.


I'm actually curious as to your specific objection/concern.  I've read the various critiques written by detractors of the BAS tests over the years, but too many of those arguments relied on willful obtuseness and eye rolling. I'd like to hear the methodology/implementation critiques from those who nevertheless agreed with the conclusions.

The point has also been made that [in the article] first I argue "ultrasonics hurt fidelity" and then cite M&M, which supposedly undermines the argument because no one could hear a difference.  In no way does M&M rebut the assertion that ultrasonics _can_ cause audible distortion.  They were using high end setups designed at expense for audiophile-grade frequency extension, and the results show they obviously weren't affected by audible IMD.  Am I missing something else?


The supposed stake through the heart of M&M is that they didn't use 'any' (later changed to 'they didn't use ENOUGH') pure DSD (i.e., not sourced from analog tape or 'standard rez' digital) recordings as test material.  This ignores the fact that subjects were often allowed to pick the discs THEY thought best showed the difference.    It also conveniently 'forgets'  that SACDs were touted as inherently improving *even analog-sourced materials* -- and indeed, most non-classical SACDs were and are sourced from analog tapes.  They're only rarely new recordings.  (Some Sony SACDs were reportedly even sourced from Redbook masters.  )

So 'skeptical' complaints against M&M on these fronts really just constituted moving the goalposts. Suddenly only 'pure' DSD recordings reveal the benefits of the format!  (Leaving aside that the relatively large ultrasonic content of such recordings could cause distortion.  Though SACD players usually have 50- or 100kHz lowpass filters after the DAC, to cut out the REALLY ultra junk.)

That said, M&M absolutely *should* have described their methods in far more detail in the original article (that information dribbled out later).  And they *could* publish subset analysis of data to see if considering only the 'pure' DSD recordings made any difference in their findings.  Or release their raw data and let others work it over.  Or, preferably, someone could do another test and gather more data.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: LaserSokrates on 2012-03-09 22:35:50
An excellent read, thank you very much! A bit deeper (as in practical) than the stuff I heard at the university, but perfectly understandable nevertheless.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: xiphmont on 2012-03-10 02:12:19
That said, M&M absolutely *should* have described their methods in far more detail in the original article (that information dribbled out later).  And they *could* publish subset analysis of data to see if considering only the 'pure' DSD recordings made any difference in their findings.  Or release their raw data and let others work it over.  Or, preferably, someone could do another test and gather more data.


Excellent, thanks.  I'll ask about it at this month's meeting.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-03-10 22:09:14
Excellent, thanks.  I'll ask about it at this month's meeting.


No problem.  Loved your article, btw.

Dave Moran addressed some of the 'skeptics' head-on in a contentious exchange on the sa-cd.net forum. Here's one possible entry point into that:

http://www.sa-cd.net/showthread.php?page=1 (http://www.sa-cd.net/showthread.php?page=1)

And from this newer thread on that forum -- about your article -- one can see right off that the True Believers already consider M&M's study to have been 'debunked':

http://www.sa-cd.net/showthread.php?page=1 (http://www.sa-cd.net/showthread.php?page=1)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: hellokeith on 2012-03-10 23:57:48
Does the statement of the benefits of mixing/mastering >44/16 also extend to recording? Or record at 44/16 and then upconvert for mixing/mastering?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Wombat on 2012-03-11 01:54:48
At first i have to congrat Monty that he gave us this article.
For some this article might be the answer as the concept of the enemy responding to the famous braindead TAS articles spreaded lately.
Unfortunately it seems that in the places that were hearing different before this only made people trip for one moment and then they returned to old thinking.
There only has to come in one person to the discussion with some weird numbers no one challenges and this article is declared wrong.
On the other hand I did read over forums like computeraudiophile lately, since JimH pointed there and enjoyed the latest writings.
Greetings to Julf this way for his hard work btw.

So again i have to thank for this article that finaly set of an avalanche of discussion on several places on the web in the right direction
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: FreaqyFrequency on 2012-03-11 03:27:02
post


For the sake of the sanity and for spared minutes of the lives of everyone reading this thread, and for the love of humanity, do not read the thread corresponding to Monty's article on the SACD threads linked.  Please.

Please.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: julf on 2012-03-11 10:26:41
Greetings to Julf this way for his hard work btw.


Thanks, Wombat! I have to say that at times CA makes me lose my will to live... :-/

    Julf

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: moozooh on 2012-03-11 15:37:08
In my opinion, the problem we need to focus on is not 24/192 distribution per se, but rather not using the best engineering practices and/or features available for 16/44.1 content. The tests that compare 24/192 content with directly derived 16/44.1 are partly missing the point because the 16/44.1 content found elsewhere is not always derived in such fashion—sometimes deliberately so to appear worse. It's not a big secret that DVD-A and SACD are usually made from different masters that technically could  be downsampled to 16/44.1 without loss, but weren't because then there'd be no point buying the more expensive DVD-A and SACD.

As long as the CD is the lowest common denominator with the worst mastering and channel content, while vinyls continue having better headroom and SACDs more channels, the initiative is rather pointless. There is still reason to buy these alternatives because they may sound better for reasons completely unrelated to the physical limitations (or quirks) of the media. (Well, number of channels aside.)

But to be brutally honest, I don't believe in its success either way, because it's basically asking companies to make less money than they do, and I don't expect them to be interested.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Wombat on 2012-03-11 16:42:59
What about recent CD versions that obviously don´t sound as good as the HD release that is released around the same time like on HDtracks.com?
I think people that discover such a problem should send back their CD as BROKEN! Maybe this will help in a way?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-03-11 18:33:38
There only has to come in one person to the discussion with some weird numbers no one challenges and this article is declared wrong.

...or bump some thread based on a presentation that could not be made here because it is not compliant with our rules by providing some update that is over two years old which doesn't make it any more compliant.

Just because the haystack is large does not mean there's a needle in there.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: 2Bdecided on 2012-03-11 19:42:20
post


For the sake of the sanity and for spared minutes of the lives of everyone reading this thread, and for the love of humanity, do not read the thread corresponding to Monty's article on the SACD threads linked.  Please.

Please.
..and then of course, I had to. You ***!

Cheers,
David.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: xiphmont on 2012-03-11 20:01:18
post


For the sake of the sanity and for spared minutes of the lives of everyone reading this thread, and for the love of humanity, do not read the thread corresponding to Monty's article on the SACD threads linked.  Please.

Please.
..and then of course, I had to. You ***!

Cheers,
David.


As bad as it was, it wasn't as bad as much of the discussion on Slashdot.  The ACs couldn't stop confusing 192kHz and 192kbps.  That was truly disheartening.

It's one thing for someone who doesn't care to not understand.  It's quite another to care deeply, have no functional grasp, and be uninterested in acquiring one.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-11 22:05:58
As bad as it was, it wasn't as bad as much of the discussion on Slashdot.  The ACs couldn't stop confusing 192kHz and 192kbps.  That was truly disheartening.


Doesn't make me facepalm that much. Compared to those who buy the TV preacher arguments for the top 96kHz octave, then mistaking 192 kHz for the fairly sensible 'so they want to sell 192 quality?' seems to me as more of ignorance and less of stupidity.

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-03-12 04:09:12
For the sake of the sanity and for spared minutes of the lives of everyone reading this thread, and for the love of humanity, do not read the thread corresponding to Monty's article on the SACD threads linked.  Please.

Please.



