Is the dynamic range of CD sufficient?

[Julf]

A story about a middle eastern country having a very long bridge built in the mountain which was designed I think by a French architect. The day came to test it, he made the French architect stand under it as the cars went by!!! Now that is one way to make sure there is extra safety margin in the bridge.

An excellent policy - and pretty much similar to how aircraft and ship designers used to have to be on the maiden flight / voyage...

I cross over the new version of this bridge every week or two and I am always reminded of the demise of the original

Ouch, yes, the Tacoma bridge resonance was pretty extreme!

Then there is the car-burning building in London...

 

[Phelonious Ponk]

I don't think anything has devolved here rbert. This discussion, predictably, seems to have expanded from dynamic range to, as you said, "whether 16/44 is adequate." Though, not to put too fine a point on it, no one is questioning if it is adequate for recording music; we're talking about playback here. Can you point me to any of this good evidence showing that 20khz will not replicate live music for a significant number of people? Statistically, I wouldn't expect that the number of people measured to hear above 20khz is significant. That they've not only been found in significant numbers, but had their perceptions of music at and above 20khz tested would be quite surprising.

Nice anecdote about headroom, Amir. I believe in headroom, in over-engineering, but in this case, it seems clear that it does harm. Did your previous thread on this subject that rbert referred to above address that issue? Did you conclude that the good of having some extra headroom to operate in outweighed the IMD caused by the ultrasonic information? Could you point me to it?

Elcorso -- so filtering FR out above 22khz distorts low frequencies? How? What's that sound/look like? Surely it has been measured?

Tim

 

[Al M.]

I believe the premises are poorly raised. It isn't only about human hearing capabilities, but about horrible brickwall filters in the regular CD format.

You sound like it's still 1984. I don't think there is 'horrible brickwall filtering' in 16/44 anymore. This is taken care of by upsampling/oversampling, see:

http://www.audioholics.com/audio-technologies/upsampling-vs-oversampling-for-digital-audio

Look at figure 3 and read the explanatory text; the subsequent figures are also instructive.

 

[JackD201]

I think it is pretty obvious that 16bit is "adequate" for recording music. Depending on the music 8bits is "adequate" (video games)

Thing is............recording has been done at 24bits for over a decade now (inching closer to 2 decades) for reasons already clearly established on this forum and elsewhere. Same goes for tape transfers. So we go back to what a previous poster (I think it was Amir) said. If you can get a copy of the original and the cost of storage isn't an issue, would you wan the copy or a downconverted version, the downconversion of which could have been botched? I know what my answer is. Just give me the original. Storage isn't that expensive anymore barring another flood in Thailand or anywhere else drives are made. Now, that's not to say everyone should buy the 24bit or invest in whatever is required to play it. That is a choice everybody is free to make. Just remember that folding it down can make it undistinguishable at best, but never better.

What wouldn't make sense is buying for a premium a 24bit version of something that was recorded at 16bit.

 

[microstrip]

I believe the premises are poorly raised. It isn't only about human hearing capabilities, but about horrible brickwall filters in the regular CD format. If you truncate so heavy it will alter lower frequencies and music integrity. (...)

Surely. People often forget that what we listen to is digitized music being played through electronic instruments. If using the extra dynamics (the original subject of the thread) or bandwidth designers can make better sounding electronics we will simply tell the whole sounds better, irrespective of human limits.

Bandwidth was later added in this thread, probably because there is almost no music in 41/24 - 24 bits is mostly associated with higher sampling rates.

 

[Phelonious Ponk]

I think it is pretty obvious that 16bit is "adequate" for recording music. Depending on the music 8bits is "adequate" (video games)

Thing is............recording has been done at 24bits for over a decade now (inching closer to 2 decades) for reasons already clearly established on this forum and elsewhere. Same goes for tape transfers. So we go back to what a previous poster (I think it was Amir) said. If you can get a copy of the original and the cost of storage isn't an issue, would you wan the copy or a downconverted version, the downconversion of which could have been botched? I know what my answer is. Just give me the original. Storage isn't that expensive anymore barring another flood in Thailand or anywhere else drives are made. Now, that's not to say everyone should buy the 24bit or invest in whatever is required to play it. That is a choice everybody is free to make. Just remember that folding it down can make it undistinguishable at best, but never better.

