Inside High-res audio: PCM vs MQA vs CD: 2L Sampler Comparison

[krabapple]

It is no accident recording engineers pretty well all know that 24-bit is handy at the recording stage and 16-bit more than enough for the commercial release.

agreed

The only thing I wonder about is home use situations where there is a lot of DSP going in. Then again, I suspect any device that does that, upconverts any source that's 16bits, to 24

 

[krabapple]

The best comparison/analysis of MQA I've seen is this:

and again, that is a a very silly headline. 'Worse than FLAC?" implies that FLAC is somehow bad. How about just "MQA: not lossless'?


The best bench breakdown of MQA performance I've seen are the series of posts by Archimago , already cited here.

 

[krabapple]

Joe Jacksons The Verdict, from the 1984 CD-issue, recorded on a 3M Multitrack with 16bit/50kHz(!) probably transferred to 44.1 in the analog domain. Unfortunately Sonic Visualiser is not very intuitive and seems to lack simple data like RMS or headroom.

EDIT: Headroom is 0dB, RMS-20.2dB

That would be better displayed by a waveform, no?

 

[krabapple]

Agreed - the finding is not news, and yet it needs to be repeated often, even constantly, because too many folks in the hobby are unaware of it and don't understand, and too many vested interests in the industry want it to be ignored. I'm particularly glad you mentioned Oohashi and Boyk, since these are the "sources" - and the only sources - cited by Bob Stuart, Hans Beekhuyzen, and everyone else who tries to claim there's scientific evidence for the nonsense they peddle.

And as you note, these aren't even two sources, as Boyk's only contribution is to point out that musical instruments can produce harmonics above 20kHz (duh) - for the crucial next-step claim that these frequencies are audible to humans, Boyk simply relies on Oohashi's original, never-able-to-be-replicated study.

Indeed, the foundational manifesto of hi-rez -- Bob Stuart's of Meridien's white paper (and AES presentation) from 1997-8 -- is a tissue of flimsy evidence in this regard. Its publication in JAES provoked a skeptical response from certain AES members including Stanley Lipshitz, in a rebuttal letter afterwards.

 

[krabapple]

I've been exploring some other music from Mahler and other composers and although there are some points reaching the same difference, they correspond to silences or end of notes. In Stravinsky's pieces however, all music content is essential regardless its volume, and that makes those pieces quite gear demanding, in the sense that without a minimum quality system they are not understandable at all, you simply lose track of the music.

That's silly. People have been 'keeping track' of these masterpieces in recordings since long before the digital era, on all sorts of gear. My first exposure to Le Sacre was an orchestral performance in the mid 1970s shown on a public TV station; I was enthralled, though the playback gear was surely below 'minimum quality' of anything available today. The LP I bought soon after was played on a big hunk of 'console' furniture, a top-loading combo record player/receiver/speaker set in a big Italianate wooden case. Again, not even 'minimum' by today's standards.

 

[tmtomh]

Just so we're all clear here, are folks suggesting that some of these very dynamic classic recordings have more than 96dB of dynamic range in them? Or to put it another way, are folks suggesting that there are dynamic classical recordings out there where the quietest sounds in the performance have been lost because 16-bit digital PCM was incapable of capturing them?

If that's what's being claimed, suggested, or implied, can someone provide an example? I get that anything is possible. But as a practical matter, I'm thinking that even in a specially isolated chamber with ambient noise of only, say, 15dB, the peak level of the orchestra would have to be 111dB in order to max out the 96dB dynamic range of a 16-bit recording (leaving aside the role of dither in enabling recovering of information beyond the LSB). Even in that case, who is going to listen to the recording at 111dB playback levels? And even if they did, who's going to be listening in a playback environment with only 15dB ambient noise - even closed-back, over-ear headphones are still going to let in that much from a typical 30-40dB ambient listening room, yes?

I'm just struggling to understand what music folks have in mind that can't be fully captured and reproduced by 16-bit in real world human recording and listening conditions.

