Thread: Where are these 24/192 files coming from?

I was just on Qobuz a French site that sells high res albums sorta a la HDtracks. In fact, some of the files that populate the French site are recent popular recordings of such acts as Gotye and Wild Beasts. Ok. So far so good, right? Well, the curious thing is virtually every popular piece of music on the site (and they list them as Studio Masters and indicate this is the native sampling and bit rates that they were recorded at) is recorded at 24/44.1. Not 24/48, not 24/96, and certainly not 24/192. Question then. In this day and age with HDtracks selling music from the 1980's as 24/192 for albums that likely never saw anything over 16 bits to begin with, are studios really recording just a bit better than CD quality in 2012? Apparently so. So if in 2012, most studios are still recording at 24/44.1, where the heck are all these 24/192 files coming from? To say nothing of your standard 24/96. So is anyone else a bit suspicious when say a 1980s album by Madonna gets the 24/192 treatment when it likely never saw anything beyond 16 of those bits? I mean if the studios are still just barely passing the cd quality mark (hey Im all for 24 bit...but it does beg the question) in 2012, just what is happening to these files? Anyway I'm not debating about whether they sound better or what have you. I'm just wondering where these files are coming from and how they are getting all this bandwidth from something that didn't have it to begin with?

I kind of get it if something began as a digital recording at 16 bit and is now trying to be peddled as 24 bit but what about analog recordings? When they are converted to digital and say it is at 24 bit even with a more limited frequency range and dynamic range of an analog source it really is 24 bit isn't it? Isn't the issue really digital recordings upsampled from 16 to 24 and not analog converted to 24?

Originally Posted by mwheelerk

I kind of get it if something began as a digital recording at 16 bit and is now trying to be peddled as 24 bit but what about analog recordings? When they are converted to digital and say it is at 24 bit even with a more limited frequency range and dynamic range of an analog source it really is 24 bit isn't it? Isn't the issue really digital recordings upsampled from 16 to 24 and not analog converted to 24?

Agreed - if a studio / release company goes back to the original (or backup or whatever) tapes and samples to digital at 24/192, that's legitimate to me (if a bit overkill for tapes that have no content over 48k or lower). But taking a digital recording that was archived at 24/44.1 (say) and reselling it upsampled to 24/192 is not what I'd consider honest or even useful.

Originally Posted by mwheelerk

I kind of get it if something began as a digital recording at 16 bit and is now trying to be peddled as 24 bit but what about analog recordings? When they are converted to digital and say it is at 24 bit even with a more limited frequency range and dynamic range of an analog source it really is 24 bit isn't it? Isn't the issue really digital recordings upsampled from 16 to 24 and not analog converted to 24?

I sorta get the analog thing too. For recordings in the 70's that would make sense. But once you get to a certain year, most of the music was done digitally. If you go from say 1990 onwards, we're talking just about everything except for the producers who are digitally averse. It's just very interesting to me. Like the Madonna album on HD. No way that was analog, though it could have been transferred to tape. I'm just lost in terms of what is being added value wise. The artists I was referring to above both have albums from 2011. And they were natively recorded at 24/44.1. Just seems very strange. I'm sure the word length certainly adds more than the samples--but it just begs the question that if we are still recording at this level at major labels where are these 24/96 files from major labels coming from (HD has had a few recent releases)? To say nothing of the other rates. I have a feeling we are seeing a LOT more upsampled work than even Audacity charts would indicate.

Originally Posted by bleedink

I sorta get the analog thing too. For recordings in the 70's that would make sense. But once you get to a certain year, most of the music was done digitally. If you go from say 1990 onwards, we're talking just about everything except for the producers who are digitally averse. It's just very interesting to me. Like the Madonna album on HD. No way that was analog, though it could have been transferred to tape. I'm just lost in terms of what is being added value wise. The artists I was referring to above both have albums from 2011. And they were natively recorded at 24/44.1. Just seems very strange. I'm sure the word length certainly adds more than the samples--but it just begs the question that if we are still recording at this level at major labels where are these 24/96 files from major labels coming from (HD has had a few recent releases)? To say nothing of the other rates. I have a feeling we are seeing a LOT more upsampled work than even Audacity charts would indicate.