What wouldn't make sense is buying for a premium a 24bit version of something that was recorded at 16bit.

Again, no one has been talking about recording at 16 bits. This is a discussion of playback media.

So we go back to what a previous poster (I think it was Amir) said. If you can get a copy of the original and the cost of storage isn't an issue, would you wan the copy or a downconverter version

From the article I linked above:

Neither audio transducers nor power amplifiers are free of distortion, and distortion tends to increase rapidly at the lowest and highest frequencies. If the same transducer reproduces ultrasonics along with audible content, any nonlinearity will shift some of the ultrasonic content down into the audible range as an uncontrolled spray of intermodulation distortion products covering the entire audible spectrum. Nonlinearity in a power amplifier will produce the same effect. The effect is very slight, but listening tests have confirmed that both effects can be audible.

http://people.xiph.org/~xiphmont/demo/intermod.png

Maybe I don't want a down converted version; maybe I want the 24/196 with a brick wall filter. But the "for" side of this conversation seems to be making a "we got the bits, why not use them" argument. Why not has been clearly articulated. Until someone can point to as clear an argument, supported by measurements, that demonstrates that down converting does more harm/is less transparent than the added IMD, that's not an argument at all. It's no more substantive than just saying you like it. Maybe you like the mastering. Maybe you like IMD. It is an utterly insubstantial position.

Tim

 

[Julf]

Surely. People often forget that what we listen to is digitized music being played through electronic instruments.

Indeed. And quite an amount of music from the 80's and 90's that is now considered highlights of the period was done using instruments and systems doing 12 bits @ 32 kHz, if even that...

Bandwidth was later added in this thread, probably because there is almost no music in 41/24 - 24 bits is mostly associated with higher sampling rates.

I have been surprised at how much "studio masters" have been sold (by Linn and others) as 44.1/24 - hopefully not just because a 16 -> 24 bit conversion is very easy to hide.

 

[JackD201]

Again, no one has been talking about recording at 16 bits. This is a discussion of playback media.

Problem is, it's 2014. For 10 bucks your Mac can play 24bit without having to change your library. The majority of stand alone DACs have been 24bit capable for quite a while too. It's not like there is a stiff barrier to entry other than paying a little bit more per album.

Given the situation, other than as an intellectual exercise, who really cares? In my first post in this thread I said "go try it out for yourself". My 24/44.1 of good ol' Bruno Mars IS better than my daughter's CD and my iTunes plus. My 24/88 Daft Punk too. Both NON "audiophile" fare. Is it because it is mastered differently or is it the medium? I don't know. For me what is is more important than why it is. If the CD is better than the LP, I'll listen to the CD and not the LP. I sadly got a cold stamper of Random Access Memories so I listen to the 24/88 more than the LP. It's not very complicated or rather I choose not to make it complicated. I have a lot of fantastic sounding CDs. Unfortunately I have orders of magnitude more crappy ones. The ratio is very different with my 24 and DSD library. Is it because the selections have already been curated? I'm fairly certain that is a big part of it. Again, it is what it is. 16bit is a least common denominator medium. MP3 is 16bit too. That makes it very much less curated. You WILL get more crap because there are just more titles. Even that isn't a problem. Getting a system to sing, whether your barometer is fidelity or something more personalized, is a lot harder to do than to dumb down a system to something safe and passable.

 

[microstrip]

(...) Given the situation, other than as an intellectual exercise, who really cares? (...)

Jack,

WBF members seem to care. Some of them even debate HiRez PCM versus DSD and now some people tell them that even 44.1/16 can not be distinguished from 44.1/24 ...
But as you say, for me it is mainly an intellectual exercise - nice name for an entertainment activity - as 99.9% of the digital music I listen only exists in redbook.

 

[Phelonious Ponk]

Problem is, it's 2014. For 10 bucks your Mac can play 24bit without having to change your library. The majority of stand alone DACs have been 24bit capable for quite a while too. It's not like there is a stiff barrier to entry other than paying a little bit more per album.

