[linux-audio-user] Digital Fidelity

>perhaps we should fork this discussion off at some point......

Done.

>> I guess, then, that *real* 24-bit resolution, or something very close to
>> it, would yield what I am looking for - if it can be achieved.
>
>Are you sure that's what you are looking for?

Well, I guess I would have to hear it to be able to answer that. :)

>Fidelity is a measure of how closely the signal you get out of your
>recorder matches what you put into it.

Agreed. That is all I am after.

>It's more likely that you are hearing differences in quality from one
>component to the next. I can hear the difference between an Apogee 24
>bit converter and a cheap no-name 24 bit converter.

I have no doubt that this is indeed a significant factor . . .

>Ah, but there are so many other differences between those effects than
>their bit depths. Let me guess, they sound better in the chronological
>order they were released in? The amount of DSP available and the quality
>of the code has changed too...

. . . in fact, I don't have any way of knowing *what* is really in a piece
of hardware, regardless of what the specs say, the company that
manufactured it, or even how much I paid for it.
I only know that some components sound much better than others - I can't
say for sure why.
If they claim to have a higher resolution, it is of course natural to
assume that has something to do with *why* they might sound better.

>Can you prove you can hear the difference between 24 or more if
>no real 24 bit converter does exist ? It is _extremely_ difficult
>and expensive even to do a valid test at 20 bits. Some people
>have done it, and they all arrive at the same conclusion.

Not without being in a lab under controlled conditions - no, of course I
can't prove it.

>> Again, what do you base this on?
>
>Working knowledge of how good analog recording is, understanding of the
>theory of sampling and quantisation, undisputed results from
>psycho-acoustic
>research, and elementary physics and mathematics.

I have some basic background in those subjects as well, yet I do not agree.
Psycho-acoustics is by its very nature *subjective* - you cannot have
'undisputed' results from this -
it is as fallible as any statistical sampling, and as easily skewed.

>Yes. A correctly dithered signal converted back to analog is
mathematically
>equivalent to the original unquantised version plus some noise. There is
no
>way, not even in theory, to detect it was ever quantised. Since it can't
>be detected, you can't hear it. But you could fool yourself into thinking
>you can, as many have done before you. After (correct) dithering the only
>'defect' that remains is noise. And with 24 bits and standard signal
levels
>this is well below the thermal noise of any analog amplifier that exists,
>and also well below human hearing thresholds.

If there is one correct way to do this, then there should be no reason for
different 'noise-shaping' algorithms.
In fact, why are there different noise-shaping algorithms if the noise
can't be heard.?

>> I can hear the distortion of the audio signal created by Dolby - and I
>> don't like it.
>
>What has that to do with this discussion ?

I merely mentioned that as another example of psychoacoustic masking that
supposedly one cannot hear - yet I can.
I can also hear the difference between a digital copy and the original
sound file, and between the same generation of digital copies on different
hard drives.
I can hear radical differences in audio quality between CDs burned at
different speeds.
Theoretically - or mathematically as you wish to present it - I shouldn't
be able to hear any of this: they are all mathematically the same, and
should sound identical - but they do not.

Perception by the human ear and human mind cannot be reduced to a
mathematical equation, however much you may wish to do so.
There are organic fragrances that never have, and never will, be able to be
synthesized, or even distilled - even with the most refined and careful
processes - for the same reason.
Perception is not mathematical - and that applies as much to that which is
perceived as to the perceiver.
 
>Buy some *good* converters, add Ardour and the result is *far* better
>than any 24 track analog machine that ever existed. That is if your
>idea of quality relates to fidelity and not to some specific typical
>analog distortion that you may like or mistake for 'correct'. By 'good'
>converters I mean at least the quality of RME, or better Apogee.