Actually, the first few Madonna albums from the early 80s *were* recorded analog, so I can see them being resampled for HD Tracks. But I thought that "Like A Virgin" was digital, so . . . ?

I really want to see the provenance of these "high res" albums published prominently.

Originally Posted by jhwalker

Actually, the first few Madonna albums from the early 80s *were* recorded analog, so I can see them being resampled for HD Tracks. But I thought that "Like A Virgin" was digital, so . . . ?

I really want to see the provenance of these "high res" albums published prominently.

According to the site though, it was recorded at 24/44.1 as was Wild Beasts Smother. That's why it is particularly unnerving to see what is really being recorded in studios (commercial studios) and what is being sold. You would think that everything the studios do these days would be 24/96 minimum. Not so if you use these two releases as examples. You bring up a disturbing trend...as I see it. It seems that at times they are taking digital masters (or presumably digital masters) and transferring them to tape to transfer them to a higher sample rate than they were originally. That's what I see happening. Or some extraordinary upsampling that is transparent in graphs and such. This was my humble opinion but I can see no other explanation for some of this weirdness. Are we to really believe that some of this stuff was archived on tape for the last 40 years? Madonna? Really? And if they are digital what gives? We know what technology was around at the time. And it appears that even today the studios aren't using exactly cutting edge word lengths and sampling rates. Gotye isn't exactly unknown right now. They seem to be pretty explicit with the provenance of this file at least. It just smells really funny to me.

To the OP, I can confirm that Gotye has made the Studio Master available also from his website so it would appear that it is legit.

24 bit masters available here - http://gotye.com/#store.html

If you like classical you can hop over to channel classics and find files listed as 24/192. I'm assuming that this is an accurate representation because the 24/96 files that I've downloaded were recorded at that level.

Originally Posted by Brian D

To the OP, I can confirm that Gotye has made the Studio Master available also from his website so it would appear that it is legit.

24 bit masters available here - http://gotye.com/#store.html

Thanks I saw reference to that as well. It seems to confirm that recordings are being done right now in 24/44.1. Not that there is anything wrong with that in particular. There isn't. But when you see stuff from 30 years ago in significantly higher resolution, when you can wiki the recording chain sometimes, like I said I get a bit suspicious. If the studios were in fact recording 24/96 as a de facto standard then I wouldn't be scrutinizing this tiny factoid quite so much. Anyway if what is being recorded today isn't even up to Madonna circa 1983 then something is terribly wrong here.

Originally Posted by LBob

If you like classical you can hop over to channel classics and find files listed as 24/192. I'm assuming that this is an accurate representation because the 24/96 files that I've downloaded were recorded at that level.

I'm aware I can get classical in 24/192 in 6 channels if I want. I do appreciate you having pointed that out. I was really speaking more to commercial 'pop' music if you will. Classical doesn't quite need the renaissance that more commercial music deserves.

Originally Posted by bleedink

Anyway if what is being recorded today isn't even up to Madonna circa 1983 then something is terribly wrong here.

Well, even if they recorded it in 32/384, it doesn't help when they then compress it to death, and let it clip badly, and compress again. Suggested reading: "Perfecting Sound Forever - The Story Of Recorded Music" by Greg Milner. A real eye-opener.

Originally Posted by Julf

Well, even if they recorded it in 32/384, it doesn't help when they then compress it to death, and let it clip badly, and compress again.

Can't argue with that. I guess like someone said above that I assumed everything today was being recorded at least 24/96 if only to rerelease it in a box set in another 40 years for 3 or 4 figures.

Originally Posted by LBob

If you like classical you can hop over to channel classics and find files listed as 24/192. I'm assuming that this is an accurate representation because the 24/96 files that I've downloaded were recorded at that level.

Channel Classsics masters are DSD.
24/96 and 24/192 are DSD to PCM conversions made with Weiss Saracon software.

For Pop and Jazz 24/44.1 is the most common figure for masters nowadays.
So it is for many classical recordings.
Some are DSD and some 96 and 192, but they are "rara avis".