Heh.  And that's even though I screwed up the URLs in that other post.  These should work better:


Thread(s) -- there's TWO of them -- where  David Moran takes on the study 'debunkers':
http://www.sa-cd.net/showthread/42987//y?page=first (http://www.sa-cd.net/showthread/42987//y?page=first)
http://www.sa-cd.net/showthread/58757//y?page=first (http://www.sa-cd.net/showthread/58757//y?page=first)


Brain-destroying thread on Xiphmont's article:
http://www.sa-cd.net/showthread/82449//y?page=first (http://www.sa-cd.net/showthread/82449//y?page=first)


..which I see now includes the classic retort "You are being confused by science,which cannot explain everything we here[sic]."
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: 2Bdecided on 2012-03-12 09:48:50
Thread(s) -- there's TWO of them -- where  David Moran takes on the study 'debunkers':
http://www.sa-cd.net/showthread/42987//y?page=first (http://www.sa-cd.net/showthread/42987//y?page=first)
You've got to feel sorry for the guy or girl (Arnaldo) who puts their fingers in their ears and goes "la la la" to every rational post, repeating the word "debunked" in response to the study to convince themselves that this is the case. Then admit they own over 800 SACDs. With that level of emotional and financial investment, they have to disbelieve M+M.

Though maybe there's no need to feel sorry for them, because M+M clearly say that most SACDs do sound great because they're often used to showcase the best recordings+mastering out there. CDs could sound just as good, but sometimes the same content+mastering isn't available on CD.

Cheers,
David.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: skamp on 2012-03-12 10:28:24
We're talking about a forum that's called "sa-cd.net". A meaningful debate can only take place on neutral ground. And while HA relies on tangible evidence, it is not so neutral either…
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-12 12:18:53
while HA relies on tangible evidence, it is not so neutral either…


.. the meaning of 'while' being 'because'?

[blockquote]If a presidential candidate were to declare that the earth is flat, you would be sure to see a news analysis under the headline ''Shape of the Planet: Both Sides Have a Point.'' After all, the earth isn't perfectly spherical.
– Paul Krugman (http://www.nytimes.com/2000/11/01/opinion/reckonings-bait-and-switch.html)[/blockquote]
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: icstm on 2012-03-12 16:57:47
We're talking about a forum that's called "sa-cd.net". A meaningful debate can only take place on neutral ground. And while HA relies on tangible evidence, it is not so neutral either…
so the debate on their site is clearly meaningless.

But on CA I got confused. Julf plots a graph that happens to have an amplitude resolution (determined by I assume a sufficient number of bits) to show that a shift (faster than NF) in a waveform at a particular sampling rate. However if I had few bits (or if I had a waveform of lower amplitude) then the shifted wave might not be captured with the correct shift.
If this shift is important to what we hear (I am not clear if these differences can be heard) then could there be a benefit from improvements in the redbook?

Miska correctly points out that changing volume changes the frequency whilst that volume change is happening, however their chart should have just shown a sine wave of changing amplitude.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bandpass on 2012-03-13 10:19:15
From the article:
Quote
Modern work flows may involve literally thousands of effects and operations

How should this be interpreted—of the order of 1000 effect invocations × order of 1000 arithmetic operations = order of a million arithmetic operations per sample?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: SebastianG on 2012-03-13 10:49:20
From the article:
Quote
Modern work flows may involve literally thousands of effects and operations

How should this be interpreted—of the order of 1000 effect invocations × order of 1000 arithmetic operations = order of a million arithmetic operations per sample?

I would not worry about the number of arithmetic operations but only about the quantization part where the high precision result of computations is converted again to 24 bit, for example. And for the 24 bit case, there's enough headroom for 1000 effects because the accumulated quantization noise would still be below the noticable threshold. I don't know current practice, but I believe most "serious" processing chains are completely in 32 bit floating point precision or even higher.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bandpass on 2012-03-13 11:24:14
Yes, so while a pure float32 system could have problems, float64 processing with float32 transport would be fine.  But SSE or similar processing presumably moves things back into the danger zone.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: pdq on 2012-03-13 13:07:13
I would hardly call float32 a danger zone. Even when the signal is greater than 0.5 of full scale (which is a small percentage of the time) float32 has a full 24 bits of resolution. When the signal is down at 2^-24 you still have 24 bits, that is 2^-48 of full scale.

When you perform repeated arithmetic operations on the data, if the rounding is uncorrelated (which it should be) then the increase in digitization error only increases as the square root of the number of operations. One thousand operations increases the error by 5 bits.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bandpass on 2012-03-13 14:24:42
Yes, but if "Modern work flows may involve literally thousands of effects" is true, and each effect might have a 1000 operations, then, on a pure float32 system (and presumably those compiled for SSE), that's over 10 bits of noise to be subtracted from 25.

I doubt though that work flows do actually involve thousands of effects, at least not on an individual track within the multi-track.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: SebastianG on 2012-03-13 14:33:31
Yes, but if "Modern work flows may involve literally thousands of effects" is true, and each effect might have a 1000 operations, then, on a pure float32 system (and presumably those compiled for SSE), that's over 10 bits of noise to be subtracted from 25.

How do you arrive at 10 bits instead of 5 bits? Don't you think the "sqrt rule" applies here? if you add N orthogonal noise signals (of equal colour) with an RMS of each X, the expected RMS of the result is X*sqrt(N).
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bandpass on 2012-03-13 15:09:44
Yes, the "sqrt rule" applies:

1000 (but I doubt it's true) effects in serial, each of 1000 (don't know that this is true either, but it's a figure to work with) arithmetic operations, = 1,000,000 arithmetic operations

log2(sqrt(1,000,000)) ~= 10

But as you said above, this doesn't apply if the effects are internally float64 and the transport is float32; in that case, it's only 5 bits of noise.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: icstm on 2012-03-13 15:12:16
I would hardly call float32 a danger zone. Even when the signal is greater than 0.5 of full scale (which is a small percentage of the time) float32 has a full 24 bits of resolution. When the signal is down at 2^-24 you still have 24 bits, that is 2^-48 of full scale.

When you perform repeated arithmetic operations on the data, if the rounding is uncorrelated (which it should be) then the increase in digitization error only increases as the square root of the number of operations. One thousand operations increases the error by 5 bits.
that sounds about right and even then that is worse case, as you could simplfy the operations before performing them in the first place thus limiting the places where rounding occurs.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: icstm on 2012-03-13 15:43:16
Does the statement of the benefits of mixing/mastering >44/16 also extend to recording? Or record at 44/16 and then upconvert for mixing/mastering?

BTW, did we get an asnwer to this?
I agree that mixing needs to be >than 44/16, but you raise an interesting question about recording.

I would like to use image processing as an analogue to this.

RAW does not only offer lossless over JPEG, it also offers the ability to store a wider range of spectrum (Blacker and Black / whiter than white / highlights / shadows / clipping /crushed and all the other terms you see in this space). When doing your mixing (or post processing) then this additional information is beneficial.

I would suggest that the same applied here that to get the most of your mixing, ie your post processing, you would want as much information as possible.

However, the difference here maybe that the dynamic range of a jpeg is not high enough let alone the fact it is lossy.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: mixminus1 on 2012-03-13 16:11:58
Quote
BTW, did we get an asnwer to this?

Yes, it was addressed in the original article, in the "When does 24 bit matter?" section.

tl;dr: Recording/tracking with 24-bit resolution allows you to set your reference level lower, leaving more headroom for unexpected peaks ("whiter than white"), while still retaining a signal to noise ratio greater than that of 16-bit.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: drewfx on 2012-03-13 16:43:16
I would suggest that the same applied here that to get the most of your mixing, ie your post processing, you would want as much information as possible.


No. You would only want as much information as necessary. Any information that, after all mixing/processing/etc., doesn't make it to the (audible portion of the) output only wastes time/resources without adding or improving anything.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: IgorC on 2012-03-14 00:43:15
How about just wide public test for 16/44.1, 22/44.1 /, 16/96, 24/96  or any other combination (with obligatory ABX).
It will be a good lesson for two guys  "I'am audio engineer"  and "I'am audiphhhh what?".
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Wombat on 2012-03-14 00:52:08
Where is Adam Savage & Jamie Hyneman when you need them? Time for another show of Mythbusters!!
We only need some idea how to blow up some audio gear to inspire them
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-03-14 06:42:45
Where is Adam Savage & Jamie Hyneman when you need them? Time for another show of Mythbusters!!


