16 bit to 24 bit.....What’s the other 8 bits?

Im convinced waveform accuracy is only indirectly linked to dynamic range, and that comparing potential distance till the noise floor misses a big part of the story. Its the reason vinyl with its theoretically higher noise floor performs so well with harmonics. Its why single bit systems like SACD also succeed.

I don't have more goods than that at the moment so i'll step back. Think about that rollercoaster ride, though. The harmonic uses the resolution it has in all four quadrants - and at the zero crossing. But to that harmonic, the zero crossings happen just about everywhere BUT where sine waves cross zero. What does it mean when when the lowest DB output of that harmonic is near peak for the system, and vice versa when the harmomic peak is EXACTLY the same as zero for a sine wave?

 

You are entitled to any conviction you like, but fact is that dynamic range is another word for the inherent "error" of the system. Take the original waveform, take the waveform after it has passed the system you are measuring, substract the two from each other. What you get back is a waveform that shows nothing but the errors that system has introduced while doing its thing, and all the errors. Or in your word, how exactly the waveform is painted. The magnitude of that error, whatever shape it might have, is measured by the figure called dynamic range.

In any system worth its money, that error should be nothing but white noise, preferrably way down. Guess what it is with CD.

 

Here's a mind experiment - 16 bit audio has 65535 possible sample values. To represent that visually in "dots" or "steps", you'd have to stack 31 4k TVs one on top of each other to see all the possible sample values as pixels.

Using incorrect analogies, as you propose to elaborate on further, does not illustrate the true principles of digital signal encoding.

For an example where simple explanations fail to describe digital audio: utilizing noise-shaped dither, the only limitation is the quantization noise floor, not the often specified dynamic range of a particular bit depth. In fact, just to prove a point, I took this little 32-bit-rendered composition of mine, normalized it to just under 0dB, and then reduced the volume by 96dB. All audio is below -96dB, which should exceed the dynamic range capabilities of 16 bit audio, we are told. Then I downsampled to 16 bit audio anyway with gaussian noise-shaped dither, which you might think would result in no sound or some crude 1-bit buzzing.

Then boosted the volume back up so you could hear it (to -23dB gain from the original).

Original (gain adjusted to match output): in-23.75db.flac
<-96dB 16-bit audio: out-23.75db.flac

If I were to tweak a bit more, I could make the noise less objectionable, but you can see that the audio is still there and its character is not affected.

 

Proof?

 

Physiology. Exceptions occur, and these are (very) few.

 

No, but changes in society is that fast. Men had been traditionally exposed to more industrial and urban noise, and war, until the last 30 or 40 years. What includes women today is that they are exposed to all of that, and more. It used to be that more men smoked tobacco, which directly affects hearing. Now, women smoke just as much, if not more. Modern factors include nightclubs and personal stereos, and car stereo. There is an equal opportunity for women to suffer hearing loss as men are. We now live with a few generations of men who have never been in a war or have been exposed to industrial noise. It evens the playing field.

While it is true that BOTH men and women have reduced hearing as they age, there are no studies that I am aware of that take heredity, health, or prior exposure into account. This is all why I continue to question the old belief that women hear better, and that old men cannot hear past a certain frequency range. There are just too many variables, and incomplete studies to conclude what you guys keep repeating. And, as long as there are people like your host who may claim to still hear maybe 15 or 16 kHz who are in their 60s, I will question the stuff you all keep parroting. I grew up military, My father served in two wars, He lived to be 80 with good hearing. My mother also had good hearing. I am in my mid-50s. I can still hear 16 kHz on a good day.

 

Ah I didn't get that you objected about the sex differences. I don't know about that and make no claim about it. All I'm saying is that according to all literature on the subject from any era, the top limit is about 20 kHz and declines from there through age. So if we design audio systems to reproduce up to 20 kHz we're good for all humans of all ages and sexes.

 

That's a lot better than people trying to say that all men have bad hearing as they grow old, and that all women hear better than men. What you posted is much more reasoned, and something I agree with.

 

The fact that your experiment produces ANY sound at all is evidence that looking at CD as 95db of dynamic range is FALSE. As I said, you cannot apply a simplistic sine wave model to analyze 16 bit behavior. If it were true, that any signal has 95db of range then kaput, you should have heard silence with your experiment.

 

I will see what I can get. In the meantime; old men hear like old men, except audiophiles.

 

If anything it would show that CD is even better than you claim. It doesn't though. His trick is noise shaping, which means that the inherent noise is moved from the mid-range - where human hearing is most sensitive and most of the music happens - to the highs where human hearing is less acute. The higher tones are drowned in even more noise in order to lower the noise floor where it most matters, but the average noise floor over the whole spectrum is still at -96dB. Without the noise shaping trick, we would probably only have heard white noise.

That term is a red herring. No-one is talking about "sine waves", the model works for all kinds of signals.

 

All I'm saying is 24 bits is better than 16 bits, and that quoting dynamic range for a sine wave is the red herring. What we actually hear with 44/16 is a complicated thing.

I also think this sort of noise floor approach leads to conclusions like why vinyl can't possibly be as good as CD, where I feel the truth is vinyl is doing a better job with complex waveforms than CD.

I've concluded that the 44.1khz is a non issue - I sincerely doubt anyone hears differences between that and 96khz or higher. But I can hear a world of difference between 16 and 24 bits, and no one yet, including me, has stated the concrete theory as to why. I suspect its all about resolution at zero crossing, but its a a theory right now.