Given the situation, other than as an intellectual exercise, who really cares? In my first post in this thread I said "go try it out for yourself". My 24/44.1 of good ol' Bruno Mars IS better than my daughter's CD and my iTunes plus. My 24/88 Daft Punk too. Both NON "audiophile" fare. Is it because it is mastered differently or is it the medium? I don't know. For me what is is more important than why it is. If the CD is better than the LP, I'll listen to the CD and not the LP. I sadly got a cold stamper of Random Access Memories so I listen to the 24/88 more than the LP. It's not very complicated or rather I choose not to make it complicated. I have a lot of fantastic sounding CDs. Unfortunately I have orders of magnitude more crappy ones. The ratio is very different with my 24 and DSD library. Is it because the selections have already been curated? I'm fairly certain that is a big part of it. Again, it is what it is. 16bit is a least common denominator medium. MP3 is 16bit too. That makes it very much less curated. You WILL get more crap because there are just more titles. Even that isn't a problem. Getting a system to sing, whether your barometer is fidelity or something more personalized, is a lot harder to do than to dumb down a system to something safe and passable.

Well, I partly want to answer the question "is it the hi-res or the mastering" out of intellectual curiosity, sure. But let's assume for a moment that it is the mastering. With as few high profile people as there are advocating for quality audio, we have people like Neil Young and audiophiles vigorously advocating for the wrong thing. I view that as a problem. Our audiophile conventional wisdom is that more bits/higher sample rate = better. There seems to be a pretty good chance that we're wrong about that, but we don't seem to want to know. I find that curious. I find the fact that the industry may be pursuing hi-res as the "quality" medium when it's possible that the quality lies elsewhere just plain sad.

Tim

 

[elcorso]

You sound like it's still 1984. I don't think there is 'horrible brickwall filtering' in 16/44 anymore. This is taken care of by upsampling/oversampling, see:

http://www.audioholics.com/audio-technologies/upsampling-vs-oversampling-for-digital-audio

Look at figure 3 and read the explanatory text; the subsequent figures are also instructive.

I'm sorry but graphics means nothing to me, by default I trust my ears. The brickwall filters applied at higher frequencies bring some help thanks to the filters forcibly pushed by upsampling/oversampling, but the problem is the source of the music tracks, like if they were recorded at 16/44.1

There is another article from Meridian's Bob Stuart from 2009;

http://www.theabsolutesound.com/art...audios-bob-stuart-talks-with-robert-harley-1/

"Robert: How much better can 44kHz, 16-bit get? Are we running out of all the performance improvements that are possible?

Bob: It’s really hard to say that it’ll never improve further. In the early days it was about improving the DACs and lowering the jitter. It was only when we were able to apply digital signal processing that we were able to get any more out of CD. From a purely psychoacoustic viewpoint, 16 bits isn’t enough to cover the dynamic range we can hear. By using noise shaping we can get a lot closer.

Most recordings don’t have an inherent dynamic range of 20 bits. In fact, if you go to a venue and look at the noise of the venue and the noise of the electronics and the microphone’s inherent thermal noise, only a few recordings are better than 18 bits. So it’s quite hard to get a signal which covers the dynamic range of human hearing. However, we also know from psychoacoustics that you can hear things below the noise floor, especially if they’re structured, and especially if they’re correlated to certain signals we’re listening to. So down in the threshold area 20 bits are okay, but 24 bits are absolutely much more than we need.

The question is whether the CD coding space [44.1kHz, 16-bit] is good enough. The answer is that it would be nice to have a little bit more, because it cannot transparently bring to a human listener everything that he can hear. In my AES paper on high resolution a few years ago, I asked the question “What are the parameters of a transmission channel that is completely transparent between the performers on a stage of a concert hall and a listener?” The answer is that it’s actually more than 20 bits, and it really has to be wider than 44kHz. "

And at the end of the article:

" Robert: How do you see the future of high resolution?

Bob: It’s incredibly important that you capture the content and archive at the highest possible resolution. Even if it’s going to be delivered on a storage channel like CD, you absolutely should capture it with all the information that we can hear."

Roch

 

[JackD201]

Tim,

Fine, lets complicate things. Mathematically higher bit depth is better because you have less quantization error. Increasing sampling makes the difference between ideal and practical SNR smaller. Mo' is betta. Period.

