Best sound quality - run 96khz samplerate, do not oversample.

[jupiter8]

1. who/what sets the "limit" of human hearing?
science? math? nature? ignorance? evolution/adaptation? Chuck Norris? Apogee?

+1

Just because something assumed/presumed to be "in-audible" does not mean it is not perceived in some way or manner with which "we" are not aware, or otherwise comprehend to any degree.

This can easily be tested.

[bezusheist]

1. who/what sets the "limit" of human hearing?
science? math? nature? ignorance? evolution/adaptation? Chuck Norris? Apogee?

How is this even remotely a problem ?

huh?
the problem is you have "tools" like Ethan Weiner spreadin' crap like a dairy farmer...
the limits of man should not be set by the limits of man's knowledge...

[A.M. Gold]

It can be about how well the converter reproduces a smooth output and it can do this better at higher rates. Now this is usually seen (on a 'scope) more than it's heard (if at all) but there's a nerd factor too.

I think the argument there was that the "reconstruction filter" on all modern (i.e. at least post 2000) converters is always enough to ensure that the output is sufficiently smoothed such that higher sample rates are unnecessary.

[jupiter8]

1. who/what sets the "limit" of human hearing?
science? math? nature? ignorance? evolution/adaptation? Chuck Norris? Apogee?

How is this even remotely a problem ?

huh?
the problem is you have "tools" like Ethan Weiner spreadin' crap like a dairy farmer...
the limits of man should not be set by the limits of man's knowledge...

The limits if human hearing is obviously set by evolution but my point is,it can easily be tested so there is little doubt what the limits of human hearing is.
It's not a mystery as some people tend to believe.

[khanyz]

It can be about how well the converter reproduces a smooth output and it can do this better at higher rates.

No, absolutely not.

Yes, absolutely yes. I have experience of this and it's usually a byproduct of the associated filter design.

I never said this could be heard, but it can be seen.

[Shabdahbriah]

.. the limits of man should not be set by the limits of man's knowledge...

Which is essentially what occurs, through "ignorance" passing for 'knowledge'...

Curious, no?

[A.M. Gold]

It can be about how well the converter reproduces a smooth output and it can do this better at higher rates.

No, absolutely not.

Yes, absolutely yes. I have experience of this and it's usually a byproduct of the associated filter design.

I never said this could be heard, but it can be seen.

Here is the crux of the disagreement. On the other forum that I was referring to, there was all kinds of back and forth about whether a test could be devised that would be fair, who would design it, who would admnister it. I supported the idea of the test but I think feathers were so ruffled during the online debate that people just gave up on the idea of ever agreeing about testing procedures.

[khanyz]

I think the argument there was that the "reconstruction filter" on all modern (i.e. at least post 2000) converters is always enough to ensure that the output is sufficiently smoothed such that higher sample rates are unnecessary.

Probably, but then you'd still see the difference on an oscilloscope so someone in marketing will roll it out as an advantage in high end stuff.

Progress does mean that high end, will eventually become mainstream, which then becomes budget so as the converters become better, the source doesn't have to do as much work.

That's not an excuse for a lazy source though.

[Compyfox]

There are no sonic drawbacks, as there are with oversampling or running 44.1khz w/o oversampling.

To stir up the discussion a bit (and add some controvery) - I disagree.

Higher sampling rates good and fine, but you're still limited to the 20Hz to 20kHz frequency range. You just have a finer resolution while recording (if you did it right of course).


The thing with the "ultimate goal to find the best possible sound" is not possible if our AD/DA's are locked to these common frequencies, while (yes, I say the evil word now) analog/old outboard equipment do not have those limitations.

Even if we don't hear it, or see it on an analysing tool (at least not on one that goes from 20-20k), there are frequencies that add to the overall sound. Things that are missing in the pure digital realm. Now... if AD/DA's would reach from 10Hz to let's say 22kHz rather than the known 20Hz-20kHz, we suddendly have more material to listen to and to work with. A totally different game!


Why waste sampling cycles if:
- we have a limited frequency range
- we overedit the recorded material
- compress the crap out of it while both mixing/mastering
- the endformat is CD / MP3 (44kHz, 16bit) or DVD-Audio (48kHz, 24bit) mostly with lossy encoding formats anyway (not counting MPEG HD-AAC/lossless, MPL and DTS licenses, since they're friggin' expensive)?


???



*edit: oh, looks like this discussion already went down that road - time to sit back, grab some popcorn and enjoy the show*

[khanyz]

Here is the crux of the disagreement. On the other forum that I was referring to, there was all kinds of back and forth about whether a test could be devised that would be fair, who would design it, who would admnister it. I supported the idea of the test but I think feathers were so ruffled during the online debate that people just gave up on the idea of ever agreeing about testing procedures.