Originally Posted by bleedink

Can't argue with that. I guess like someone said above that I assumed everything today was being recorded at least 24/96 if only to rerelease it in a box set in another 40 years for 3 or 4 figures.

Naah, it will be re-released as "mastered for iPop"

Originally Posted by Manelus

For Pop and Jazz 24/44.1 is the most common figure for masters nowadays.
So it is for many classical recordings.

i guess the record industry has to cut costs somewhere, as those pesky pirates eat away their business, and hard disk space is sooooo expensive these days... :-/

Analog record have near infinite low level of quantization (analogically bit-depth for digital record).

Why "near infinite"?
It restricted by physical grains of analog record medium (vinyl, tape).

Increasing of bit-depth lead to effect comparable with recording by "physical grains".
For low levels of digital musical signal we have rather high level of distortions.
"Physical grains" is equal high bit-depth.
For low levels of analog musical signal we have low level of distortions.

We have decrease low level distortions by other ways also.
These ways is hard released for exists situation.
Therefore new records increase bit-depth (easy way).

24-bit is not level of "physical grains". Using 32-bit we get more precision of measurement analog master or acoustical source.
But we have question: hear it or not?

Originally Posted by audiventory

Analog record have near infinite low level of quantization (analogically bit-depth for digital record).

Not really. One (simplistic, but useful) way to measure dynamic range (and that is what amplitude resolution provides - I assume that is what you mean by "low level of quantization". You either have quantization, or you don't) is signal-to-noise ratio. I have not come across a tape system with more than 90 dB SNR, and often it is in the 70-75 dB range, so less than the 96 dB that CD/Red Book provides even without dithering. 24 bit theoretically provides 144 dB.

Therefore new records increase bit-depth (easy way).

Not suer I understand what you mean.

24-bit is not level of "physical grains". Using 32-bit we get more precision of measurement analog master or acoustical source.
But we have question: hear it or not?

Indeed. So 146 dB is not enough? I don't think going to 192 dB would make any difference - show me a listening room with background noise that low

Audiventory,

Originally Posted by audiventory

Audiophile Inventory ConverteR 48x44 - professional sample rate converter WAV, AIFF, FLAC for sound production and exact audiophiles. Used for suppressing of playback intermodulations for some HD-audio files and equipment.http://samplerateconverter.com

Is that a commercial product? Can you perhaps tell us what advantages it provides compared to a free, open source tool such as sox or audacity?

Not really. One (simplistic, but useful) way to measure dynamic range (and that is what amplitude resolution provides - I assume that is what you mean by "low level of quantization". You either have quantization, or you don't) is signal-to-noise ratio.

I'm agree. Model "physical grains" is not scientific of course.
We can call as "Grains" the magnetic domains of ferrum (tape material).

For analog we have:
1) Max level - where input signal level increase - but output level fixed.
2) Min level - level of noise.
Between 1) and 2) - dynamic range. It measured as relation of absolute values. As example, near 60 dB for tape.
"Analog quantisation" - is resolution of "physical grains". As example, 2¹²⁸ or 2⁶⁴²
Really we can't measured "Analog quantisation". But analog resolution into 60 dB range is very good. Better than digital.
For constant absolute levels resolution may be different for different materials. I don't know about instruments (microscope?), for measured "Analog quantisation". It's area of nano-technology.

24 bit theoretically provides 144 dB.

Theoretically - yes.
Practically: lower level - more harmonic distortions (little number of "quantization steps" for correct DAC-restoring of signal).
I say about real (non-ideal) DAC's output low frequency filter and stochastic (random) musical signal.
Ideal DAC 16 bit/44 kHz will make ideal restoring for sinusoide with infinite duration.
For musical signal with restricted time of each its components (sinusoids or harmonics) - not.

Indeed. So 146 dB is not enough? I don't think going to 192 dB would make any difference - show me a listening room with background noise that low

Real noise of modern 24-bit ADC is minus 115-120 dB.
We not hear it, of course. Only ear into speaker

In record pause we can make noise level is minus infinity (edit audio file by musical editor) even.

