24bit vs 16bit, the myth exploded!

[gregorio]

Quote:

Originally Posted by Acix /img/forum/go_quote.gif ... you still believe that the leading expert at ADAM and Yamaha and SPL and all the other manufacturers have just created this range out of thin air to use as a marketing tool.

No, as I understand it, the 24bit 192kFs/s standard was decided upon because it was way beyond what was ever likely to be needed. If you create a standard which is so ridiculously beyond what humans are capable of perceiving that will hopefully result in this format never going out of date or requiring a new format incompatible with old formats. Bare in mind this format was agreed before any converters existed which were capable of 192kFs/s.

ADAM, Yamaha and SPL did not invent the format, as far as I know the first to support this format were DigiDesign. I can't remember off the top of my head the name or composition of the standards organization responsible. I should also mention, that professionally we've been using higher than 16bit for almost 20 years, although it's only relatively recently that it's been called Hi-Res and there has been a consumer demand for it. Now there is a market, I cannot blame Yamaha or ADAM for taking advantage of it, even if the demand is almost entirely superfluous for the consumer. If you are talking about professional studio monitors, then as mentioned previously 24bit is useful for recording and mixing, it's as a playback consumer format that 24bit is irrelevant.

G

 

[JaZZ]

Sennheiser MKH 8000 series:

.

 

[gregorio]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif Sennheiser MKH 8000 series

Yes, those Sennheiser mics look interesting. I haven't had chance to see them yet. They were only released last year and have only become available in the last few months. In fact, AFAIK some of the 8000 series models have still not been released yet. I presume they are designed to replace the MKH40 series, which I have used and which have a maximum frequency of 20kHz, good mics though.

To be honest I'm looking forward to having a play with the 8040 and seeing exactly what their frequency response is. At the moment though I don't have access to one and I don't know anyone who has done more than just seen them at a trade show. Again though, all but the last few kHz of their (quoted) maximum frequency response (50kHz) can be encoded at 96kFs/s. They go nowhere near the 96kHz of 192kFs/s.

As they are so new though, they cannot account for differences that consumers think they hear with existing 192kFs/s recordings.

As I said before, there are a few mics which go up to 40kHz (Schoeps for example) and it looks like there is a new one which might go up to 50kHz. All the most commonly used studio mics though, max out at 20kHz or lower.

You can try nit-picking and finding any scrap of information which may provide some proof that some relatively insignificant detail of what I have stated is disputable. But the main substance of what I have stated is correct, namely: 24bit is a complete waste of time for the consumer and so is 192kFs/s.

G

 

[JaZZ]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif ...the main substance of what I have stated is correct, namely: 24bit is a complete waste of time for the consumer and so is 192kFs/s.

I'm not pretending the opposite -- since I have no experience with 24/196. But in contrast to you I just leave it open that the higher data resolution (from 96 to 192 kHz) might possibly provide a noticeable sonic advantage -- as 96 kHz does compared to 44.1 kHz, despite the fact that it shouldn't.
.

 

[gregorio]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif I'm not pretending the opposite -- since I have no experience with 24/196. But in contrast to you I just leave it open that the higher data resolution (from 96 to 192 kHz) might possibly provide a noticeable sonic advantage -- as 96 kHz does compared to 44.1 kHz, despite the fact that it shouldn't. .

Theoretically there is good a reason why 96kFs/s could sound better than 44.1kFs/s. The anti-alias filters at 44.1k have to be very steep (to stay out of the hearing range) and it's relatively difficult and expensive to create good steep brick wall filters without noticeable artifacts. The anti-alias filters are much smoother (less steep) with 96kFs/s and so are much easier and cheaper to implement without noticeable artifacts. So in theory it is potentially possible that a difference could be perceived. There is some anecdotal evidence to support this potential difference but so far as I'm aware no one has yet managed to get a significant result in DBT. Most good studios use good quality ADCs though, so in most commercial releases it is extremely unlikely (but not provably impossible!) that the anti-alias filter at 44.1kFs/s is going to have any noticeable artifacts which could be improved by the smoother filters in 96kFs/s.

Again though, even if there is a perceivable difference with 96kFs/s it is not related to data resolution. I conducted my own tests in 2002 with both 96kHz and 192kFs/s and neither I nor any of my colleagues could tell a difference, except that my hard disk space seemed to get eaten up at an alarming rate and my processing ability was decimated!

These days, when recording for TV/Film I use 24bit-48kFs/s and then dither down to 16bit. For CD I generally use 24bit - 44.1kFs/s or if I'm not convinced with the quality of the ADC, I might use 24bit-88.2kFs/s then dither back to 16bit-44.1k. Beyond for testing purposes, I don't bother with 192kFs/s at all.