That's why I mentioned the Natalie Merchant album as an independent test. Sure it will be mastered differently, as the limits of the format are differnt, but it would have to pass the same judgement on acceptability for release. It could be done with a few albums for a better sample.

I don't think you can come up with a good test, as there's opinion at every level of it. What's the best source material, the best dither etc.

However, going back to the original topic. As a general rule you can never recover quality and so it's better to start with the best quality you can.

If you then process it with more than enough headroom and detail, you are trying to maintain the quality level until you are forced to degrade it.

So the argument is what headroom (dynamic and frequency) are you comfortable working in. For me, the 2k provided by 44.1k is too low and the few dB I can get by combining several 16bit tracks into a single 16bit track is not enough. The solution is to work at higher rates and bit depths. I then have room to play and will only lose quality when, and how, I decide to.

The fact that things have to be mastered differently for 24bit, 16bit and lossy compression is also a compromise I may not be happy with. So I can present my music as I intended (24-bit), as I would accept (16bit) and begrudgingly (lossy compression). People can then pick how they want to listen to it.

I'm tired now, so I'm pulling over and letting someone else drive for a bit.

Cheers,
Nigel

[Urs]

The problem is that the added harmonics in a signal chain add to each other. That is, a nonlinear process will also add harmonics to the already added ones from the previous process. So each time you double the samplerate all you get is one more plugin in the signal chain to end up with the same level of artifacts.

The best way to keep aliasing low is to bandlimit each single process in the chain. Thus deploying steep filters inbetween. 'tis the only way.

[mandolarian]

Sorry, I'm late to the science and marketing party. I'm just gonna comment on the basics 'outside' of the box and how it gets there.

First: It's Nyquist Theory, not Nyquist Law.

But, it's been the best theory for digital audio recording so far. I'm open to better theories and applications. And $50 mics with 100k bandwidth, off axis.

Dan Lavry will go with you up to 88.2/96K. And then abandon you to the bat whispers and wide bandwidth marketing. He demonstrates the phenomena that 192k, never mind 384k actually degrades accuracy. A good, if equation-riddled read at: http://www.lavryengineering.com/forum_i ... _Audio.pdf

Dan advocates an 'optimal' sampling frequency. As an audio hardware engineer he's well aware of the difficulties in putting theory into practice. His practices are among the best.

Most engineering solutions are compromises - however, like internet forums, marketing departments are unfettered by such constraints.

High quality A/D/A convertor at 192K/384k is more marketing myth than science. Great for hardware vendors, tho. That's the only real benefit. Unless you believe Nyquist theory has only theoretical merit. Or your music is based on bit-crushed bat sonar.

[jupiter8]

First: It's Nyquist Theory, not Nyquist Law.

It's a theorem actually,completely different thing.

[khanyz]

The problem is that the added harmonics in a signal chain add to each other. That is, a nonlinear process will also add harmonics to the already added ones from the previous process. So each time you double the samplerate all you get is one more plugin in the signal chain to end up with the same level of artifacts.

The best way to keep aliasing low is to bandlimit each single process in the chain. Thus deploying steep filters inbetween. 'tis the only way.

How much headroom do you allow within each process for it to behave as intended (dB/Bit Depth and Sample Rate)? How far above 20kHz would you place the filter for it not to encroach significantly on the intended range of use?

How much difference is there in the theoretical and practical values?

[A.M. Gold]

First: It's Nyquist Theory, not Nyquist Law.

It's a theorem actually,completely different thing.

Actually, I brought up Nyquist but I don't think he has much to do with the actual argument. The real argument is about 20 KHz being a cutoff for "relevance" in matters of human hearing (and some will argue it's usually below that---which is true for many older people).

People like Ethan like to think that there are only a few torpedoes that have to be launched at a relatively easy target in order to sink this argument once and for all. One is that he strongly adheres to the "20 KHz limit". Another is that he thinks possible issues with D/A reconstruction are sub-audible. He also rejects that there is any other source of perceptible distortion or degradation in modern A/D/A systems, even low budget ones.

So with that he essentially rises to the position that CD quality audio is a perfect replication of whatever live source is fed into it, limited only by the reproductive capacity of the speakers and the quality of the amp. He has often characterized it as far superior to vinyl or studio tape.

But the argument is two-fold. One argument involves measurable issues and the arguments are all about hearing and perception. The other is far more nebulous as comes into play when people assert that the "four parameters" must not be the only relevant acoustic phenomena, or that there is some kind of interaction ging on between them that isn't understood, etc., etc.