Skip to main content

Topic: 24/96 releases sometimes just upscaled 16/44? (Read 18017 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
24/96 releases sometimes just upscaled 16/44?
Reply #25
Original analogue master tapes will certainly contain content above 24kHz, and a level of depth beyond that of 16-bits.
I agree about the 24 kHz, although most of it will be noise from the tape and the analogue recording chain. I'm less sure about the 16 bits, since IME tape noise is quite a bit higher than 16-bit dither. Tapes with Dolby-SR and Telcom noise reduction will likely perform better, but I didn't test that myself.
For those interested, I've put online a 24/96 sample of tape silence which was made during a large archiving project several years ago. Studer A820, no noise reduction, straight into a dCS ADC.
StuderA820TapeNoise2496.wav


If a recording doesn't show noise almost all the way up to the Nyquist frequency, there's probably some digital filtering involved after the ADC.
BTW, for the archiving project the client (a major record company) insisted that their old 16/44.1 masters (mostly U-matic format) were converted to 24/96. I don't know if they can easily see the difference between these upsampled versions and genuine 24/96 recordings. A spectogram will show the origin right away of course.
  • Last Edit: 04 June, 2012, 04:41:08 AM by Kees de Visser

24/96 releases sometimes just upscaled 16/44?
Reply #26
Original analogue master tapes will certainly contain content above 24kHz, and a level of depth beyond that of 16-bits.

I agree about the 24 kHz, although most of it will be noise from the tape and the analogue recording chain. I'm less sure about the 16 bits, since IME tape noise is quite a bit higher than 16-bit dither. Tapes with Dolby-SR and Telcom noise reduction will likely perform better, but I didn't test that myself.


My benchmark for judging this issue is to measure the level of audio bandpass filtered  2.5 to 5 KHz (- 3dB) with 4th order Butterworth filters at each end.

                           Left   Right
Min Sample Value:   -1.87   -1.4
Max Sample Value:   1.8   1.4
Peak Amplitude:   -84.88 dB   -87.37 dB
Possibly Clipped:   0   0
DC Offset:                 0    0
Minimum RMS Power:   -99.2 dB   -100.77 dB
Maximum RMS Power:   -97.61 dB   -99.46 dB
Average RMS Power:   -98.47 dB   -100.07 dB
Total RMS Power:   -98.47 dB   -100.07 dB
Actual Bit Depth:   32 Bits   32 Bits

Using RMS Window of 100 ms

I think that 16 bits will still do it but noise shaping would be required.


  • uart
  • [*][*][*][*][*]
24/96 releases sometimes just upscaled 16/44?
Reply #27
Hi Arnold, I love to see actual measurements like that.  Now just making sure I'm understanding your procedure. You're sampling a 2.5kHz interval of a "blank" mastering tape with high quality (24 bit?) ADC and the above are the statistics of the scanned waveform. Is that correct?
  • Last Edit: 04 June, 2012, 11:17:38 AM by db1989

  • jamie_P84
  • [*]
24/96 releases sometimes just upscaled 16/44?
Reply #28
LPs have a FR that extends beyond 20kHz, but what's there, at what level,  and how distortion-free is it?

Depends, as always, upon the condition of the record and the quality of the recording/playback equipment.
As an example though, at least one commercial quadraphonic LP system (Quadradisc) used a pair of modulated carriers for the extra channels' content, cut to the vinyl in the range of 18-45kHz. High end record players can reproduce even higher frequency signals.

Quote
I have many AAD CDs which show no signs of roll-off at 20kHz, incidentally.

So, how have you determined that these CDs you refer to have 'no signs of rolloff' at the redbook limit?

I use an old WinAmp plugin called "DSP Spectrum Tool" (it crashes on newer WinAmp versions) to reliably check for roll off.
CD content derived from an oversampling ADC often has no roll off at 20kHz - frequency response can remain flat up to 22.05kHz.

'fails gracefully?  "subjectively usable dynamic range'?  capturing extra 'information'?

Welcome to analogue recording - things aren't so clean-cut as with digital are they.

Quote
PS. It would be rather unfortunate if this thread were to turn into "yet another analogue vs digital debate", all because my original post insulted certain people's "religion".


PS I took you not to be a troll, but now I wonder.

My original post was about sourcing genuine native 24/96 recordings to try out on my new soundcard - nothing more.
I anticipated the instinctive rush to defend the holy redbook, which is why I begged my superiors to show me mercy. What more did you want?