Oh, no, this will only apply to ordinary consumers' hearing. The fact that they cannot pick up what we have heard, proves that audiophiles hear better ...


We only need some idea how to blow up some audio gear to inspire them


Turn up the '40 kHz' button until test subject notices and then watch the tweeter burn? We will have to think up a really creative visual effect to make this look dramatic.


Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: 2Bdecided on 2012-03-14 09:33:53
I would suggest that the same applied here that to get the most of your mixing, ie your post processing, you would want as much information as possible.


No. You would only want as much information as necessary. Any information that, after all mixing/processing/etc., doesn't make it to the (audible portion of the) output only wastes time/resources without adding or improving anything.
What would be a real waste of time / resources would be to try to figure this out for a given mix beforehand

Adding more bits to capture is solving a problem that isn't there. "Adding more bits to processing" would be rather a simplistic thing to support or denounce - it makes a huge difference whether we're talking about the accumulator within an IIR or FIR filter, coefficients, the pipeline between effects, etc etc etc. Huge numbers of bits has been common in some of these for years.

Cheers,
David.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: wakibaki on 2012-03-14 14:07:52
I thought he made good points, but he quoted Meyer and Moran. I just think it's better that we're all informed as to the counter-arguments rather than ending up with egg on our faces.


I'm actually curious as to your specific objection/concern.  I've read the various critiques written by detractors of the BAS tests over the years, but too many of those arguments relied on willful obtuseness and eye rolling. I'd like to hear the methodology/implementation critiques from those who nevertheless agreed with the conclusions.

The point has also been made that [in the article] first I argue "ultrasonics hurt fidelity" and then cite M&M, which supposedly undermines the argument because no one could hear a difference.  In no way does M&M rebut the assertion that ultrasonics _can_ cause audible distortion.  They were using high end setups
designed at expense for audiophile-grade frequency extension, and the results show they obviously weren't affected by audible IMD.  Am I missing something else?


Sorry, I haven't looked at this thread for a while.

The suggestion is not that M&M rebuts the assertion that ultrasonics can hurt fidelity but it demonstrates that ultrasonics did not hurt fidelity.

I don't suggest that the reference to M&M should have been omitted in order not to draw attention to the fact that it demonstrates that ultrasonics did not hurt fidelity, I merely draw attention to the fact that it demonstrates that ultrasonics did not hurt fidelity in the case examined.

The article contends that building to accommodate ultrasonics necessarily sacrifices performance in the audible range. This may be true, but it is not demonstrated that the degradation is audible. Technology, moreover, moves forward apace, so that even if there is audible degradation, this may not always be the case.

All this merely leads to the suggestion that equipment must necessarily be built to a higher standard i.e 'designed at expense for audiophile-grade frequency extension'.

While it is probably possible to establish reasonably accurately at what  point THD becomes audible, it is preferable in some ways to sidestep any argument by exceeding the threshold of audibility by some margin, since, in the case of amplifiers anyway, this is technologically feasible. It may not be desirable to resist too strongly exceeding the threshold of audibility in terms of frequency response where this is feasible without degrading performance to the point where it no longer offers a margin over the threshold of audibility in other areas.

w
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: drewfx on 2012-03-14 15:41:05
I would suggest that the same applied here that to get the most of your mixing, ie your post processing, you would want as much information as possible.


No. You would only want as much information as necessary. Any information that, after all mixing/processing/etc., doesn't make it to the (audible portion of the) output only wastes time/resources without adding or improving anything.
What would be a real waste of time / resources would be to try to figure this out for a given mix beforehand

Adding more bits to capture is solving a problem that isn't there. "Adding more bits to processing" would be rather a simplistic thing to support or denounce - it makes a huge difference whether we're talking about the accumulator within an IIR or FIR filter, coefficients, the pipeline between effects, etc etc etc. Huge numbers of bits has been common in some of these for years.

Cheers,
David.


Yes. Higher bit depths (and floating point) and upsampling/downsampling is common during processing and is almost always used today where useful or necessary.

I didn't quote icstm's entire post, but my impression was that he may have been talking not just about during processing, but for recording as well. So my response was more to that (perhaps incorrect) interpretation.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-03-14 15:44:58
There is no point in distributing audio to consumers in a 24bit/192kHz format.  The only possible convenience I see is that (AIUI) modern AVRs commonly convert incoming audio signals to 24 (32?) bits before applying DSP.  Some also upconvert
sample rate to 96kHz.  If the AVR does those functions poorly, providing the audio already 'upconverted' would be a way to avoid degradation.  But I have no evidence that AVRs are doing it poorly. And too not everyone wants to use DSP when they play music.  Often the upconversion can be turned off by setting the AVR to a 'Pure' mode.

However, there is a fascinating discussion going on on one of the pro audio lists that reminded me of one possible legitimate use of really high SR ADC for digital *capture* of taped (analog) audio:  use of tape bias tones to
correct wow and flutter as is done by Plangent Processes (http://www.plangentprocesses.com/) .  Such tones can be well into the hundreds of kHz depending on the tape machine originally used to make the
recording.    So some of these tones are well beyond the capabilities of even 192kHz SR to capture.

It turns out, though, that what PP actually do is use a circuit to 'downshift' the ultrahigh frequency bias tones into a range that can be captured by common sample rates.  It's described in a scant detail here (http://www.plangentprocesses.com/aes117.pdf) )
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: icstm on 2012-03-15 15:34:34
Yes. Higher bit depths (and floating point) and upsampling/downsampling is common during processing and is almost always used today where useful or necessary.

I didn't quote icstm's entire post, but my impression was that he may have been talking not just about during processing, but for recording as well. So my response was more to that (perhaps incorrect) interpretation.
I completely agree about the processing.
What I was saying about the recording (and this may not be true) is that there could be cases where you want to shift the information that is in above audible range down or there could be cases where you wish to expand the difference between 2 sounds.

The example I giving was in image processing where you are trying to change how the highlights are shown. HDR photography I would have thought is analogous to this?

Also, though completely unrelated, if much of the sound energy of keys jangling is above the range of hearing, why would not try to map this down to better capture this lost power?

If you are going to process above 16/44, even if you are going to playback at 16/44 I would have thought there are cases where recording at 16/44 may not be enough if you are going to PP?
(or is that rubbish?)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: drewfx on 2012-03-15 16:35:27
I'm unaware of any audio processing where inaudible frequencies are used for enhancing audible ones somehow in general audio production/reproduction. But perhaps someone else knows of something? If there was something, then a higher sampling rate might indeed make some sense.

Now if you wanted to, say, pitch shift ultrasonics down by several octaves, or intentionally create audible inter-modulation distortion from inaudible frequencies for some reason, it would indeed make sense to record the higher frequencies. But I'd say those sorts of things would be unusual exceptions that requires special procedures, not a general case.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-03-15 16:48:41
It should be pretty clear from the title of the discussion that recording and processing is not on-topic.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: 2Bdecided on 2012-03-15 16:58:55
So you're substantially claiming that non-ABXable lossy-encoded sound is equal to a lossless one ?
If no one can ABX two sounds under any conditions, then they're perceptually equivalent.

Quote
If true, than what's the meaning of lossless encoders ?

1) lossless existed before lossy.
2) where is this 100% unABX-able lossy encoder?
3) where is this store that guarantees to use a particular lossy encoder that I have decided I'm happy with?
4) lossless is a perfect source for re-mixing, lossy encoding, broadcasting, etc etc - any number of things I might want to do myself.


Quote
BTW - once again - my position is that we don't need more Hz, but we need more bits !
You don't need more bits. (http://www.hydrogenaudio.org/forums/index.php?s=&showtopic=93753&view=findpost&p=788658) Which is also explained perfectly in the article this thread is about.

DSD seems to work quite well - 2.8224 MHz / 1 bit. With just 3 or 4 bits it could be essentially perfect (120dB+ SNR, 50kHz+ bandwidth, zero distortion). Aren't dither and noise shaping amazing?

