Bybee's Lament

[Fungus]

I have done a few in my time and failed them all even between sacd and cd

That's because SACD is audiophoolery.

CD's 41kHz@16bit is about as much as humans can hear, there's simply no need for more.

You could maybe argue the case for 48kHz@16bit but not for 96kHz or 24 bits.

(for playback purposes - mixing/recording is another matter)

[Fungus]

The most given arguments are that humans have no audio memory to compare two sources in different times.

So how do golden ears know the magic stones have improved the sound?

[Fungus]

That's because SACD is audiophoolery.

Worse that that, it's a con.

SACDs have normal CD audio on the same disc, the high-def SACD audio is in a separate layer of metal underneath that.

I once ripped the ordinary CD audio off a SACD disc because it was supposed to be a new MFSL "gold" remaster using all sorts of fancy equipment to get the best possible sound. I wanted to see if it was better than the standard CD.

Guess what? It was massively low pass filtered. Not subtly, massively...

I'm guessing there's a button on the SACD player to switch between CD and SACD audio so they low pass filtered the CD version to make the SACD sound better when the salesman presses the button "for comparison".

Here's a GIF of the two versions:

[grumpydoc]

I have done a few in my time and failed them all even between sacd and cd

That's because SACD is audiophoolery.

CD's 41kHz@16bit is about as much as humans can hear, there's simply no need for more.

You could maybe argue the case for 48kHz@16bit but not for 96kHz or 24 bits.

(for playback purposes - mixing/recording is another matter)

The argument for faster sample rates is for easier filter topology (3dB/octave from 20ish kHz rather than needing a brick wall anti-alias filter). I don't think there's any argument for 24bits playback.

[madires]

CD's 41kHz@16bit is about as much as humans can hear, there's simply no need for more.

You could maybe argue the case for 48kHz@16bit but not for 96kHz or 24 bits.

(for playback purposes - mixing/recording is another matter)

Wasn't that 44.1kHz? And for some music, especially classic, 18 or 20 bits would be great.

[Fungus]

You could maybe argue the case for 48kHz@16bit but not for 96kHz or 24 bits.

(for playback purposes - mixing/recording is another matter)

The argument for faster sample rates is for easier filter topology

Sure, hence the case for 48kHz - to allow some wiggle room in less-than-ideal hardware reconstruction filters. 

OTOH this is the 21st century and we live in a world of $2 audio DACs with integrated DSPs. If the DSP inside the DAC can do sin(x)/x reconstruction on the 44.1kHz data then higher bitrates simply aren't needed. You have a perfectly reconstructed signal inside the chip which can be 'resampled' methematically at any bitrate you or your output hardware could possibly desire.

nb. For this to happen the signal on the CD needs to bandwidth limited to about 20kHz at the mixing/mastering stage (hence the justification for higher bitrates there).

[polli]

for some music, especially classic, 18 or 20 bits would be great.

The quieter parts might need more bit depth if they were on their own, but if the whole piece is normalized, at what volume are you listening that you can hear the quantization on the low volume parts?

[Fungus]

CD's 41kHz@16bit

Wasn't that 44.1kHz?

Oops, brain fart...   

And for some music, especially classic, 18 or 20 bits would be great.

Time to repost this? 



(bonus: Lots of analog test gear porn...)

[Cerebus]

The most given arguments are that humans have no audio memory to compare two sources in different times.

We don't have a very good color memory, either. You can perceive subtle differences in color (each Munsell color coordinate is perceptually distinct) but remembering specific colors over any timespan is nigh impossible. The same is likely to be true for all subjective experiences.

I think this [colour memory] is one of those things where there is a lot of variation in human abilities and, as with pitch, some outliers where there are some people with uncannily accurate colour memory, just as there as some people with perfect pitch. I used to know a printer who could, maybe 7 times out of 10, name a Pantone colour just from looking at a sample swatch. If he didn't get it exactly right he'd still have named one of the adjacent colours in the Pantone book.

[Cerebus]

for some music, especially classic, 18 or 20 bits would be great.

The quieter parts might need more bit depth if they were on their own, but if the whole piece is normalized, at what volume are you listening that you can hear the quantization on the low volume parts?