To be clear, I'm not questioning the utility of 24 or 32-bit for recording, mixing, and mastering - instead I'm asking, do folks citing super-dynamic classical recordings really think these recordings have 96+dB of dynamic range?

 

[krabapple]

here's data from a peculiar example - a Telarc release of Mahler's 2nd ( Slatkin conducting the Seattle Symphony, first released in 1982) that used the 50kHz/16bit* Soundstream Digital Tape Recorder for recording . Perceptually it is *extremely* 'dynamic' -- very quiet when quiet, damn loud when loud -- though actual DR is ~70dB. One of those records where you need a very quiet room to hear everything at a constant 'normal' playback level, otherwise you ride the volume control at the risk of releasing a sound bomb. I can't imagine wanting or needing more than this for playback of a consumer release.

Measured via the current version of Adobe Audition. These numbers are for the entire ripped album from the CD layer of the SACD (released 2006), concatenated into one track shown above:

    Left    Right
Peak Amplitude:    -0.04 dB    -0.06 dB
True Peak Amplitude:    -0.03 dBTP    -0.06 dBTP
Maximum Sample Value:    32612    31953
Minimum Sample Value:    -32540    -32527
Possibly Clipped Samples:    0    0
Total RMS Amplitude:    -28.48 dB    -27.60 dB
Maximum RMS Amplitude:    -7.45 dB    -8.67 dB
Minimum RMS Amplitude:    -78.32 dB    -78.12 dB
Average RMS Amplitude:    -40.98 dB    -40.28 dB
DC Offset:    0.00 %    0.00 %
Measured Bit Depth:    16    16
Dynamic Range:    70.87 dB    69.45 dB
Dynamic Range Used:    57.50 dB    58.10 dB
Loudness (Legacy):    -21.14 dB    -20.76 dB
Perceived Loudness (Legacy):    -18.55 dB    -17.92 dB
ITU-R BS.1770-3 Loudness: -21.16 LUFS

0dB = FS Square Wave
Using RMS Window of 50.00 ms
Account for DC = true

* : " The filtered analog signal passed through a custom sample and hold and was digitized by an Analogic MP8016 16-bit Analog-to-Digital Converter operating at a 50 kHz sample rate. A three-bit sync pattern and an even-parity bit were added to each 16-bit sample to form a 20-bit word that was serialized and transmitted by interface electronics to the tape transport where each audio channel's data were written to two separate tape tracks. " https://en.wikipedia.org/wiki/Soundstream

 

[Herbert]

That would be better displayed by a waveform, no?

Sonic Visualiser does not display RMS or headroom when showing a waveform.

 

[Herbert]

Funny that you should quote this as Dr. Doi was my boss at Sony (two levels up)! He would tell me many stories about working with Philips. None positive.

So you both might be happy to hear that I bought a CDP-101 NEW in 2014 because it was stored in the cellar of a HiFi dealer here in Germany for 32 years, three of them were still there; the sealed box had a thick layer of dust. Inside a. catalogue and a red cloth for wiping CD. Grease was dried, after cleaning an lubricating it sings

 

[restorer-john]

So you both might be happy to hear that I bought a CDP-101 NEW in 2014 because it was stored in the cellar of a HiFi dealer here in Germany for 32 years, three of them were still there; the sealed box had a thick layer of dust. Inside a. catalogue and a red cloth for wiping CD. Grease was dried, after cleaning an lubricating it sings

Did you get the Sony demonstration CD in the box? The first run came with a CD because there were none (not easily) to buy. What is the production date on your machine? (it's on the rear, bottom of the external transformer housing)

 

[AudioSceptic]

Confirmed:

https://www.philips.com/a-w/research/technologies/cd/beginning.html

Thanks. Of course, I now need to read all of it. ;-)

 

[restorer-john]

Thanks. Of course, I now need to read all of it. ;-)

Make sure you read Sony's side of it too on their corporate site.

Personally, based on all the books and research I have done over the years, I don't trust a thing Philips say. Philips were lucky to have Sony, not the other way around.