Anyway, here is an example spectrum plot of a 24/96 DVD-A release with sharp roll-off just below 22.05kHz (the display range is 0 - 48kHz and -130dB - 0dB):

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #29
Arnold is just performing some processing and then running an analysis using Adobe Audition. The sample was provided by Kees who works in the industry.

I haven't looked at the sample myself, but I would imagine that it won't be ABX-able from a 16-bit rendering when peak level is played back at normal listening levels, even without the use of dither.
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • uart
  • [*][*][*][*][*]
24/96 releases sometimes just upscaled 16/44?
Reply #30
Arnold is just performing some processing and then running an analysis using Adobe Audition. The sample was provided by Kees who works in the industry.

I haven't looked at the sample myself, but I would imagine that it won't be ABX-able from a 16-bit rendering when peak level is played back at normal listening levels, even without the use of dither.

OK thanks Greynol, I see it now. I was just wondering where the sample came from as I originally didn't notice the link to the wavfile in Kees's above post.

  • Kohlrabi
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #31
As an example though, at least one commercial quadraphonic LP system (Quadradisc) used a pair of modulated carriers for the extra channels' content, cut to the vinyl in the range of 18-45kHz. High end record players can reproduce even higher frequency signals.

Why would you want that?

'fails gracefully?  "subjectively usable dynamic range'?  capturing extra 'information'?

Welcome to analogue recording - things aren't so clean-cut as with digital are they.

You can use the same terms and methods for both analogue and digital audio, since you can sample your analogue signal into a digital form without loss of information. Then you can compare both in the digital domain. You don't have to make up terms which you only define for analogue recordings, to imply that there is some "hidden" information in the analogue recordings.

My original post was about sourcing genuine native 24/96 recordings to try out on my new soundcard - nothing more.

From my understanding you wanted to have a reason why there are 24/96 releases with sharp roll-offs, which are apparently sourced off of lower sampled releases:
So, is it normal for these high resolution releases to be upscaled?

And this dissolved into (albeit interesting) banter about technicalities of high-res audio recording.
The real question was never investigated: Why do some recordings look like they are just upsampled? And to me the answer is quite simple: Either there were no master tapes which higher frequency content, or the content was deliberately removed. Or the person in charge was unable to create a high-res master. In any case some consumers will just buy anything with large enough numbers on it, so it is a good business decision to just deliver what the customer wants and cut some corners in the process, if another approach takes too much effort.
  • Last Edit: 04 June, 2012, 12:03:56 PM by Kohlrabi
It's only audiophile if it's inconvenient.

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #32
I guess they didn't count on their customer base to listen with their eyes?

13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • uart
  • [*][*][*][*][*]
24/96 releases sometimes just upscaled 16/44?
Reply #33
Average RMS Power:   -98.47 dB   -100.07 dB
Total RMS Power:   -98.47 dB   -100.07 dB
Actual Bit Depth:   32 Bits   32 Bits

Using RMS Window of 100 ms

I think that 16 bits will still do it but noise shaping would be required.


Isn't that already a little worse than 16 bit quantization noise before noise shaping? Since that's only a 2.5kHz interval then wouldn't the noise over the 20 kHz spectrum be +10log(8) = +9 dB. That would make the tape about -90dB compared to the roughly -96dB quantization noise of 16 bit?

So I guess that would make 96k-24bit remasters from tape a bit of a waste of time, unless of course you're a bat. 

Edit: And a bat who really likes listening to noise + the squeal of studio monitors at that. 
  • Last Edit: 04 June, 2012, 12:18:03 PM by uart

  • uart
  • [*][*][*][*][*]
24/96 releases sometimes just upscaled 16/44?
Reply #34
BTW. I assume the gain settings for the recording that Kees made were appropriately adjusted so as to avoid clipping at the maximum signal strength that the tape can hold.  If the gain settings were cranked up then it wouldn't be a fair (for the tape) comparison. Similarly if the gain was underdone then it would make the tape look better than it really is. So I'm hoping the gain settings were well chosen for a variety of tape content (not just for the silence), since this was part of an archival session.
  • Last Edit: 04 June, 2012, 12:15:02 PM by uart

  • jamie_P84
  • [*]
24/96 releases sometimes just upscaled 16/44?
Reply #35
You don't have to make up terms which you only define for analogue recordings, to imply that there is some "hidden" information in the analogue recordings.

Which is precisely why I've done nothing of the sort. I suggest you re-read my comments in post #21, rather than krabapple's negative parody of them.

From my understanding you wanted to have a reason why there are 24/96 releases with sharp roll-offs, which are apparently sourced off of lower sampled releases:

The reason is certainly of interest to me, but primarily just I want to hear what genuine 24/96 sounds like on my system!