G

 

[JaZZ]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Theoretically there is good a reason why 96kFs/s could sound better than 44.1kFs/s. The anti-alias filters at 44.1k have to be very steep (to stay out of the hearing range) and it's relatively difficult and expensive to create good steep brick wall filters without noticeable artifacts. The anti-alias filters are much smoother (less steep) with 96kFs/s and so are much easier and cheaper to implement without noticeable artifacts. So in theory it is potentially possible that a difference could be perceived. There is some anecdotal evidence to support this potential difference but so far as I'm aware no one has yet managed to get a significant result in DBT. Most good studios use good quality ADCs though, so in most commercial releases it is extremely unlikely that the anti-alias filter at 44.1kFs/s is going to have any noticeable artifacts which could be improved by the smoother filters in 96kFs/s. Again though, even if there is a perceivable difference with 96kFs/s it is not related to data resolution.

We already had this. I think it is, nonetheless. Because in this case I'm referring to the «Dragon Boats» -- and the comparison between the original 96-kHz and the downsampled 44.1-kHz version. No hardware filtering involved.
.

 

[linuxworks]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif then as mentioned previously 24bit is useful for recording and mixing, it's as a playback consumer format that 24bit is irrelevant. G

exact same analog in photography. you want to capture at high bit depth, but even more important is STAYING at that depth during the processing. in photo, we start with a raw image (non compressed) and we keep data at 16bit/pixel and only at 'save as' time do we 'dither down' to 8bit jpg for regular old joe's use on the web or even for print.

edit needs high res because of math roundoff. but final user presentation does NOT apply to the same standard and 8bit is just fine for photo use.

I do believe that 16bit audio is 'just fine' for end consumer use. a GOOD clean 16 bits, but that's all it really takes.

also, do remember, that in the real world, storage is not infinite. photos, songs, movies - they eat up disk. that's NOT a good thing (I'm thinking blue ray, which is hella expensive in storage for what it SHOULD be taking, but I digress..). so even if there's some slight test-equip measurable diff by going deeper in bit depth or sampling freq, you have to remember that portable players and even home disks have to store this stuff.

a good technical design makes a tradeoff between absolute perfection and the real world. 24bit and 96k do not seem 'real world' to me as final formats for end user listening. still too expensive and they don't justify their piggish storage needs.

 

[JaZZ]

Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif exact same analog in photography. you want to capture at high bit depth, but even more important is STAYING at that depth during the processing. in photo, we start with a raw image (non compressed) and we keep data at 16bit/pixel and only at 'save as' time do we 'dither down' to 8bit jpg for regular old joe's use on the web or even for print.

Yes. -- Pixel number and sharpness on the other hand would represent the sampling rate. So if you want the pictures to be enjoyed on a large poster format, better leave it in hi-rez. Some audiophiles listen to their music in poster format (think electrostatic headphones, HD 800, high-end speakers...).
.

 

[gregorio]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif We already had this. I think it is, nonetheless. Because in this case I'm referring to the «Dragon Boats» -- and the comparison between the original 96-kHz and the downsampled 44.1-kHz version. No hardware filtering involved. .

Filtering has to be involved, there is no choice or option. The Nyquist limit for 44.1kFs/s is 22,050Hz anything above this frequency must be removed otherwise the sampling (or re-sampling) process either fails entirely or results in alias images which cause serious artifacts. It's not hardware filtering, just as it isn't hardware filtering in ADCs but it must brick wall filter, this is a basic "law" of digital audio.

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif Some audiophiles listen to their music in poster format (think electrostatic headphones, HD 800, high-end speakers...). .

No they don't, there is no equivalent in audio to "large poster format". With bit depth it would mean the audiophile listening at a volume which would at the least make them permanently deaf and in sample rate it would mean pitch shifting the highest frequencies in the music down by more than 2 octaves, while leaving the lower frequencies unaltered (otherwise they would be too low to be heard!).

G

 

[JaZZ]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Filtering has to be involved, there is no choice or option. The Nyquist limit for 44.1kFs/s is 22,050Hz anything above this frequency must be removed otherwise the sampling (or re-sampling) process either fails entirely or results in alias images which cause serious artifacts. It's not hardware filtering, just as it isn't hardware filtering in ADCs but it must brick wall filter, this is a basic "law" of digital audio.

Now you lost me. Are you saying that WaveLab doesn't do the low-pass filtering right during downconversion? Which kind of filter(ing) would do it right, then, in your opinion? BTW, I also compared the original 44.1-kHz version of the «Dragon Boats» with the hi-rez version, to the same avail.

It seems that you can't avoid brickwall filtering (no news) -- hence you can't avoid the flaws you attribute to it.
.

 

[JaZZ]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif No they don't, there is no equivalent in audio to "large poster format". With bit depth it would mean the audiophile listening at a volume which would at the least make them permanently deaf and in sample rate it would mean pitch shifting the highest frequencies in the music down by more than 2 octaves, while leaving the lower frequencies unaltered (otherwise they would be too low to be heard!).