Cheers,
David.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: sld on 2012-03-15 18:09:15
I don't think it matters—when you're sitting  in front of the speakers, doing an ABX test, you can use any organ of your body you like to help make the determination (you might want to lock the door first though).

Well, in this perspective lossless is useless if you can't ABX lossy... 

No, ABX isn't the unique method to evaluate sound quality, IMHO.

BTW just think about low frequency effect on your floor (then to your feets): even if your ears hear certain frequencies, the listening experience wouldn't be the same if the floor would not vibrate. 

ABX the vibrations?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-03-15 18:14:46
It's really a moot point since 16/44.1 can reproduce the same vibrations.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-03-15 23:39:33
So you're substantially claiming that non-ABXable lossy-encoded sound is equal to a lossless one ?


Try running the concept of 'non-ABX-able lossy' past some codec tweakers.  That is, people who have trained themselves to be very sensitive to lossy artifacts , so they can improve the codecs.

Even 320kbps LAME is ABX-able, as evidenced by reports here on HA (look up posts by user \mnt for example).  It's just that such listeners are rare.  For most people who have reported trying, high bitrates using a decent codec produce lossy versions that are *effectively* indistinguishable from source by ABX.  (The source counts too ...considering 'killer' samples and all that.)


Quote
BTW - once again - my position is that we don't need more Hz, but we need more bits !


We don't.  More bits are useful during recording and production, far less so in playback at home.

If you've  been reading HA regularly since 2001, you should know all this already.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: icstm on 2012-03-16 10:22:33
It should be pretty clear from the title of the discussion that recording and processing is not on-topic.

which is why my first post to this thread was that the orginial article linked from the OP finally answers my post here on HA where I was asking about the playback format!
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-10-11 20:53:49
xiphmont: evolver's reposting of your article this week

http://evolver.fm/2012/10/04/guest-opinion...-make-no-sense/ (http://evolver.fm/2012/10/04/guest-opinion-why-24192-music-downloads-make-no-sense/)

lead me down the rabbit hole to a post of yours on slashdot from earlier this year

http://slashdot.org/comments.pl?sid=2857759&cid=40038991 (http://slashdot.org/comments.pl?sid=2857759&cid=40038991)

where you take some shots at the AES, e.g.

Quote
It's not an attack, it's more a statement of truth. The AES publishes all sorts of things. Papers with interesting ideas and no data (eg, the J. Dunn 'equiripple filters cause preecho' paper, which presents a fascinating insight, even if it doesn't work out in practice), papers with data that are effectively WTFLOL (the famous Oohashi MRI paper) and papers that are more careful controlled studies. It runs the whole gamut on both sides, just as I said.


Just want to point out that there's a sustantial diff in terms of peer-review between AES convention presenations/publications, and JAES publications.  Oohashi et al. never made it past convention, as far as I can tell.  Their work ended up in a low-impact neurophysiology journal.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bandpass on 2012-10-12 05:51:41
Quote
Papers with interesting ideas and no data (eg, the J. Dunn 'equiripple filters cause preecho' paper, which presents a fascinating insight, even if it doesn't work out in practice)

Dunn refers to R. Lagadec and T. G. Stockham, ‘Dispersive Models for A-to-D and D-to-A Conversion Systems’ for data.  Is there data elsewhere to support that it doesn't work out in practice?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Patrunjica on 2012-10-13 23:51:10
Fascinating read, there has been one sticking point for me though that relates to the Nyquist frequency and I'd appreciate if anyone can offer an answer to my question.

Is a sampling rate of 44.1kHz sufficient to accurately reproduce all waves under 20kHz? I ask because I have been doing some tests using wave generators operating under various sampling rates and I stumbled upon something that confused me greatly - since my understanding of the Nyquist theorem is very limited.

The following is a 13001Hz sine wave generated using SineGen 2.5 using three different sampling rates and then imported in Reaper

(http://oi47.tinypic.com/2cmokeq.jpg)

Why doesn't the first sine wave resemble a sine wave anymore and does this mean anything as to the resolution necessary to fully and accurately reproduce one in the first place?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-10-13 23:59:45
Your wave editor is connecting the samples with straight lines rather than with sinc pulses.  This is not even remotely how it is supposed to be done either in theory or in practice.

To answer your question, a 44.1kHz is perfectly adequate to capture any frequency below 22.05kHz.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: saratoga on 2012-10-14 00:01:45
Is a sampling rate of 44.1kHz sufficient to accurately reproduce all waves under 20kHz?


Yes, and more actually.

Why doesn't the first sine wave resemble a sine wave anymore and does this mean anything as to the resolution necessary to fully and accurately reproduce one in the first place?


Your software isn't actually trying to draw the waveform that those PCM samples would generate.  Thats just a linear interpolation of those points, not the PCM waveform.  Just ignore it.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Patrunjica on 2012-10-14 00:13:57
Your software isn't actually trying to draw the waveform that those PCM samples would generate.  Thats just a linear interpolation of those points, not the PCM waveform.

Figured as much, but I can't quite wrap my head as to how exactly a constant PCM sine wave can be generated from what looks to be like (and by all accounts should be) chaotic, non-repeating data.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Wombat on 2012-10-14 00:26:45
Why doesn't the first sine wave resemble a sine wave anymore and does this mean anything as to the resolution necessary to fully and accurately reproduce one in the first place?


http://www.hydrogenaudio.org/forums/index....nction+audacity (http://www.hydrogenaudio.org/forums/index.php?showtopic=93496&hl=sinc+function+audacity)

Edit: i answered while Patrunjica had a similar picure in and a relating question as in the thread i linked to. His post was edited while i answered.

Edit2: If the above sentense makes no sense you are absolutely right! Me didn´t scroll up to see the pic and didn´t realize the answeres inbetween, sorry. Nonetheless the thread i linked to should help you Patrunjica
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: drewfx on 2012-10-14 00:28:38
Figured as much, but I can't quite wrap my head as to how exactly a constant PCM sine wave can be generated from what looks to be like (and by all accounts should be) chaotic, non-repeating data.


What makes you think it should be chaotic and non-repeating?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: saratoga on 2012-10-14 00:46:36
from what looks to be like (and by all accounts should be) chaotic, non-repeating data


Look again.  Your points are not chaotic and actually do repeat, even with linear interpolation.  You posted several periods.  Now fit a sensible function between those points instead of a straight line and it'll repeat like the original sin wave. 

If you want to know which function you need to use, look up how PCM works.  Or just assume that since PCM very clearly does work, the needed function does in fact exist and whoever made your sound card and stereo implemented it. 
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Patrunjica on 2012-10-14 01:13:21
True, but each time they repeat they are different then before, dividing up the sampling rate to that particular frequency results in a number with an infinite number of decimal places, approximating the location of one of the sampling points will influence the position of the next sample point and so on and so forth so that each time the wave repeats it shifts out of phase relative the sample points. And since the respective frequency is closer to the Nyquist any such deviation is more noticeable compared to a deviation that might occur using a higher sampling rate.

I know the waveform can be rebuild since obviously I can hear it, but I can't help but doubt that it's a more of an imperfect reconstruction then compared to pretty much anything lower then a 11025Hz sine.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-10-14 01:24:30
I know the waveform can be rebuild since obviously I can hear it, but I can't help but doubt that it's a more of an imperfect reconstruction then compared to pretty much anything lower then a 11025Hz sine.

You've placed your doubt in the wrong thing.  Why you're bothering to continue down the wrong path instead of acknowledging what has been stated to you very plainly (that your wave editor is not connecting the samples together properly) is a mystery to me.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: splice on 2012-10-14 01:39:13
... I know the waveform can be rebuild since obviously I can hear it, but I can't help but doubt that it's a more of an imperfect reconstruction then compared to pretty much anything lower then a 11025Hz sine.


Play the waveform back through a digital to analog converter. Look at the resulting waveform with an oscilloscope. It's a smooth sine wave. Magic.