If you hear rustling in the bushes behind you sometime in the next few days, it's a classical music recording engineer sneaking up behind you to garotte you with a mic cord because he's taken deep offence at the suggestion that a classic recording engineer would ever normalize the levels on a recording. 

I can't think of an actual piece where this happens, but it's quite plausible that there exists a piece of music that has both a solitary piccolo solo and full orchestra playing fortissimo [For the non music readers that's the Italian notation for very loud]. Under those circumstances it's quite plausible that the noise floor becomes audible during the piccolo solo.

Lets see how that pans out in numbers.

You can debate the dynamic range of human hearing, but it's at least 120 dB. The dynamic range of a full orchestra is about 50 dB (again, open to debate). Domestic background noise when it's 'quiet' is usually around 40 dBA SPL. 16-bit SNR is 96 dB, 24-bit SNR is 144 dB.

So 16-bit with a 50 dB orchestral dynamic range would leave that theoretic piccolo 46dB above the recording noise floor, for 24-bit it'd be 94 dB above the noise floor. It's quite clear that 94 dB SNR is a non-starter, you'd never hear the noise. What about 46 dB SNR? If you set up so that the piccolo was clearly audible, let's say 60 dBA SPL (quiet speech) the noise floor of a 16-bit recording would be -26 dB to the noise floor of the listening room and the full orchestra would be at 110 dBA SPL, just 10 dB shy of most people's auditory pain threshold.

Conclusion: 16-bit is adequate for even the widest dynamic range of orchestral music when listened to in domestic surroundings with 40 dBA SPL background noise. Even in a music studio listening room (typical background 20 dBA SPL) it would still be acceptable. If you wish to listen to full dynamic range orchestral music in an anechoic chamber, then you will need 24-bit encoding.

[Alex Nikitin]

Conclusion: 16-bit is adequate for even the widest dynamic range of orchestral music when listened to in domestic surroundings with 40 dBA SPL background noise. Even in a music studio listening room (typical background 20 dBA SPL) it would still be acceptable. If you wish to listen to full dynamic range orchestral music in an anechoic chamber, then you will need 24-bit encoding.

And IMHO it is a wrong conclusion, at least for the "CD standard" 44.1kHz sampling rate. We are not normally using audio equipment to listen for some noise or continuous tones, we usually listen to music, human voice or a combination of both (I am discounting special effects in movies as these can be generally put into the "noise" category). It appears however that we use "noise, distortion, dynamic range" as relevant values where in practice they are not the most relevant, otherwise I wouldn't be able to hear more music (meaning more details, more ambiance)  from a good cassette recording (inferior on all these points) than from a (properly made) CD recording (assuming a quality hi-res or analogue master recording is used to create both). For me it means that we are missing something in the usual way we evaluate the audio equipment quality.

Cheers

Alex

[newbrain]

Conclusion: 16-bit is adequate for even the widest dynamic range of orchestral music when listened to in domestic surroundings with 40 dBA SPL background noise. Even in a music studio listening room (typical background 20 dBA SPL) it would still be acceptable. If you wish to listen to full dynamic range orchestral music in an anechoic chamber, then you will need 24-bit encoding.

And that conclusion is drawn with the very conservative 96dB estimation (6dB/bit) for the noise floor.
If (shaped) dithering is used, that figure goes up a lot, allowing resolve signals with amplitude of 1/4 of a bit.

From Monty's video:
https://youtu.be/cIQ9IXSUzuM?t=814

[Cerebus]

Conclusion: 16-bit is adequate for even the widest dynamic range of orchestral music when listened to in domestic surroundings with 40 dBA SPL background noise. Even in a music studio listening room (typical background 20 dBA SPL) it would still be acceptable. If you wish to listen to full dynamic range orchestral music in an anechoic chamber, then you will need 24-bit encoding.

And that conclusion is drawn with the very conservative 96dB estimation (6dB/bit) for the noise floor.
If (shaped) dithering is used, that figure goes up a lot, allowing resolve signals with amplitude of 1/4 of a bit.

From Monty's video:
https://youtu.be/cIQ9IXSUzuM?t=814

To be more precise, I'm assuming an Effective Number Of Bits of 16-bits. ADC/DACs being what they are, and data sheet specsmanship being what it is, most 'audio' ADC/DACs are a good couple of bits worse than what they claim. To be realistic, you'd have to run my numbers again substituting something closer to the ENOB actually achieved.