 

No-one is disputing that. The point is that when you get above what humans can hear, or what is practical in a normal living room, the additional effort and cost (storage space, bandwidth, hardware, compatibility issues) is pointless.

Well it can't, that is a given.

CD handles any signal more truthful than vinyl. That is not a debatable point, it's a testable fact.

I don't think you can. When comparing different commercial releases you hear the different mixing and/or mastering, regardless of the bit depth. The conclusion that the differences you hear are due to bit depths is wrong.

If you had a 16bit and a 24bit recording positively made from the same master (or the 16 bit one downconverted from the 24 bit one), you wouldn't hear any difference. Except a slightly higher noise floor on 16 bits, but that would be discernable only on quiet passages and only when the volume was cranked up way higher than you or your neighbours could stand. No other differences.

 

I find this part of the discussion between @Lemon Curry and @anorak2 very interesting.

Lemon Curry, here's why I am inclined to agree with anorak and disagree with your claim that bit-depth is about something other than, or more than, dynamic range:

If increased bit-depth were truly about "higher resolution" in the way you say - in other words if it were not simply another way of describing dynamic range - then 24-bit music would be more accurate than 16-bit in a way that would manifest itself as something other than decreased noise floor. The only possible other way it could manifest would be as harmonic distortion, or I guess as inferior frequency response: in other words, the 16-bit version would have less fidelity to the original, manifesting some form of distortion or other deviation from the original.

It's sometimes hard to get our heads around, but these types of flawed fidelity simply are not how analog-to-digital conversion and digital-to-analogue conversion work. The inaccuracy, the lack of resolution, the quantization error, manifests itself as noise, not distortion, not inferior frequency response, not poor stereo imaging, not blurred transients, or any of those analogue phenomena.

If your theory about bit-depth being about "resolution that's different than dynamic range" were correct, then SACD/DSD (and the 100s of millions of delta-sigma DACs out there) would be impossible: DSD uses 1-bit sampling, at a very high sample rate, with massive amounts of noise shaping. The native dynamic range of 1-bit digital is about 6.2dB I believe - and yet even with that, the analogue signal is still accurately captured, and the noise shaping moves it up out of the audible range - and SACD format never has audible distortion, graininess, or whatever compared to the much higher bit-depth of CD format.

 

Except for one flaw.
As long as that 500hz signal and 5khz signal is within the bandwidth of 16/44.1, (which it obviously is) it will still be captured flawlessly.

24 bit will also capture it flawlessly and exactly the same.
There is no such thing as "Deeper" resolution that 16 bit can not capture.
Any signal will be captured fully, as long as it is within about 20-20khz and from Full Scale (Zero) loudness down to about -90 db.

More "Dots" as the term you use, signify higher frequencies in the horizonal, and dynamic range in the vertical scale.
Since 16/44.1 encompasses the entire range any human can hear and a bit of a safety zone more, there can be no more resolution.

The bits apply to ALL SOUND Waves, whether simple or very complex.
Sorry, but this is digital sound basics 101.

 

I am of the thinking, that not all have complete understanding of how the basics of digital sound work totally.
That is perhaps what started many of the "ideas" that 24 bit could resolve more minute information, or capture more detail, when in reality, its barking up the wrong tree.

If there is an advantage to 24 bit in the home playback environment, (which I think there "may be"...) it is related to possible differences in DAC's, or something to do with things unrelated to bit depth or sampling rate directly.
I am not sure, as mastering is a far bigger variable...

 

Facts, 24 bits in studios makes sense. Because any digital processing or workstation options work better and more transparent in 24 bit than in 16 bit, and with less artifacts even when dithered down to 16 bit for final release formats.

 

The nature of SACD is what makes it immune, in my estimation, from some of the same issues that affect CDs.

CD has a 16 bit word that describes the amplitude of the signal at a particular moment. SACD describes only the relative change from the current position: one bit up or down, and when that occurs in time. The concept of "zero crossing" has no meaning in SACD. I think this is a brilliant approach. It also, of course, makes the low pass filtering easier also so the recording is less inclined to have nyquist-related noise problems.

In my mind, a SACD "draws" a waveform more like vinyl does.

 

I engineered early audio sound cards back in the 80's and early 90's, so well aware of digital sound basics

 

Your statement "old men hear like old men" means absolutely nothing.

 

Yes, but in reality, CD and SACD reproduce waveforms exactly the same, A vinyl waveform seen on a scope, is different and not the same as either digital system you mentioned.
The vinyl waveform actually is less accurate and has noise and distortion along with the waveform, along with mild wow and flutter, the waveform visibly "moves" around a bit.
I have seen both CD and vinyl on a scope, and can tell you first hand, although I love my vinyl, it is a mildly flawed analog representation, and is not ideal like the same waveform on 16/44.1

 

Yet you are repeating one of the typical "myths" between 16 and 24 bit digital sound.
The idea that 24 bit somehow can resolve more detail or creates a waveforms better?

 

Take a test mentioned in another thread & post your report...

 

Yes, there is a difference with digital audio. For most it cannot be intuitively understood. There is no free lunch, it takes a lot of studies.

 

Of course it does. It means that less hearing of e.g. high frequencies is an aging phenomena.

I'm not saying for most it's a really big problem, but it means we cannot hear over tones and sharp edges in music anymore.