 

[xaviescacs]

That's silly. People have been 'keeping track' of these masterpieces in recordings since long before the digital era, on all sorts of gear. My first exposure to Le Sacre was an orchestral performance in the mid 1970s shown on a public TV station; I was enthralled, though the playback gear was surely below 'minimum quality' of anything available today. The LP I bought soon after was played on a big hunk of 'console' furniture, a top-loading combo record player/receiver/speaker set in a big Italianate wooden case. Again, not even 'minimum' by today's standards.

I wasn't talking about modern DACs and 115 SINAD, I'm talking about a minimum of dynamic range and resolution, which in the context of this forum is assumed, but it is not the same for the general public.

Perhaps "lose track" was inaccurate. I meant that you are missing important music content, more than just edge of notes.The fact that you were able to enjoy it doesn't mean that you weren't loosing something. It is almost impossible to audition The Rite of Spring or The Frebird on a modern TV a don't miss some of the music, bug chunks of it. I just wanted to point out that in some Stravinsky's music, there are a lot of musical content, not just a few moments, at less than -20db, which makes it very difficult to follow if you haven't a system with a minimum of resolution and dynamic range. It can be a state of the art system or a 40 years old one, but it can't be a 20 bucks earbuds while traveling, or a TV while having dinner, which are the situations when many people listen to music.

Talking about music, if you don't realize that the strings chord with the beat at the beginning of Action rituelle des ancêtres is highly dissonant, or you lose track of the winds line at some point, or you are unable to follow the percussion off the beat all the time, even at -25db, then you are just missing part of the music, and therefore not really listening to it as it was composed, and therefore you are unable to "follow it". In the Firebird, to "follow" the piece, you need to hear carefully the initial double bass theme to be able to recognize its transformation later on, but in doing so without adjusting the volume, you will have to face some really high volume peaks at some points that could be annoying.

In the Firebird's recording I have, the Boulez - Deutsche Grammophon, the introduction is between -15db and -20db all the time, with some passages below that.

However, in the Infernal Dance, there are many points where the loudest moments are simply cut.

It is so obvious, that I assume the engineers had to find a balance between audibility of the quiet fragments and maximum loudness. But I don't have any experience in that field so I'm just speculating.

Just so we're all clear here, are folks suggesting that some of these very dynamic classic recordings have more than 96dB of dynamic range in them? Or to put it another way, are folks suggesting that there are dynamic classical recordings out there where the quietest sounds in the performance have been lost because 16-bit digital PCM was incapable of capturing them?

To clarify, I wasn't. I think 16 bits is more than enough, as you and many have explained. I think it is a commercial problem, and a quite unsolvable one sometimes, as some classical music is composed to be performed and listened in very quiet environments.

With respect to that

though actual DR is ~70dB

I would add that this is the difference between the loudest moment and the quietest one, which will be most of the times a silence in the music, but it doesn't really mean that there is "music" that low. You have to listen or look at the spectrogram to look for the lowest points in music.

However, the recording I have of this first movement of the 2nd indeed has very quiet passages at well below -20db, and at -25db at some point. Which is quite low indeed!

I wanted to say thanks to everyone in this forum. I'm learning a lot thanks to your posts and answers.

 

[Frank Dernie]

here's data from a peculiar example - a Telarc release of Mahler's 2nd ( Slatkin conducting the Seattle Symphony, first released in 1982) that used the 50kHz/16bit* Soundstream Digital Tape Recorder for recording . Perceptually it is *extremely* 'dynamic' -- very quiet when quiet, damn loud when loud -- though actual DR is ~70dB. One of those records where you need a very quiet room to hear everything at a constant 'normal' playback level, otherwise you ride the volume control at the risk of releasing a sound bomb. I can't imagine wanting or needing more than this for playback of a consumer release.
View attachment 128881
Measured via the current version of Adobe Audition. These numbers are for the entire ripped album from the CD layer of the SACD (released 2006), concatenated into one track shown above:

    Left    Right
Peak Amplitude:    -0.04 dB    -0.06 dB
True Peak Amplitude:    -0.03 dBTP    -0.06 dBTP
Maximum Sample Value:    32612    31953
Minimum Sample Value:    -32540    -32527
Possibly Clipped Samples:    0    0
Total RMS Amplitude:    -28.48 dB    -27.60 dB
Maximum RMS Amplitude:    -7.45 dB    -8.67 dB
Minimum RMS Amplitude:    -78.32 dB    -78.12 dB
Average RMS Amplitude:    -40.98 dB    -40.28 dB
DC Offset:    0.00 %    0.00 %
Measured Bit Depth:    16    16
Dynamic Range:    70.87 dB    69.45 dB
Dynamic Range Used:    57.50 dB    58.10 dB
Loudness (Legacy):    -21.14 dB    -20.76 dB
Perceived Loudness (Legacy):    -18.55 dB    -17.92 dB
ITU-R BS.1770-3 Loudness: -21.16 LUFS

0dB = FS Square Wave
Using RMS Window of 50.00 ms
Account for DC = true

* : " The filtered analog signal passed through a custom sample and hold and was digitized by an Analogic MP8016 16-bit Analog-to-Digital Converter operating at a 50 kHz sample rate. A three-bit sync pattern and an even-parity bit were added to each 16-bit sample to form a 20-bit word that was serialized and transmitted by interface electronics to the tape transport where each audio channel's data were written to two separate tape tracks. " https://en.wikipedia.org/wiki/Soundstream

Yes this is a very dynamic recording. Way, way more than most peoplr have heard, I would wager, yet 16-bit is, as you illustrate, plenty.
This is absolutely the nub of my scepticism when people propose the need for more than 16-bits and always has been.
I think people are willing to look at the "numbers" without experimented on themselves with the actual sound levels involved and what they mean, to get an idea of what is credibly audible.
This is the case for both the dynamic range of music and the loudness of distortion artifacts IMHO.

 

[restorer-john]

Typical Telarc liner notes of the era, and recording details:

Note the Studer SFC-16 Sample Rate Converter used to take the 50kHz Soundstream to 44.1kHz. It was brand new, just introduced in1983

 

[Hayabusa]

That's not different than using the spectral view in Audition, as Amir does. You can alter the colors there too if you like.

I could not see thse tones in the video in the spectral view, but I assume you can change the range/colors in such a way that better visible...

Also audacity is for free

 

[Rottmannash]

Typical Telarc liner notes of the era, and recording details:
View attachment 128968

View attachment 128969

Note the Studer SFC-16 Sample Rate Converter used to take the 50kHz Soundstream to 44.1kHz. It was brand new, just introduced in1983

Cool. How does it sound? I have a CD of The Alan Parsons Project "On Air" which was DDD. Sounds great and very dynamic.

 

[AudioSceptic]

Not really as the ultrasonics are very likely below vinyl noise.Wikipedia states that 50kHz are possible, proven by test recordings, but I assume these are signals close to full scale.

The JVC/RCA CD-4 quad system of the 70s required response up to 45 kHz. This resulted in better-shaped styli such as Shibata being developed, which continue to be used to play stereo vinyl today.

 

[Nathan Raymond]

and again, that is a a very silly headline. 'Worse than FLAC?" implies that FLAC is somehow bad. How about just "MQA: not lossless'?

The best bench breakdown of MQA performance I've seen are the series of posts by Archimago , already cited here.

Archimago's posts are great. My intention wasn't to diminish anything Archimago has done, and Goldensound links to Archimago in the info section of that video. Correct me if I'm wrong, but I don't believe Archimago published any audio to MQA format for analysis which is what Goldensound did and I felt that brought something new to the discussion. As far as his video title, I took that to be the typical attention-draw approach for a YouTube video, and I translated FLAC to lossless in my head when I read it (I don't have any FLAC files in my collection since I turn them all into ALAC since it is a more convenient lossless format for my hardware and software). I felt the title wasn't entirely unwarranted in the sense that MQA does have a lot of (undeserved) popular mindshare in a way that FLAC doesn't, and FLAC hasn't had any licensing fees since day 1 so if someone wants to give more attention to FLAC, fine by me.