Weird analogies -- and definitely inadequate. The equivalent to a high bit depth would be a high bit depth -- so high dynamic accuracy at ultra-low levels. The equivalent to increased sampling rate would be finer detail (e.g. by finer printing screen) and increased sharpness.

The equivalent to your pitch shift in the field of photography would be a shift in the electromagnetic spectrum (e.g. green would turn to ultraviolet or the like).
.

 

[gregorio]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif Now you lost me. Are you saying that WaveLab doesn't do the low-pass filtering right during downconversion? Which kind of filter(ing) would do it right, then, in your opinion?BTW, I also compared the original 44.1-kHz version of the «Dragon Boats» with the hi-rez version, to the same avail. It seems that you can't avoid brickwall filtering (no news) -- hence you can't avoid the flaws you attribute to it. .

"Doing it right" is a contentious issue. I know that a lot of design and processing power is thrown at "doing it right" in high end ADCs. What I do know is that it's very difficult to avoid artifacts with a steep brickwall filter. But brickwall filters aren't all created equal, there are quite a number of different ways an anti-alias filter can be implemented. I posted a link earlier in this thread to a really good paper on digital filters. The exact implementation of what is done by Wavelab or indeed inside ADCs with regards to brickwall filters is beyond my knowledge and to an extent probably considered by some to be a trade secret. To be honest Wavelab is not generally used in commercial studios and I don't believe is considered "professional" quality more of a pro-sumer product. I don't know what dithering algorithms are used by Wavelab either.

Sorry, I haven't really answered your question. I would like to run the test file "Dragon boat" myself to find out what is really going on but my system is currently configured for a project I'm working on and I won't want to mess with it until I'm finished on the 7th April.

G

 

[linuxworks]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif Yes. -- Pixel number and sharpness on the other hand would represent the sampling rate. So if you want the pictures to be enjoyed on a large poster format, better leave it in hi-rez. Some audiophiles listen to their music in poster format (think electrostatic headphones, HD 800, high-end speakers...). .

shooting 'for posters' does NOT need high res!

if you are standing next to the poster, sure. but photogs don't shoot 'for that' - they shoot regular slr style res and when you get noise, well, you get noise. posters, like high def tv's, only really look good when there is PROPER viewing distance. you can't judge a photo by what it looks like an inch away from your face (so called pixel-peeping).

I might continue the analogy, though; that headphone listening IS more like looking at a photo closer-up than using home speakers would be. flaws in the material and the amp chain are more noticeable in phones than speakers.

funny that I find that almost all my music sounds 'fine' on speakers but once I start to plug in phones, the flaws all come out and sound ugly. having TOO high a res can hurt you, in some situations (like playback)

 

[gregorio]

Quote:

Originally Posted by JaZZ /img/forum/go_quote.gif Weird analogies -- and definitely inadequate. The equivalent to a high bit depth would be a high bit depth -- so high dynamic accuracy at ultra-low levels. The equivalent to increased sampling rate would be finer detail (e.g. by finer printing screen) and increased sharpness. The equivalent to your pitch shift in the field of photography would be a shift in the electromagnetic spectrum (e.g. green would turn to ultraviolet or the like). .

Exactly, to hear the accuracy at very low levels in 24bit audio you would have to turn your amp up so high that when a sound came in near the high end of the dynamic range it would probably put the listener in a coma (no joke)!! But as I've already mentioned there is no equipment out there that can actually resolve 24bit audio. And yes, your analogy with using a higher sample rate would mean at 192kFs/s three quarters of the colours used would be beyond the light frequencies visible to a human being.

In other words, a correct analogy would be a poster which included colours so bright that looking at them would cause permanent blindness and where three quarters of the colours used could only be seen or appreciated by bees!! Unless of course you are a 'visiophile' in which case those ultra-violet colours and x-rays definitely improves the percievable quality of the image

G

 

[JaZZ]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Exactly, to hear the accuracy at very low levels in 24bit audio you would have to turn your amp up so high that when a sound came in near the high end of the dynamic range it would probably put the listener in a coma (no joke)!! But as I've already mentioned there is no equipment out there that can actually resolve 24bit audio. And yes, your analogy with using a higher sample rate would mean at 192kFs/s three quarters of the colours used would be beyond the light frequencies visible to a human being. In other words, a correct analogy would be a poster which included colours so bright that looking at them would cause permanent blindness and where three quarters of the colours used could only be seen or appreciated by bees!! Unless of course you are a 'visiophile' in which case those ultra-violet colours and x-rays definitely improves the percievable quality of the image

Your examples are funny, but not adequate. No need to defend 24 bit or 14/16 bit in photography other than for editing purposes, but I definitely see an audible advantage in hi-rez sampling rates and don't even exclude a benefit from 192 kHz. Your reasoning isn't entirely convincing to me. Your excuses for the brickwall filter being difficult to implement properly rather come across as an attempt to defense your current point of view. If it's indeed just the implementation which is so hard to do perfectly, then it lacks perfection in every environment I've compared hi-rez to 44.1 kHz.
.