The magic trick is that the D to A converter doesn't just "join the dots". It passes its output through a reconstruction filter that only passes frequencies below half the sampling rate. Say you digitise a 20 KHz signal at a 44.1 KHz sampling rate.  That 20 KHz signal can only ever be a sine wave. For it to be any other shape it would have to contain harmonics, and those harmonics would start at 40 KHz. The harmonics would be filtered out at the input to the A to D converter.

When you think about it, any digitised signal above 11.025 KHz must be a sine wave. It may itself be a harmonic of a lower frequency signal, but it won't in turn have any higher harmonics because they would be above 22.05 KHz.

So if you take your "jagged" join-the-dots line and join the dots with a sine wave curve, you'll find that there's exactly one curve that will join all the dots: one with the same frequency as the original frequency. Drawing that curve is exactly what the D to A reconstruction filter does.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Patrunjica on 2012-10-14 01:55:07
So if you take your "jagged" join-the-dots line and join the dots with a sine wave curve, you'll find that there's exactly one curve that will join all the dots: one with the same frequency as the original frequency. Drawing that curve is exactly what the D to A reconstruction filter does.


No disagreements there, I was just wondering how it can perfectly reproduce the wave when the sample points have to be approximated, the ideal position for them lies at the end of an infinite series of decimals so naturally those get truncated at one point or another and what I'm wondering is if this truncation implies any errors in the reproduction concerning amplitude or the resulting frequency itself.

The deviations can't be that large since obviously PCM works but I'm wondering how large can they be for those frequencies that creep up closer to the given sampling rate?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-10-14 01:59:21
If you were to look at the waveform in an editor that actually assumes proper reconstruction you will in fact see a proper sine wave regardless of the frequency.  This is all about the specific application you used to display the waveform.

Your application is displaying it incorrectly!
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Wombat on 2012-10-14 02:06:04
The deviations can't be that large since obviously PCM works but I'm wondering how large can they be for those frequencies that creep up closer to the given sampling rate?

And another link:

http://www.hydrogenaudio.org/forums/index.... (http://www.hydrogenaudio.org/forums/index.php?showtopic=91126&st=0&p=771163&)

Now read the post especialy from Woodinville. I don´t understand all this but since you asked... Btw. when you use search in here you´ll find several similar threads dealing with such things.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: mjb2006 on 2012-10-14 02:15:55
(http://skew.org/tmp/hydrogenaudio/13001_comparison.png)
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-10-14 02:29:14
Thanks for that!
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: splice on 2012-10-14 02:34:02
No disagreements there, I was just wondering how it can perfectly reproduce the wave when the sample points have to be approximated, the ideal position for them lies at the end of an infinite series of decimals so naturally those get truncated at one point or another and what I'm wondering is if this truncation implies any errors in the reproduction concerning amplitude or the resulting frequency itself.


The displayed sample points only appear to be chaotic. They are actually accurately positioned to the limits of the numeric resolution. There is one sine curve that exactly (to the limits of the resolution) passes through all of the sampling points. Any inaccuracy introduced by "truncation of an infinite series" appears as noise in the output from the D to A conversion. This noise will be equal to or less than that represented by the least significant bit of the digital signal. 

When you digitise a signal to 16 bits, each sample has to be represented by one of 65535 different numbers. If it falls part way between two possible values, it is set to the nearest. This means the digitised value can be up to half a "bit's worth" different than the original input. When you reconvert it back to analog, the output can be up to half a bit plus or miinus different than the original. This is noise, because it is different than the original signal.

For a 16 bit signal, this noise is insignificant. Try it for yourself. Use Reaper or Audacity to generate a 1 KHz signal using all 16 bits (0dBfs). Do it again using only the least significant bit (-96 dBfs). Play the 16 bit signal at the loudest level you can stand. Without changing the volume setting, play the 1 bit signal. Can you hear it?  (I doubt it).

Once you understand this, we'll move on to dither.

Edit: mjb2006's picture would have saved me a thousand words if I'd seen it in time...
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: saratoga on 2012-10-14 03:59:59
True, but each time they repeat they are different then before,


I didn't realize you had picked such an unround number as 13001, so the periods you showed don't actually repeat exactly.  But the function is still periodic, just with a much longer period.  They have to repeat every LCM(44100,13001) samples given that the underlying sin function is itself periodic. 

dividing up the sampling rate to that particular frequency results in a number with an infinite number of decimal places, approximating the location of one of the sampling points will influence the position of the next sample point and so on


This isn't true because subsequent values of sin(x) do not depend on previous values.  How could they?  sin(x) has an infinite domain, so if it could only be computed recursively it would be impossible to compute it at all.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Kees de Visser on 2012-10-14 11:50:51
Your application is displaying it incorrectly!
No need to shout
Why not propose him an application that does it right (or at least better) ? My favorite (and cross-platform) is iZotope RX, which has a free demo version available here (http://www.izotope.com/products/audio/rx/download.asp).
You can vary the "Waveform interpolation order" in the Preferences/Display between 0 and 64.
Nothing beats empirical evidence IMO.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: greynol on 2012-10-14 12:13:40
Probably because I am WindowsAdobe Audition-centric and don't know of other programs.  I'm shouting because I provided an immediate (and correct!) explanation to the dilemma which was ignored and now the discussion has needlessly gone off-topic.  I'm tempted to split it off now.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-10-14 23:41:13
No disagreements there, I was just wondering how it can perfectly reproduce the wave when the sample points have to be approximated


By 'knowing', in the appropriate sense, that it is a sine wave. In principle, not very far from the way that I can perfectly reproduce a line using merely two distinct points: by knowing that it is a line.

Not that it matters. A 20 kHz sine wave and a 20 kHz triangle-shaped wave (or whatever shape!) sound the same, as the difference is above the hearing limit. Only the sine wave part of the triangle will contribute to the ?20 kHz range.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: yourlord on 2012-10-15 01:48:19
No disagreements there, I was just wondering how it can perfectly reproduce the wave when the sample points have to be approximated


Look up the Nyquist–Shannon sampling theorem. Because it's known that the highest possible represented frequency is X, there is only 1 possible frequency below X that can match any 2 sample values. Based on that fact the DAC will "connect" the 2 sample points with a voltage curve that matches that single possilbe frequency below X...

The fact that the samples can only be quantized to 65536 levels introduces what's known as quantization noise. At 16 bit resolution you can basically ignore it for audio delivery.

Not that it matters. A 20 kHz sine wave and a 20 kHz triangle-shaped wave (or whatever shape!) sound the same, as the difference is above the hearing limit. Only the sine wave part of the triangle will contribute to the ?20 kHz range.


It should be stated that a 20kHz triangle wave can't be represented with 44.1kHz sample rate. It would be low pass filtered to a sine wave before being sampled.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2012-10-15 07:37:08
there is only 1 possible frequency below X that can match any 2 sample values. Based on that fact the DAC will "connect" the 2 sample points with a voltage curve that matches that single possilbe frequency below X...


That is wrong. The number of samples required, approaches infinity as frequence approaches sampling frequency / 2. For any frequency below, you can do with a finite (but possibly large) number of samples.

Simplify down to 3 bits, i.e. values 0 through 7. Pick two successive samples, say, a 4 and a 3. Matched by a 7654321012345676543210123456 sequence, a 6564321234565432345 sequence, a 543234543234543234 sequence and a 4343434343434343434343.  What's the lowest frequency in each of those signals?
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Arnold B. Krueger on 2012-10-15 13:32:41
I know the waveform can be rebuild since obviously I can hear it, but I can't help but doubt that it's a more of an imperfect reconstruction then compared to pretty much anything lower then a 11025Hz sine.


I'll put your mind at rest. A digital reconstruction of any digitized or synthesized analog signal is always imperfect for theoretical and/or practical reasons.

I think you're asking the wrong question. The most important question is not whether the reconstruction is imperfect, but rather how those imperfections compare to what you get when you do the same thing by other means, for example by keeping the signal in the analog domain.

When it comes to storing music, digital from day one of consumer digital audio back in 1982 has always been vastly more perfect than analog media.