[Fungus]

To be more precise, I'm assuming an Effective Number Of Bits of 16-bits. ADC/DACs being what they are, and data sheet specsmanship being what it is, most 'audio' ADC/DACs are a good couple of bits worse than what they claim. To be realistic, you'd have to run my numbers again substituting something closer to the ENOB actually achieved.

Yep. There's nothing technically wrong with using 192kHz @ 24bit for recording/mastering.

(although finding a microphone with that much sensitivity will be challenging)

[Kjelt]

I have done a few in my time and failed them all even between sacd and cd

That's because SACD is audiophoolery.
CD's 41kHz@16bit is about as much as humans can hear, there's simply no need for more.
You could maybe argue the case for 48kHz@16bit but not for 96kHz or 24 bits.
(for playback purposes - mixing/recording is another matter)

Well there have been new developments (couple years back)  backed by experiments and published by the AES that (sample)time is crucial for pinpointing the source of a sound signal and that humans are able to go as low as 10us. That would mean a samplerate of 200kHz not for frequency but for correct timealignment.
Part of this is now used for MQA which I personally find also audiophoolery since they use filtering and folding and copy protection so they can get license fees for their drm while a native wav sample of 255kHz would be far superior but that aside.

[schmitt trigger]

OTOH this is the 21st century and we live in a world of $2 audio DACs with integrated DSPs.

In 1979 there was a song called "video killed the Radio star", by the one-hit-wonder group The Buggles.

The parody most likely today would be named:  "MP3 killed the SACD"

[Mr. Scram]

There is an obvious reason quantum physics and effects are mentioned so often in product descriptions concerning these kinds of products. It's how they make something that's amazing to some, but sucks to other. It's Schrödinger all over again.

[f5r5e5d]

Well there have been new developments (couple years back)  backed by experiments and published by the AES that (sample)time is crucial for pinpointing the source of a sound signal and that humans are able to go as low as 10us. That would mean a samplerate of 200kHz not for frequency but for correct timealignment...

not right, 10 us iatd resolution has nothing to do with hearing or needing to record > 20 kHz or sample at more than just adequate by Nyquist rates

it is a remarkable result but it relies on the power of our neural nets for doing correlation of entirely 'in band' 'conventional audio frequency' 20 - 20 kHz

as for encoding fractional sample time differences in digital audio, Monty shows, and its trivial to do yourself with a Spice with wavefrom analysis - make a raised sin envleoped tone burst, define a 2nd with fractional sample delay, pass thru .wav i/o @ 16/44 - you can then see in the fft analysis nanosecond resolution of the tone fundamental phase delay after the 16/44 step

[Cerebus]

And IMHO it is a wrong conclusion, at least for the "CD standard" 44.1kHz sampling rate. We are not normally using audio equipment to listen for some noise or continuous tones, we usually listen to music, human voice or a combination of both (I am discounting special effects in movies as these can be generally put into the "noise" category). It appears however that we use "noise, distortion, dynamic range" as relevant values where in practice they are not the most relevant, otherwise I wouldn't be able to hear more music (meaning more details, more ambiance)  from a good cassette recording (inferior on all these points) than from a (properly made) CD recording (assuming a quality hi-res or analogue master recording is used to create both). For me it means that we are missing something in the usual way we evaluate the audio equipment quality.

Cheers

Alex

Oh Alex, now you've gone and torn it. I was with you on the whole Double Blind thing, I got what you were on about even if it seems to have passed over the heads of a number of people.

But now I think that you're heading off into the audio la-la land of the purely subjective. As a man who used to sit somewhere between a dirty great 24 track Studer A800 and a thumping great pair of Lockwood monitors and thus has quite some experience of really, really listening to what is going on I have to take issue with the claim that a cassette tape recording somehow contains "more music (meaning more details, more ambiance)" than a CD. I've listened to the 'detail' disappearing from what I heard in the studio as it's gone off to the pressing plant (yup, my era was vinyl and a completely analogue signal chain*) or cassette duplicating plant. Beyond a shadow of a doubt my impression is that a well recorded CD retains much more of the character of what I've heard first hand than a well recorded cassette ever could. Yes, that is a subjective judgement, but it's based on experience of hearing every part of the reproduction chain from the actual music, through the reproduction of that in the monitoring room, the master tape, the mixed master tape and the finished LP/cassette.