That, I would be most happy to do.
I thought RME was the pinnacle of quality, but if you say Apogee is even
better - we will certainly try it (finances permitting, that is. :)

>You still probably won't believe me, but the fidelity of a £100 card
>like an audiophile 24/96 will be greater than that of 24 track 2".
>
>The audiophile will have a lower noise floor, better linearity, no
>scrape flutter or wow, much lower cross talk between channels, much less
>IMD, wider frequency response (and a more solid bass end).... but it
>might not sound as 'good'.
>
>I don't know if you have ever worked with tape, but you really did have
>to be so much more careful than digital about getting a good level to
>cut down noise, putting non critical tracks on 1 and 24 as they always
>got a bit knackered on reels and transport, recording at lower levels if
>the source has lots of hf content, line up and bias.... all this stuff
>was a total pain in the arse. Most everyone used some kind of noise
>reduction, unless they were pushing the tape really hard, in which case
>the distortion figures are laughable compared to digital.

Yes, I realize that there are all kinds of problems - especially noise -
with tape, and it is not the 'perfect' way to record, and I am equally well
aware of the great advantages digital has over analog.
My point has not at all been to criticize digital recording technology.
I want very much for digital recording to match or exceed analog in quality
- I am just not yet convinced this has been achieved.

>I submit that the perceived 'superior' perfomance of good analog tape
>recorders, of any track width, is more a long term ear training result
>than anything else, after all, we have been listening to such
>machinery, faithfully encoded even on our cd's, and before that on our
>lp records of yesteryear, for 3 or 4 generations now. Do that for 55
>years, and the ear thinks thats what its supposed to sound like,
>effectively becoming its 'Gold Standard'.

You certainly have a good point there - I think there's a lot of truth to
that.
But what I'm after isn't a perceived 'Gold Standard' of analog recording.
I'm just after the most faithful possible reproduction of what I hear live.
That is what I mean by fidelity.

>That diff of tape or no tape isn't always that obvious, and I had that
>hammered into me one evening in about 1961 when I had a chance to
>listen to one of Emory Cooks 78 rpm lp's that had been recorded in
>trinidad, live to disk, of some of their then infamous steel drum
>bands. No tape in the path, straight from Altec M21 mics thru the
>preamps & to the cutter head making the master.
>
>The hair stood up on the back of my neck, it was that real. There was
>stuff from the background crickets at 17khz or more that was as live
>and real as if I had been standing in the middle of those crickets
>myself. Even the whispers of the drummers as they kept each other in
>step, probably 55db below the drums, could be heard well enough to
>understand it if they were using english, which some didn't.

Look - I do understand what you guys are trying to say, and respect the
fact that you have some science and experience to back it up.
I will just say this:
We have an old Tascam portable 8-track, which is now ready for the junk
heap, but we got close to perfect fidelity (after a lot of hard work) of
what we recorded on it with respect to the live sound.
If I wasn't looking, I couldn't tell if my husband was playing live, or
playing back his recordings.
Our early attempts to record that live sound through our Gina card directly
to the hard-disk sounded just plain bad: harsh, strident, thin - cold, but
more to the point - not at all like the live sound.
The analog recordings have a warmth to them - a midrange 'fullness' that I
don't hear digitally.
Digital can sound very sterile.
(When we attempted this through our earlier Pinnacle Multisound, it sounded
like a midi guitar.)

When we record now through our hdsp9632, the fidelity is very good - very
clean (almost *too* clean), but still not quite the live sound - though
very close.
When I am unable to tell whether my husband is playing live, or playing
back a digital recording of his music - then I will believe that the
digital technology has matched analog.
If all it is going to take is a better quality AD converter - then I will
be thrilled!

>I know by now I've bored all the knowitalls here to tears, so I'll go
>back to my corner now.
>
>--
>Cheers, Gene

Not at all, Gene, I enjoyed your comments.

>Dogma warning: You're not taking all the potential phenomena into
>account that have not been scientifically explained yet.
>
>I'm not saying Maluvia can hear a difference, I'm just saying you don't
>know that she can't.

Thank you, Carlo. ;)

- Maluvia
Received on Mon Feb 27 08:15:10 2006