 

[AudioSceptic]

There are some subtleties here.
When sampling it is a requirement that dither is used. This is often not appreciated. An ADC that has no dither applied is incorrectly implemented and will yield objectionable signal correlated quantisation noise. This is true in any digital system, whether audio or any other system. Dither is not an option. It is a mathematical necessity to obtain correct operation from the quantiser. Dither is minimally AIWN of one half LSB. But it can be other distributions. Correct dither doesn't reduce the available bit depth. Incorrect dither might. Dither other than AIWN can lead to some surprising benefits. (Noise this low in amplitude is often below the inherent noise floor of the recording process. So long as there is AIWN in the chain - and most microphones will oblige, certianly any small diaphragm mic, there is often no real need to add it in explicitly. In early digital recordings this was often all that was done. But you need to know and be sure.)

The overarching theory is of course Shannon. I highly suggest that anyone interested read his original paper. It is perhaps my favourite scientific paper. Well written, groundbreaking, and one of the most important of the last century.

There are a lot of subtleties in the paper that go unappreciated.
The core point is signalling in a noisy channel. Every communication channel is a bandwith limited channel with a defined signal to noise ratio. The maximum information transfer rate is exactly determined by this pair of metrics. However SNR need not be constant over the channel. It is the area under the curve of SNR/frequency that determines the total information capacity. (This is the key insight that allows ADSL to work.)
It is perfectly possible to resolve signal that is well below the noise. Humans can, and radar and sonar systems exploit this to the hilt. There is signal below tape hiss. Detecting it is covered by Shannon's theorem. They key is that you need time to get enough information into the system to allow a signal to be found. And there is no free lunch, you can't detect arbitrary signals, the signal must itself have a low information content, and overall the books will balance. Signals that are self correlated are detectable, since self correlation limits their information content. Eventually you exchange ability to detect signal in one part of the band with ability in another. But it can be done any way you like by perturbing the overall SNR/frequency curve, so long as the area remains constant. The bit depth does not automatically set the signal dynamic range in the channel. It constrains the overall information content, but not the dynamic range at any one frequency.

If we note that the human ear has poor SNR at high frequencies - say the top octave, and certainly above 15kHz, there is a lot of wasted channel capacity. There is no chance any human, even with perfect hearing, can hear down to the LSB at 15kHz. Zip. Even the lowest couple of bits. So, in the digitisation process we can exchange SNR across the audible band. And this is exactly what all the various audio dither mechanisms do. They effectively perturb the dither in such away that the SNR in the human ear critical mid bands is better than 96dB, whilst it is reduced in the higher frequencies.
You can be sure that nearly every CD you ever bought has been so dithered. Back in the early days companies were very proud of their dither algorithms, and sometimes named them on the liner notes. Every DAW you can buy has a final dither step, and usually a choice of dither algorithms.

What is critical is this, dithering is pretty much a final step. It is applied in a DAW when the entire production and mastering is complete and the result is created for distribution. Typically the internal 24 bit samples are being truncated to 16 bit. Again, even if no specialised dither is applied, you cannot truncate the bits from 24 to 16. You get quantisation artefacts. In exactly the same way as ADCs must dither, truncation must dither.

Whilst shaped dither is related to noise shaping in DACs, as both are described by the same theory, the implementation and effects of the two are quite different. Dither results in in-band shaped noise. But the shaping is controlled so that it remains inaudible.

Internal to DAWs there is a lot of care and management needed in the digital chain. In the ideal case there is never any truncation or re-sampling applied to a signal. But if there is it is critical that the algorithms do not lose the below LSB signal. Different plugins can be poorly behaved in this manner, and silently introduce quantisation noise into the chain. A 16 bit audio signal commonly has information below 96dB at some frequencies, and the DAW must ensure that is preserved.

This is also important when considering the ultimate quality of a DAC. A DAC that shows 96dB SNR does not actually provide perfect reproduction of CD quality music. Typically there is another 6 odd dB of real signal in the mid bands. Any DAC aspiring to be CD transparent needs to hit over 100dB SNR in the mid bands.

I know it's only May, but this will be one of the messages of the year!