The real question was never investigated: Why do some recordings look like they are just upsampled? And to me the answer is quite simple: Either there were no master tapes which higher frequency content, or the content was deliberately removed. Or the person in charge was unable to create a high-res master. In any case some consumers will just buy anything with large enough numbers on it, so it is a good business decision to just deliver what the customer wants and cut some corners in the process, if another approach takes too much effort.

If a 24/96 master is not available, why bother releasing on the high resolution formats at all?
I wouldn't expect to buy a movie on Blu-Ray for example, only to find that it had been upconverted from DVD-quality video, when a high definition master was known to exist.
  • Last Edit: 04 June, 2012, 01:07:58 PM by jamie_P84

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #36
primarily just I want to hear what genuine 24/96 sounds like on my system!

Unless there is a problem likely no different than not genuine 24/96. Don't bother claiming otherwise unless you can also provide double-blind test results indicating such.

I wouldn't expect to buy a movie on Blu-Ray for example, only to find that it had been upconverted from DVD-quality video, when a high definition master was known to exist.

Maybe you can come up with a better analogy.  This red herring falls pretty flat.
  • Last Edit: 04 June, 2012, 01:16:42 PM by greynol
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • jamie_P84
  • [*]
24/96 releases sometimes just upscaled 16/44?
Reply #37
primarily just I want to hear what genuine 24/96 sounds like on my system!

Unless there is a problem likely no different than not genuine 24/96. Don't bother claiming otherwise unless you can also provide double-blind test results indicating such.

Well that's the closest you've come to making a useful contribution to my thread, so congratulations I guess. It still doesn't answer my original question, but at least it's not personally insulting or sarcastic this time.

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #38
I suggested you consider whether you're actually using your auditory system earlier.  It would have saved you a lot of trouble.
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • jamie_P84
  • [*]
24/96 releases sometimes just upscaled 16/44?
Reply #39
I suggested you consider whether you're actually using your auditory system earlier.

A typical response, which I'd anticipated in post #1.

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #40
...which I read, I assure you.  So much for trying to save you a lot of trouble.

Seriously, if you wish to listen with your eyes, you've come to the wrong place.
  • Last Edit: 04 June, 2012, 01:54:35 PM by greynol
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • jamie_P84
  • [*]
24/96 releases sometimes just upscaled 16/44?
Reply #41
...which I read, I assure you.  So much for trying to save you a lot of trouble.

Seriously, if you wish to listen with your eyes, you've come to the wrong place.

I wish to listen with my ears - that's why I'm seeking genuine 24/96 content (again covered in post #1).
But of course, when people listen with their ears and then share personal anecdotes of the experience, it violates the TOS.

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #42
Ok, but again, how did you determine there is a "problem"?

Anyway, if you're interested in determining the performance of your hardware, there are more objective means.  Have you considered something like the rightmark analyzer?

FWIW, it is perfectly OK to talk about listening experiences so long as they are supported with evidence in the way of double-blind test results in order to make sure what is being said is based solely on what is heard "with the ears" and not subject to expectation bias.
  • Last Edit: 04 June, 2012, 05:49:19 PM by greynol
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • Kohlrabi
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #43
If a 24/96 master is not available, why bother releasing on the high resolution formats at all?

You get extra money from people who buy 24/96 DVD-As just for the higher numbers. Didn't you prove that yourself by buying 24/96 DVD-As which clearly are just upsampled CDs? To be fair, no one can know beforehand that they're being screwed over, but what they can know is that they essentially buy snake oil, or a solution to a nonexistent problem.

I do see some merit in DVD-A releases though, in cases where the CD masters are so fucked up by loudness war engineers that you want to get your hands on a presumably proper master. But certainly not for the ultrasonics.
  • Last Edit: 04 June, 2012, 04:42:26 PM by Kohlrabi
It's only audiophile if it's inconvenient.

  • Ron Jones
  • [*][*][*][*]
24/96 releases sometimes just upscaled 16/44?
Reply #44
I can provide you with some non-musical samples guaranteed not to be rolled off at 22 kHz, if that's all you really want.

  • db1989
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #45
Some people: Stop the stupid troll-bait.

Some other people: For the love of all that is holy rational, stop responding to said troll-bait with more of your own.

Look what you did.

  • krabapple
  • [*][*][*][*][*]
24/96 releases sometimes just upscaled 16/44?
Reply #46
LPs have a FR that extends beyond 20kHz, but what's there, at what level,  and how distortion-free is it?