16bit advocates do not dispute this. What they do say is that anything more than 16bit depth and 44.1k samples is beyond the practical limit because anything else would be inaudible at least for most. Let's take a step back here and ask, is it really? Bit depth allows for more quantization levels between the highest input amplitude (Xmax) and the lowest (Xmin). Xmax in a 24bit device need not be the same as maximum corresponding voltage output of a 16bit device that being levels that would either hurt you or blow up your gear whichever comes first. Most likely max voltage out is the same for both limiting the dB but answering my own question I asked before, apparently amplitude q levels (steps) can be different. It's not as sexy as saying with 24bit you get higher dynamic range (which you can't use anyway) but it does say you get a better approximation of the continuous analog signal discreet in amplitude and time. Call it higher relative resolution or fidelity to the source, in this instance it is the same thing. So forget the extremes in amplitude for now and look at the quantization figures with 16 vs 24 bit in an ADC or DAC with a max of 2v.

Getting more to the point, why would anybody who can get a 24bit recording want to get a down converted version which will have either higher harmonic distortion or higher noise depending on how the down conversion was done because you have now created quantization error? I will guarantee that the answer will be based on practicality rather than what is really better/higher fidelity/higher performance EMPIRICALLY.

 

[Gregadd]

Maybe we should nove on from adequate it to optimum.

 

[DonH50]

Mathematically higher bit depth is better because you have less quantization error. Increasing sampling makes the difference between ideal and practical SNR smaller.

First point is true, I do not understand the second. Are you still talking about increased bit depth or sampling rate? Higher sampling rates often increase the noise (wider bandwidth, more settling artifacts). And what do you define as ideal and practical SNR? Aside: For high-resolution converters, SNR is often set by the analog noise floor rather than quantization noise. Similarly I am not aware of any converter that actually approaches the ideal SFDR in practice; IME distortion and other noise sources generally put the noise floor within 10 - 20 dB'ish of the SNR. That said, I have not looked closely at many audio DAC (chip) specs.

 

[Julf]

Mathematically higher bit depth is better because you have less quantization error.

Yes. Nobody disagrees with that. The issue is that even 16 bits gives you a quantisation noise level that is way lower than the noise level of your source material.

A very wide garage door is useless if your driveway is much narrower.

 

[Julf]

MP3 is 16bit too.

No, it isn't. MP3 has up to 20 bits of effective dynamic range.

 

[rbbert]

Yes. Nobody disagrees with that. The issue is that even 16 bits gives you a quantisation noise level that is way lower than the noise level of your source material..

I think the article Amir wrote and the Bob Stuart interview established that isn't true, as well as the likelihood that not even allowing for a "safety margin" 44.1 sampling probably isn't quite enough...

 

[JackD201]

Yes. Nobody disagrees with that. The issue is that even 16 bits gives you a quantisation noise level that is way lower than the noise level of your source material.

A very wide garage door is useless if your driveway is much narrower.

That is kinda what I said right? No one disputes this.

My question to you is if price or storage space was not an issue, would you take the 24 or the decimated 16?

My buddy Tim, and he really is my buddy I'm not being sarcastic, is a fidelity kind of guy. Whether one can tell the difference or not, what is closer to the continuos analog signal, 16 or 24? I'm not asking what is good enough but what really is.

As for mp3 being 20bit, I'll let J-J give me a good smack on the back of my head for that error. I won't let you though

 

[Atmasphere]

I think the article Amir wrote and the Bob Stuart interview established that isn't true, as well as the likelihood that not even allowing for a "safety margin" 44.1 sampling probably isn't quite enough...

The 44.1 sampling frequency is based on a mis-application of the Nyquist theorem; yes, it is certainly not enough.

 

[Julf]

My question to you is if price or storage space was not an issue, would you take the 24 or the decimated 16?

Well, if storage space and network bandwidth was no issue, I would take the 24 (as I do - 24 bit FLAC on home server, mp3 copies for mobile/portable use).

As for mp3 being 20bit, I'll let J-J give me a good smack on the back of my head for that error. I won't let you though

Ah, yes. More than happy to give JJ that privilege!