And "ambience"? There's an airy fairy audiophile word if ever I heard one. Unless one means it in the strict sense of 'ambient sounds' it's a pretty useless term that seems almost to be picked for the difficulty in pinning down what it means. If it is meant in the strict sense then it's very difficult to see how 'ambient' and 'direct' sounds are conceivably processed differently by a cassette or a CD. How will a CD "know" that a particular sine wave is 'ambient' or 'direct'? The only place in the reproduction chain where there is a real difference in ambient and direct sounds is at the microphone.

I'm sorry if that all sounds a bit harsh, but you can't flip from raising (entirely justifiable) concrete, quantifiable, concerns about objective test methodologies to using phrases like "more music" to support a position. The subjectivity of latter completely undermines your taking an objective point on the former.

*We had one bit of outboard equipment that was digital, an Eventide Harmonizer.

[Alex Nikitin]

And IMHO it is a wrong conclusion, at least for the "CD standard" 44.1kHz sampling rate. We are not normally using audio equipment to listen for some noise or continuous tones, we usually listen to music, human voice or a combination of both (I am discounting special effects in movies as these can be generally put into the "noise" category). It appears however that we use "noise, distortion, dynamic range" as relevant values where in practice they are not the most relevant, otherwise I wouldn't be able to hear more music (meaning more details, more ambiance)  from a good cassette recording (inferior on all these points) than from a (properly made) CD recording (assuming a quality hi-res or analogue master recording is used to create both). For me it means that we are missing something in the usual way we evaluate the audio equipment quality.

Cheers

Alex

Oh Alex, now you've gone and torn it. I was with you on the whole Double Blind thing, I got what you were on about even if it seems to have passed over the heads of a number of people.

But now I think that you're heading off into the audio la-la land of the purely subjective. As a man who used to sit somewhere between a dirty great 24 track Studer A800 and a thumping great pair of Lockwood monitors and thus has quite some experience of really, really listening to what is going on I have to take issue with the claim that a cassette tape recording somehow contains "more music (meaning more details, more ambiance)" than a CD. I've listened to the 'detail' disappearing from what I heard in the studio as it's gone off to the pressing plant (yup, my era was vinyl and a completely analogue signal chain*) or cassette duplicating plant. Beyond a shadow of a doubt my impression is that a well recorded CD retains much more of the character of what I've heard first hand than a well recorded cassette ever could. Yes, that is a subjective judgement, but it's based on experience of hearing every part of the reproduction chain from the actual music, through the reproduction of that in the monitoring room, the master tape, the mixed master tape and the finished LP/cassette.

And "ambience"? There's an airy fairy audiophile word if ever I heard one. Unless one means it in the strict sense of 'ambient sounds' it's a pretty useless term that seems almost to be picked for the difficulty in pinning down what it means. If it is meant in the strict sense then it's very difficult to see how 'ambient' and 'direct' sounds are conceivably processed differently by a cassette or a CD. How will a CD "know" that a particular sine wave is 'ambient' or 'direct'? The only place in the reproduction chain where there is a real difference in ambient and direct sounds is at the microphone.

I'm sorry if that all sounds a bit harsh, but you can't flip from raising (entirely justifiable) concrete, quantifiable, concerns about objective test methodologies to using phrases like "more music" to support a position. The subjectivity of latter completely undermines your taking an objective point on the former.

*We had one bit of outboard equipment that was digital, an Eventide Harmonizer.

I can understand your position very well, however I also have many years of experience in the sound equipment design, as well as in the sound recording and reproduction, and what I say, I am not saying lightly. I wish (with all my engineering background and experience) that I could make more sense out of it and find a way to measure directly what my ears tell me, however I've found so far no way to do it. On the other hand, if I would not follow my ears and my "instincts" on the sound quality, I would never be able to design some pretty decently sounding electronics. I've spent countless hours trying to measure things and even more time trying to make sense out of these measurements. But at the end it was a simple choice - either I do follow what I hear with a hope to get a good result, or I don't - and make useless crap.

And I fully agree, that the majority of cassettes, especially mass produced, did sound pretty bad. You are quite welcome to visit my home and listen to a cassette made right  .