Our ears listen 0 to 140 dB (to damage). We can improve resolution or bit-depth (for fidelity of restoring of audio signals) into ear's dynamic range.

Is that a commercial product? Can you perhaps tell us what advantages it provides compared to a free, open source tool such as sox or audacity?

That is about my converter (FREE version available also). According terms of CA I can't discuss it.
I shortly (off topic for this thread) say about main ways quality of pro sample rate converters.
1) Low level of harmonic distortions (aliasing).
2) Linear phase characteristic of conversion.
3) Time of ringing for impulse (parasitic additional signal after conversion for sounds of piano, guitars, drums, etc.).

It's voluminous theme. About details I wrote in article "Sample rate conversion".
If interestingly, I ready to discuss SRC-processing.

Originally Posted by audiventory

I'm agree. Model "physical grains" is not scientific of course.
We can call as "Grains" the magnetic domains of ferrum (tape material).

Yes, that would be a more correct term.

For analog we have:
1) Max level - where input signal level increase - but output level fixed.
2) Min level - level of noise.
Between 1) and 2) - dynamic range. It measured as relation of absolute values. As example, near 60 dB for tape.
"Analog quantisation" - is resolution of "physical grains". As example, 2¹²⁸ or 2⁶⁴²
Really we can't measured "Analog quantisation". But analog resolution into 60 dB range is very good. Better than digital.

This is where I get confused. How is 60 dB better than the 96 dB that 44.1/16 (CD/Red Book) provides?

Ideal DAC 16 bit/44 kHz will make ideal restoring for sinusoide with infinite duration.
For musical signal with restricted time of each its components (sinusoids or harmonics) - not.

Right. The difference between the ideal signal and the digital one is the digitization noise, and that is at -96 dB even with CD material (without dithering), and -144 dB with 24 bits.

Real noise of modern 24-bit ADC is minus 115-120 dB.
We not hear it, of course. Only ear into speaker

Indeed. And that is just "normal" thermal/electronics noise, the quantization noise is much lower still.

Our ears listen 0 to 140 dB (to damage). We can improve resolution or bit-depth (for fidelity of restoring of audio signals) into ear's dynamic range.

The figure for hearing damage most often quoted is 120, and the pain threshold 130 dB. The 0 dB lower limit is in ideal conditions at 1 kHz. Much higher at other frequencies. So in any case, 24 bits give you a range well in excess of the human hearing range.

That is about my converter (FREE version available also). According terms of CA I can't discuss it.

OK, fair enough.

I shortly (off topic for this thread) say about main ways quality of pro sample rate converters.
1) Low level of harmonic distortions (aliasing).
2) Linear phase characteristic of conversion.
3) Time of ringing for impulse (parasitic additional signal after conversion for sounds of piano, guitars, drums, etc.).

Sure. So without discussing your own product, can you tell us in what way the numerous algorithms available in the free, open-source SOX package fall short on those 3 criteria?

First I am NOT being facetious.

Just want to get that out of the way.

After reading the last couple posts the only thing I can add is: Huh?
I know I can't be the only one on the site with a rudimentary understanding of audio concepts...any chance of transcribing what you guys have added recently in more accessible way? Went right over my head...particularly how it related to the topic. It might even be on topic for all I know. Like I said not being facetious, I'm just not as educated in some of these concepts as you folks and would like to be able to understand well enough to keep reading!

Originally Posted by bleedink

I know I can't be the only one on the site with a rudimentary understanding of audio concepts...any chance of transcribing what you guys have added recently in more accessible way? Went right over my head...particularly how it related to the topic. It might even be on topic for all I know.

Unfortunately the "how is this related to the topic" one is one I can't answer. I just replied to the posting by audiventory. I probably shouldn't have done that. I was just trying to make sense of his/her claims.

Originally Posted by Julf

Unfortunately the "how is this related to the topic" one is one I can't answer. I just replied to the posting by audiventory. I probably shouldn't have done that. I was just trying to make sense of his/her claims.

What exactly was he claiming? That analog is better than digital or that his upsampling product can take this 24/44.1 stuff to the moon for future 24/96 box sets? I guess I missed the entire point...