When it comes to playing music, the digital outputs of music players are always vastly more perfect than their analog outputs, at any price from high to low.

When it comes to passing music through even a single audio component such as an AVR the same thing is true.  Digital (DSP-based) AVRs have always been vastly more perfect than analog AVRs at any price from high to low.

When it comes to loudspeakers, if we had implementations of speakers that were as sophisticated as our AVRs, the speakers that were more fully implemented in the digital domain would have a greater potential to be accurate.  That day is coming!

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: krabapple on 2012-10-15 21:17:24
I know the waveform can be rebuild since obviously I can hear it, but I can't help but doubt that it's a more of an imperfect reconstruction then compared to pretty much anything lower then a 11025Hz sine.


I'll put your mind at rest. A digital reconstruction of any digitized or synthesized analog signal is always imperfect for theoretical and/or practical reasons.


I'd note here that an analog recording of an acoustic event -- say, a sound recorded on magnetic tape  -- is always imperfect.  And an analog reformatting of a recorded signal (e.g., an LP pressing sourced from a tape, or from digital ) -- is also always imperfect.  Look at how the imperfections pile up!   

Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: xiphmont on 2013-07-18 18:40:30
Just want to point out that there's a sustantial diff in terms of peer-review between AES convention presenations/publications, and JAES publications.  Oohashi et al. never made it past convention, as far as I can tell.  Their work ended up in a low-impact neurophysiology journal.


Ah, I suppose that would be an important distinction, and one I wasn't aware of.  I've been an AES member on and off over the past 15 years, but mostly for the journal access.  I find it hard enough to make the local BAS meetings :-)

My point was that the publication (presentation) of the paper did not automatically elevate it beyond a status of 'let's discuss', not that the AES should be faulted for presenting it.
Title: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: xiphmont on 2013-07-18 18:50:04
Quote
Papers with interesting ideas and no data (eg, the J. Dunn 'equiripple filters cause preecho' paper, which presents a fascinating insight, even if it doesn't work out in practice)

Dunn refers to R. Lagadec and T. G. Stockham, ‘Dispersive Models for A-to-D and D-to-A Conversion Systems’ for data.  Is there data elsewhere to support that it doesn't work out in practice?


Dunn says something a little different than Lagadec, although Lagadec was certainly hinting hard at the direction Dunn went. 

In any case, Dunn presents simulations of an approximation that's a partial fit to the frequency responses of several DAC reconstruction filters in the field at the time.  The partial-ness of the fit is not accounted for in the simulated figures, nor does Dunn show any measurements of the actual filters to show if and how well they match the approximation.  Thus we don't know how good or bad the approximation is in practice.

But my primary thought is that it's a filter strategy and family that aren't used in DACs anymore.  This work is from the bad old days of high-order analog equiripple IIR anti-imaging filters, which at the time were dominant.  This is not a fault of the paper in any way, but we do not use these filters today and so it would seem the concerns simply do not apply anymore.

Of course, we also have test equipment today that should make verifying that assumption pretty easy.  I've been wondering when I'd have a lazy weekend afternoon with nothing better to do than go explore the Dunn paper experimentally.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: muaddib on 2021-10-19 13:30:22
http://people.xiph.org/~xiphmont/demo/neil-young.html (http://people.xiph.org/~xiphmont/demo/neil-young.html)

The link is down. Also there is nothing on the Wayback Machine.
Does somebody still have a backup of the page?
Any ideas why it was removed?
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Rollin on 2021-10-19 15:20:06
there is nothing on the Wayback Machine.
It is there. https://web.archive.org/web/20170602140930/http://people.xiph.org/~xiphmont/demo/neil-young.html
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: muaddib on 2021-10-19 15:45:39
It is there. https://web.archive.org/web/20170602140930/http://people.xiph.org/~xiphmont/demo/neil-young.html

For some reason I couldn't access it this morning, but now also works for me.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Makaki on 2022-01-06 07:04:43
There was quite a bit of content within the "people.xiph.org/~xiphmont" web, and it seems it's now all gone. IDK If it's intentional, but it seems weird.

There were tech demos related to Opus, Vorbis, and AFAIK, some video formats too.

IDK for how long the archive on wayback machine is held for, but I think it would be worth restoring or setting a proper mirror of the content that used to be there.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-06 10:37:41
768kHz now ;D
https://www.soundliaison.com/index.php/studio-masters/856-ray-carmen-gomes-inc
Some DACs accept 32-bit 1536kHz and DSD1024, so it is still not the end.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Chibisteven on 2022-01-06 11:39:31
768kHz now ;D
https://www.soundliaison.com/index.php/studio-masters/856-ray-carmen-gomes-inc
Some DACs accept 32-bit 1536kHz and DSD1024, so it is still not the end.

The size of those files are just awful.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: binaryhermit on 2022-01-06 14:44:19
Because everybody needs to be able to hear 700kHz sounds... /s
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: guruboolez on 2022-01-06 15:48:25
768kHz now ;D
https://www.soundliaison.com/index.php/studio-masters/856-ray-carmen-gomes-inc

I was curious about recording sampling rate. I tried to find this info on the booklet, which is freely available on nativedsd:
https://www.nativedsd.com/product/sl1052a-ray/

Quote
Recording, mixing and mastering by Frans de Rond. Produced by Peter Bjørnild. Music arranged by Peter Bjørnild with lots of help from Carmen, Folker and Bert. Recorded at MCO, Studio 2, Hilversum, The Netherlands, on the 12th of May and the
24th of August 2019. Total time: 46:17
Catalog Number: SL-1052A
Original recording format DXD 352,8 kHz
The original recording is analog mixed and mastered to tape using a Studer A80. All other formats are converted versions of the original.

So if I understand correctly:
— sound is digitally recorded at DXD "format" (PCM-352,8 KHz) [A-D encoding]
— mixing and mastering is then recorded on analog tape [D-A]
— then it's converted again in the digital domain… at twice the original sampling rate  :o
The whole processing looks curious to me.


For the sake of curitosity I bought a 32 bit DXD triple album:
https://trptk.com/catalogue/miscellanea/
The downloaded tracks were in .wav format and zip compressed: easy to use but not very efficient.
Uncompressed wav = 29,3 GB and 22579 kbps ; zip ≈ 26 GB ; WavPack = 18,3 GB and 14038 kbps “only”.
A multichannel version is also available on their catalogue and size must therefore reach 80 GB!

I converted it to lossy for my smartphone and size shrinked to less than 200 MB. Bitrate is now much more prosaic but sound quality is as enjoying ;)
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-06 16:14:29
Perhaps to capture the so-called "tape sound" from some vintage Studer machines. And yes, flac can no longer handle these sample rates, and no usable 32-bit encoder. WavPack has a chance to dominate the Hi-Res market now 8)
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-06 22:11:24
That 768 Ray Charles track is available for free ... if a GB download is free where you are.  So I picked it up for curiosity. It WavPacks to around 70 percent, which is on par with a downsample to 48/24.

This stupid file size takes some CPU to process. Monkey's Insane not making it to 2x realtime decoding - how's that for a trip down memory lane, folks? ;-)
Actually, this track fools Monkey's Insane into making 1.7 percent larger files than Monkey's Extra High. Size order Extra high < High < Normal < Insane < Fast. (Since I was at it: Running WavPack at -hx4 brought the file sizes down to below Monkey's Extra high. Half realtime encoding, saves power if you want to listen to it a few times. OptimFrog wins on size, even at default setting.)


yes, flac can no longer handle these sample rates, and no usable 32-bit encoder. WavPack has a chance to dominate the Hi-Res market now 8)
OptimFrog needs that all-important DSD mode to compete!