Cheers

Alex

P.S. - I agree that the term "ambience" is ambiguous, and would for the purpose of this discussion to define it as background sounds which can be easily masked by loud sounds.

[Cerebus]

And I fully agree, that the majority of cassettes, especially mass produced, did sound pretty bad. You are quite welcome to visit my home and listen to a cassette made right  .

It's OK I don't need to, got rid of the Nakamichi years ago (when it was still worth something) and somewhere, under a layer of dust, there's still a quite decent 3 head Teac that hasn't been used in more years that I care to remember. I might drop in sometime to compare our ideas of what 10V looks like, but that's a completely different topic.

OK. Here's what I don't get. How can you design something to 'sound right' if you don't know what, technically, 'sounds right' equates to? Randomly drop components in, randomly adjust values, randomly substitute semiconductors? At what point does that become indistinguishable from sacrificing chickens and goats?

I do understand that the latter is standard, accepted RF design methodology, as are the chanting and the black robes. 

[Kjelt]

Well there have been new developments (couple years back)  backed by experiments and published by the AES that (sample)time is crucial for pinpointing the source of a sound signal and that humans are able to go as low as 10us. That would mean a samplerate of 200kHz not for frequency but for correct timealignment...

not right, 10 us iatd resolution has nothing to do with hearing or needing to record > 20 kHz or sample at more than just adequate by Nyquist rates

it is a remarkable result but it relies on the power of our neural nets for doing correlation of entirely 'in band' 'conventional audio frequency' 20 - 20 kHz

as for encoding fractional sample time differences in digital audio, Monty shows, and its trivial to do yourself with a Spice with wavefrom analysis - make a raised sin envleoped tone burst, define a 2nd with fractional sample delay, pass thru .wav i/o @ 16/44 - you can then see in the fft analysis nanosecond resolution of the tone fundamental phase delay after the 16/44 step

So the whole multimillion dollar MQA story that they need parts of the 255kHz sampling in order to reproduce the original sample timing within 10us  is a hoax?

[Alex Nikitin]

And I fully agree, that the majority of cassettes, especially mass produced, did sound pretty bad. You are quite welcome to visit my home and listen to a cassette made right  .

It's OK I don't need to, got rid of the Nakamichi years ago (when it was still worth something) and somewhere, under a layer of dust, there's still a quite decent 3 head Teac that hasn't been used in more years that I care to remember. I might drop in sometime to compare our ideas of what 10V looks like, but that's a completely different topic.

OK. Here's what I don't get. How can you design something to 'sound right' if you don't know what, technically, 'sounds right' equates to? Randomly drop components in, randomly adjust values, randomly substitute semiconductors? At what point does that become indistinguishable from sacrificing chickens and goats?

I do understand that the latter is standard, accepted RF design methodology, as are the chanting and the black robes. 

Who said anything about Nakamichi (or Teac for that matter)?!

And on the design side - let's say, it requires a combination of knowledge, experience, good ears and a sizeable amount of luck. And all the measuring equipment you can afford, borrow or steal .

Cheers

Alex

[Cerebus]

And all the measuring equipment you can afford, borrow or steal .

And there's the nub. If you don't know what 'sounds right' equates to in objective terms, what good is the measuring equipment doing you? (Beyond allowing you to design for mundane things like distortion, transient response etc. etc.)

[Alex Nikitin]

And all the measuring equipment you can afford, borrow or steal .

And there's the nub. If you don't know what 'sounds right' equates to in objective terms, what good is the measuring equipment doing you? (Beyond allowing you to design for mundane things like distortion, transient response etc. etc.)

Measurements are essential. Practically all changes can be traced to a measurable difference, and especially in production it is important to be consistent. When a design is competed, accurate and extensive measurements allow for a tight quality control. Without measurements you are blind and it is not possible to rely only upon your ears all the time. You learn to check and double check your listening experience, but without measurements it is easy to make a crap design from a technical point of view. I always put the technical excellence first - unfortunately, it is not enough to produce a good sound. So normally you get the measurements right and only then "tune" the design by ear. And after that you measure the differences (usually small) and do a second round of listening. And third, and so on. You can easily make a mistake in one session, for a variety of reasons, but usually a long term listening, in various situations and conditions, is the best bet to get things right. 

Cheers

Alex