Depends, as always, upon the condition of the record and the quality of the recording/playback equipment.
As an example though, at least one commercial quadraphonic LP system (Quadradisc) used a pair of modulated carriers for the extra channels' content, cut to the vinyl in the range of 18-45kHz. High end record players can reproduce even higher frequency signals.



Hmmm, you quote and answer one question of mine out of several.

For a quad carrier signal -- that's the one case where it really is demonstrably important to get that ultrahigh frequency signal onto vinyl intact, and to preserve it.  Not so you can *hear* that frequency, mind you.  And btw analog quad is a dead technology, have you noticed?

So, for a typical analog product  (stereo LP not being quite so dead as quad LP),  what is the demonstrable importance of having  >=20kHz signals pressed to vinyl undistorted and at full level? And how commonly is that achieved?


Quote
Quote
I have many AAD CDs which show no signs of roll-off at 20kHz, incidentally.

So, how have you determined that these CDs you refer to have 'no signs of rolloff' at the redbook limit?

I use an old WinAmp plugin called "DSP Spectrum Tool" (it crashes on newer WinAmp versions) to reliably check for roll off.
CD content derived from an oversampling ADC often has no roll off at 20kHz - frequency response can remain flat up to 22.05kHz.



No one disputes that Redbook-rate ADC can yield an flat FR across practically all its bandwidth, with respect to the input signal, maybe fading a little at the very top. My question meant, how do you know there wasn't any rolloff  earlier in the recording chain, if the input is coming from, say, an old analog tape?  What 'signs' would you look for? 



Quote
My original post was about sourcing genuine native 24/96 recordings to try out on my new soundcard - nothing more.
I anticipated the instinctive rush to defend the holy redbook, which is why I begged my superiors to show me mercy. What more did you want?


I pretty much know what to expect from you now,  I think. 


Quote
Anyway, here is an example spectrum plot of a 24/96 DVD-A release with sharp roll-off just below 22.05kHz (the display range is 0 - 48kHz and -130dB - 0dB):



Yes, some 'high rez' releases don't really exploit the touted advantages of 'high rez' (and did you know, many CDs don't exploit the 16bits of dynamic range available to them?).  Yes, some 'high rez' release are just upconversions of previously lowpassed signals.  And actually, even if they were bona fide 96kHz/24bit recordings from start to finish, it might be very difficult tell them apart from a Redbook downconversion in a fair listening test, and even then the difference, if heard, might be due to hardware, rather than the audio format.  Implying that as a consumer audio delivery format, 'high rez' might be more a marketing ploy than anything else.


Last but not least, if you do plan to 'use your ears' to determine if a true 24/96 recording differs audibly from the same recording at 16/44, does that mean you'll do the test blind, or is that sort of thing  too religious for a rebel like you?
  • Last Edit: 04 June, 2012, 11:14:22 PM by krabapple

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
24/96 releases sometimes just upscaled 16/44?
Reply #47
IIRC and do correct me if I'm wrong, the modulation scheme used for quadraphonic vinyl doesn't require a flat frequency response, something vinyl (as a storage format irrespective of playback hardware) isn't known for doing very well at high (~20k) frequencies.
  • Last Edit: 05 June, 2012, 01:04:48 AM by greynol
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

24/96 releases sometimes just upscaled 16/44?
Reply #48
IIRC and do correct me if I'm wrong, the modulation scheme used for quadraphonic vinyl doesn't require a flat frequency response, something vinyl (as a storage format irrespective of playback hardware) isn't known for doing very well at high (~20k) frequencies.


We're talking CD-4 right, because SQ and QS were totally uncaring about HF FR.

Generally true of CD-4. The decoders of the day just required a HF carrier that was above some meager threshold. It didn't have to be clean and it wasn't.

The actual performance of the decoding was very poor by any standards, even the standards of the day.  It worked something like stereo FM, but stereo FM was a picture of stability and reliability in comparison. 

Most CD-4 recordings could only be played a small number of times, and then the best stylii of the day (or modern ones) damaged the carrier so badly that it wouldn't "Turn on the light".

24/96 releases sometimes just upscaled 16/44?
Reply #49
But of course, when people listen with their ears and then share personal anecdotes of the experience, it violates the TOS.


That is arguably libel. There can be no problem with anecdotes that are meaningful when taken at face value. But anecdotes that are all about how someone is in denial about their biases and normal human limitations get old really fast. I can make them up all day long and tell them to myself if that is what I wanted to do. ;-)