(Say what you will about DSD; some of the SACD masters are different than their CDDA counterparts - likely in order to fool users into thinking that better sound was due to format. And, I like WavPack.)
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-06 22:13:18
So if I understand correctly:
— sound is digitally recorded at DXD "format" (PCM-352,8 KHz) [A-D encoding]
— mixing and mastering is then recorded on analog tape [D-A]
— then it's converted again in the digital domain… at twice the original sampling rate  :o
My pet bat says the Studer's high range is "Not bad for a human", and then does her best Alien screech impersonation. Just kidding

Quote
For the sake of curitosity I bought a 32 bit DXD triple album:
Did they use float or integer?
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: guruboolez on 2022-01-07 06:46:44
32 bit floating for this recording.
Thanks for the lossless comparison !
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-07 09:46:47
32 bit floating for this recording.
Float protects against clipping, so it makes sense to use in processing. (Cf. the topic; Monty took care to make the point that the considerations applied to formats to deliver to end-users.)

Thanks for the lossless comparison !
Since you mentioned .zip; the 768 Gomes track was delivered as .wav, but:
Windows send to zip: 94 percent (while NTFS compression saved 38 ppm ...)
7z ultra: 80 percent
Audio compressors: 70 percent

Other vendors would also deliver single files as they are, but start zipping when there are more than one; so not primarily for compression, but for delivering a "folder". Using anything but the lowest common denominator compression algorithm will get customer support overworked I guess.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-07 10:32:41
Monkey's Insane not making it to 2x realtime decoding - how's that for a trip down memory lane, folks? ;-)
Reminds me that FhG WinPlay3 has an option to instruct mp3 to decode for 80486 or Pentium class processors.

If this trend continues .wav will no longer be usable too. While it is possible to have 64-bit float and multi MHz .wav files, the legitimate file size is 4GB. Individual software can have their hacked version of .wav but it will cause compatibility issues with other software. There are uncompressed formats like w64 and caf but I suppose these formats are not usually supported in standalone players and streamers.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: kode54 on 2022-01-07 10:54:13
That godawful site disables the in-page text search feature.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-07 11:00:06
That godawful site disables the in-page text search feature.
Firefox user here. Ctrl-F doesn't work but Edit > Find in Page works.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: ktf on 2022-01-08 17:59:54
And yes, flac can no longer handle these sample rates
Well, actually, the FLAC format goes up to 2^20 (1'048'576Hz), but it is not subset and currently not supported by the reference encoder. Adding it is trivial however, and there is a pull request waiting for merge: https://github.com/xiph/flac/pull/219
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-08 19:34:36
So there is no point to distribute music in 24 bit/192 kHz because flac supports up to 1048576Hz :D
The FAQ says floats will not be supported, but would there be 32-bit integer support?
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-08 22:09:06
The FAQ says floats will not be supported, but would there be 32-bit integer support?
I would imagine that supporting the less common (for good reason) of the 32-bit formats, will increase the noise to signal ratio in the bug report system.
But that hypothesis has been put to the test? ALAC supports 32-bit integer but not float (well at least that goes for refalac).

Then on the other hand, if the reference encoder wants to stick to "--lax" being required for non-subset (... and, *checks notes*, not reassign the "reserved" 011 to 32) then one doesn't don't need to tout a new 32-bit support as loud as the warnings of "non-subset, may not play, --lax required").

Edit: some SVN version of flake could do 32 bits (https://hydrogenaud.io/index.php?topic=83520.msg721992#msg721992), without that leading to hordes of desperate users asking HA to decode their .flac files  ;)
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Rollin on 2022-01-09 00:08:01
some SVN version of flake could do 32 bits (https://hydrogenaud.io/index.php?topic=83520.msg721992#msg721992), without that leading to hordes of desperate users asking HA to decode their .flac files  ;)
Here is .exe in attachment for curious ones. fb2k (as for all 32 bit integer, it will be converted to 32 bit float) and ffmpeg can decode 32 bit integer FLAC created by it if stereo decorrelation is disabled (option -s 0).
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: ktf on 2022-01-09 15:27:52
ffmpeg can decode 32 bit integer FLAC created by it if stereo decorrelation is disabled (option -s 0).

It probably causes signed integer overflow in the decoder though. This might work, but is undefined behaviour, so it might also stop working at some point.

I didn't want to say this right away (I too think 32-bit int audio is completely nonsensical) but as it is being discussed now anyway....

Here's a very recent (i.e. sent in yesterday) patch for ffmpeg (https://patchwork.ffmpeg.org/project/ffmpeg/patch/20220108142437.756529-1-mvanb1@gmail.com/) to create such 32 bit files backwards compatible with libFLAC (not the flac command line utility though) from 1.2.1 onwards and ffmpeg from May 2015 onwards. It is specifically crafted to not cause overflow issues. If it cannot find a work-around for a certain subframe, it falls back to using a verbatim (i.e. uncompressed) subframe.

I haven't checked, but I don't think flake did the same. This seems like a safer way.

Still, I like to stress this again, I think it is dumb to use 32-bit int for audio, but if it happens, I prefer a backwards-compatible approach.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: kode54 on 2022-01-10 02:04:57
I'm surprised that stereo decorrelation can't just take advantage of largely different values having a small difference if you allow numbers to wrap within the target bit size. But then you'd need special versions of the math functions that are either supposed to saturate or wrap, depending on the purpose.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: ktf on 2022-01-10 06:35:38
But then you'd need special versions of the math functions that are either supposed to saturate or wrap
Yes, and as those functions aren't available in hardware on many platforms (especially embedded platforms in for example a receiver or a portable audio player) these functions would have to be implemented in software. I think you'd be looking at a 5x slower decoder in that case, for very, very little gain.

.... but this has nothing to do with why this 32-bit ffmpeg patch or the flake binary can't use stereo decorrelation. That is simply because libFLAC rejects 33-bit subframes outright, and because creating a work-around on frame level instead of at subframe level is a lot more complicated.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-17 13:24:42
So it is better to assume flac won't support any 32-bit format. In fact, both 32-bit formats are useless as distribution formats. 32-bit DAC/ADC use integer math. It makes perfect sense because they are used to convert to and from analog so 1500dB of dynamic range is useless. DSP involved in these devices like interpolation, decimation filters and modulators are more about precision within a reasonable range instead of 1500dB. If I were to propose a 32-bit float format for these kinds of operations, I would rather stripping 3 exponent bits and use them as mantissa.

On the other hand, some field recorders use floating point recording format as a form of marketing, even provide sample files to impress consumers:
https://www.sounddevices.com/sample-32-bit-float-and-24-bit-fixed-wav-files/

Basically, these things record two copies of the same audio at the same time, but with different analog input levels, then route the ADC output to a floating point DSP. When one of the ADC clips, it seamlessly corssfade to another ADC. However it is common sense that the preceding physical and analog chain (mic and preamp) can still clip, faulty DSP logic (instead of "not enough" bits) can also create glitches when combining different ADCs. Here is a review of a Zoom floating point recorder:
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/

Here are some RMAA results of several converters:
https://www.audiosciencereview.com/forum/index.php?threads/rmaa-tests-welcome-to-add-others.16332/post-532248

In fact, high quality 24-bit traditional converters have far better results. Also, even if floating point math is involved in combining different ADCs, it stills makes much more sense for the recorder to scale the resulting waveform to normal range and save as 24-bit or below. Just like the old Pro Tools TDM is externally 24-bit, but internally 56-bit fixed point (48-bit processing with 8 headroom bits), and the recent versions are 32/64-bit float, as well as some other DAWs like Reaper and others.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-17 14:22:12
I suspect that if one forty years ago had discussed a floating-point format for the single purpose of audio, we might have had a 12+4 or something? Or, any votes for 13+3?

If I were to propose a 32-bit float format for these kinds of operations
So one settled for an already-established general purpose format, with some limitations: yes you can get the pain threshold and either a mosquito or a nuclear explosion at the same time, and even all three in the same file - but for the seconds you nuke the Earth, the mosquito will be lost.
As audio might very well live happily with a volume control that cannot compress air into metal, you might argue that it wasn't an even trade-off between mantissa and exponent, but ... ...

... but unless a good nuking doesn't makes you desperate for that mosquito, then I'd say that it is just fine to use an already established general purpose 32-bit float format, a standard which has been hardware-supported since long before before the famous Pentium bug. At least when it losslessly contains your 24-bit signal.

On the other hand, some field recorders use floating point recording format as a form of marketing, even provide sample files to impress consumers:
https://www.sounddevices.com/sample-32-bit-float-and-24-bit-fixed-wav-files/
Doesn't that describe the reason for float? If you think your recording might be 30 dB low, then boos it 30 dB without ever having to worry about clipping.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-17 14:54:00
I suspect that if one forty years ago had discussed a floating-point format for the single purpose of audio
For example, there are also bfloat16 which is optimized for AI, and half-precision float for GPUs: RGBA for colors and XYZW for coordinates, 16 bits for each component. Many older GPUs for examples only support 32/64 bit float and the additions of 16-bit floats save a lot of memory. bfloat and half-float are both 16 bits but blfoat has more exponent bits while half-float has more mantissa bits.
https://en.wikipedia.org/wiki/Bfloat16_floating-point_format
Quote
Doesn't that describe the reason for float? If you think your recording might be 30 dB low, then boos it 30 dB without ever having to worry about clipping.
30dB is quite different from 1500dB and can be perfectly handled with integer math. For example, 32-bit integer with 5 bits of headroom and 27 bits of integer are still plenty. The headroom doesn't need to be exposed to end users: 0dBFS to the eyes of users but adds several bits on top of that during DSP. Actually, many products use this approach, including RME:
https://www.google.com/search?q=rme+totalmix+42-bit
Thousands of dB make more sense for mixing hundreds of tracks or software synth/samplers with a lot of complex patches and voices (polyphony), on top of other insert and auxiliary effects. The RMEs have FPGA mixers for basic signal routing task with low latency, but never as complex as a complete DAW.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-17 16:14:53
30dB is quite different from 1500dB and can be perfectly handled with integer math.
Not if you started out anywhere close to peak-normalized. Which, so it happens, often is the case. No, certainly it does not have to be that way, but it seems to be. Integer formats go to full volume, and more bits -> more at the bottom.
(Someone here - I don't remember who - once championed float for end-user format if only to screw up the loudness war. I'd give that a "fascinating thought experiment!".)

Quote
Actually, many products use this approach, including RME:
Google link led me to https://archiv.rme-audio.de/en/support/techinfo/hdsp_totalmix_hardware.php , which is interesting. One format per operation type?!
The "multiplier" uses 24 bits integer, but now you got a full mighty sixteen bits volume control, that's more like it. ("65563" :-) )
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-17 16:53:18
Not if you started out anywhere close to peak-normalized. Which, so it happens, often is the case. No, certainly it does not have to be that way, but it seems to be. Integer formats go to full volume, and more bits -> more at the bottom.
From the perspective of a file format, integer has to pad to the MSB for consistent level to end users. For the perspective of processing, the DSP designers can have their own implementations. That means, regardless of integer or floating point math, for a field recorder, it only needs to spew out a normalized 24 or 16-bit file, and that's why I said in Reply #139 "both 32-bit formats are useless as distribution formats". In fact, integer math often involve "accumulator" -- the internal bit-depth, even for apparently simple hardware like DAC chips, with 48-64 bits. It is just a trade off between processing speed and memory bandwidth. Floats need processing power to shift bits, integers need more bandwidth so no shifting math is used. Dither is not always used, but if used, only when converting from the accumulator's bit-depth to the destination bit depth, instead of in every intermediate step.

Quote
Google link led me to https://archiv.rme-audio.de/en/support/techinfo/hdsp_totalmix_hardware.php , which is interesting. One format per operation type?!
The "multiplier" uses 24 bits integer, but now you got a full mighty sixteen bits volume control, that's more like it. ("65563" :-) )
Consider these products either output to analog or SPDIF/AES3/ADAT (24-bit integer), or in reverse, record from these sources, more precision means nothing other than added cost or fewer simultaneous real time operations.

Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-18 11:35:40
Download the pdf files and see how poor the results, even after firmware update. Really doesn't justify the use of 32-bit float file format, as users ultimately need to scale down the misconfigured float file to see the full waveform and listen to it anyway. It could be bad DSP logic and/or broken analog stage led to these results.
https://www.audiosciencereview.com/forum/index.php?threads/zoom-f6-portable-field-recorder-review.15668/post-786854

Attached RMAA results of my 12-bit version of UA-law for comparison, the Zoom's 32-bit float 192kHz file is not really better than 12-bit 96kHz version of UA-law. Notice when the signal is stronger, UA-law's noise floor and noise shaping are also stronger, and vice versa. The internal math is not 12-bit of course, the programming language used simply doesn't have these data types. Even 24-bit is created by combining 3 bytes to 32-bit integer.
https://hydrogenaud.io/index.php?topic=121181.msg1005031#msg1005031

Actually in ASR, MC, RME's boss, pointed me to that old TotalMix article, while I and other members were speculating whether TotalMix FX (FX = DSP effects including reverb, EQ etc) is completely done on the FPGA or not. We were talking about Chord's poor use of FPGA processing power on the M-Scaler, a very expensive hardware resampler.
https://www.audiosciencereview.com/forum/index.php?threads/chord-hugo-m-scaler-stereophile-review-measurements-also.11868/post-658146

Scroll down to see MC's replies, as well as the next pages.
https://www.audiosciencereview.com/forum/index.php?threads/chord-hugo-m-scaler-stereophile-review-measurements-also.11868/post-677200
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-18 22:11:37
Off topic alert, but:
Who the f**c is the apparent forum joke Rob Watts? The musician? Has he made claims about how his ears need higher resolution audio than Neil Young's or something? Released a sad song about "My dog left me out of envy for my hearing and he took my wife with him"?
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-19 06:13:46
Rob Watts = Chord's boss.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: NateHigs on 2022-01-19 09:09:39
If I'm paying for something, I'll take as many bits as possible please! I don't care about a race to the bottom, or calculating the exact frequency that I can no longer hear - if this music was produced in 192kHz @ 32bit, I'll take that please.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: bennetng on 2022-01-19 10:26:24
https://hydrogenaud.io/index.php?topic=108864.msg948272#msg948272
If I am the one who sells music I'd be extremely happy to have these kinds of customers.

http://melancholyaudio.blog.fc2.com/blog-entry-17.html
Google translate:
Quote
Next, let's take a look at the graph with DSEE HX. The result is ... Oh, the high frequencies are higher than I expected. What surprised me most was that it did not drop at all around 20kHz to 22kHz and was decaying naturally.

What the industry needs is more these kinds of plugins to fool the customers.
Title: Re: xiphmont’s ‘There is no point to distributing music in 24 bit/192 kHz’
Post by: Porcus on 2022-01-19 11:45:53
If I'm paying for something, I'll take as many bits as possible please! I don't care about a race to the bottom, or calculating the exact frequency that I can no longer hear - if this music was produced in 192kHz @ 32bit, I'll take that please.

If what I am offered is a file straight from the artist's DAW - the artist likely not even taking note that it is a different kind of .wav than in the stores - then yes please, float or not.
Not because of the audio quality, but because it is closer to the artist's hand. Just like if a painter releases a litho series and there is a video with "and this time I finally learned to do the damn printing by hand without help" - it is not because it makes the printing anything more accurate. Heck, when I realized that WavPack handles all the quirks of 32-bit formats and can restore the original .wav bit by bit, I went back to the .wavs and re-WavPack'd them, using the WavPack exe rather than fb2k.

Especially if the .wav is for free, less incentives of faking it. (I have a hunch that 32-bit .wavs tend to disappear from Soundcloud when fans ask why they cannot play them.)

One example still up on Soundcloud: https://soundcloud.com/termo-records/sets/the-opium-cartel . One of the label owners' own projects, with Tim Bowness of No-Man. The last file is as weird as 44.1 kHz/32-bit. And then there's a nice little Blue Öyster Cult cover.