[NTLK] More digital talk

From: R Pickett (emerson_at_hayseed.net)
Date: Wed Oct 31 2001 - 17:19:25 EST


On Wed, 2001-10-31 at 08:24, newtontalk_at_newtontalk.net wrote:

(Digest here, so I'm responsing to several snips; apologies in advance
if I misattribute anything)

> From: Robert Benschop <rbenschop_at_mac.com>
> on 31-10-2001 4:23, Jon Glass at jonglass_at_usa.net wrote:
>
> > The comment was that many audiophiles, when CDs first came out complained of
> > this phenomenon. I also gave a one explanation as to why they thought this.
> > Personally, I thought this argument went away years ago. :-)
>
> Why ?

Because the bulk of the industry stopped mastering CD's as if they were
vinyl, once the phenomenon was understood. If one is still listening to
old (pre-1985 or so) CD's versus LP's of the same material, though, one
will still hear the objectionable frequencies, and bring it back up as
an 'issue with CD audio' all over again.

> From: Robert Benschop <rbenschop_at_mac.com>
> on 31-10-2001 8:22, The Compulsive Splicer at splicer_at_paroxysm.com wrote:
> >(Try it sometime--burn a CD-R with uncompressed WAV or AIFF
> > files of songs from your favorite CD and I think you'll find they fit).
>
> Yes, and it will sound worse compared to the original.
> Not as bad as MP3 though.

A straight digital copy of an AIFF or WAV, burned to CD, will sound
worse than the original file? No. It will not. It will sound exactly
the same. It's the same bits. If it sounds different, it's an artifact
of the CD player, the sound card you're listening back through, or some
other thing. But the sound data are _exactly_the_same_, definitionally.

> From: Marco Mailand <newton2k1_at_mac.com>
> I've tried what happens with a AIFF file burned in Audio format but with
> lower sampling rate (in my case 22.05kHz) in a normal CD player. It simply
> plays as twice as fast and is not useable. But it is feasable to slightly
> change the sampling rate from 44.1kHz to e.g. 43.5kHz in order to squeeze
> out a minute or two. I've tried this too but I got a distortion with the
> difference frequency of 600Hz on all songs, which made the whole try
> worthless. The distortion was very likely caused by imperfections of my
> audio hardware (not perfect anti-aliasing filters of the built-in Macintosh
> audio chip).

Yes, this will not work. The CD Red Book spec specifies 44.1kHz
playback. Your audio CD player will play back at 44.1kHz. If you've
put 43.5 kHz material on there, it will play back some tiny percent fast
and sound weird. If you put 48kHz data from a DAT onto a CD, it'll play
back about 10% slow, and be dropped in pitch by two half-steps or so.
Compressing audio and putting it onto a Red Book CD (one that will play
back in a player that has the "CD" logo on it) will just make it play
weird. There's not any concept of multiple sampling rates like on a
computer.

> From: Marco Mailand <newton2k1_at_mac.com>
> Michael J. Hu=DFmann wrote:
> > But that amounts to re-defining "compression". In any case, there
> > is no infinite number of values to record -- remember quanta?
> Okay, but looking on the groove of the good old vinyl disc you could
> theoretically capture infinitly much needle positions.

No, no. That's what he's SAYING. 'Quanta,' without getting into a big
physics lesson, tells us that nature has a sampling rate. There is a
smallest distance that something can move; there is a lowest amount of
energy that can be exerted. Nothing smaller, nothing lower-energy,
exists.

So, the idea of analog == infinite is flawed. Analog has a vastly
superior sampling rate than digital, but it's still a finite amount of
information. Add to this the concept that all of this has to go through
our eardrum, which has a threshold of how much energy it takes even to
get it moving at all, and there IS an absolute bottom limit on not only
analog recordings, but actual live performances, to nature itself.

CD's parameters of 16 bits precision and 22.05k frequency response were
chosen in the late 70's as the format was being developed. These were
FANTASTICALLY better than all but the most audiophile systems could
reproduce at the time. Conveniently, it also allowed for album-length
recordings at the media density that could be created at that time.

These days, all professional and semi-pro digital recording is done at
the new 24bit resolution, 96k sampling rate (48kHz freq response), and
only 'squashed down' to CD at mastering time. 24 bits is _plenty_ to
cover the entire range of human perception and then some, and if anyone
can tell me that frequencies above 48k actually cause a perceptible
difference in the listening experience for humans... well, I'll be GLAD
to sell them expensive snake oil accessories to their listening systems.

As another aside, I can't find the reference, but I have read at least
twice that 32 bits of resolution could theoretically encode all values
of sound from full-compression shockwave down to molecular heat
jiggling.

Saying that digital is bad because you lose stuff at the top and bottom
is a whole lot like saying that film is bad because you lose the x-ray
and microwave information.

> From: Jon Glass <jonglass_at_usa.net>
> In an analog recording, the frequencies fade in a rather smooth curve on up
> to and way beyond the range of human hearing.

No. They plunge in a rather abrupt curve wherever the limits of the
recording medium are. Typically higher than CD's theoretical frequency
limit of 22.1k, but not on beyond zebra. Good tape can get you maybe
30k, but then we're also needing to account for the frequency response
of the microphones, recording console, etc.

And, of course, even the best speakers in the world tend to roll off
pretty sharply at 25k or so. Some piezo can go up to 40k, but there's
nothing recorded up there, realistically, ever.

Recording _is_ distortion.

> In the digital realm, that
> same recording is full blast up to the ceiling of 20k, where it is chopped
> off, wham, right at the ceiling.

Depending on whether a brickwall filter is applied. It typically _is_,
but actually CD audio can theoretically encode higher frequencies, at
algebraically increasing levels of distortion. 25k, still kinda there.
30k, basically turned into triangle waves, and also creating
objectionable subharmonic distortion down in the 'audible' bands.

> This creates an audible difference in the playback.

Hmm. No. That is, this is not a -false- statement, as much as it's far
too general. This _can_ create an audible difference in the playback.
Not if we're recording, as an extreme example, 100Hz sine waves for a
reference disk -- nobody will be able to tell the difference between
analog and CD digital. Nobody. Probably not even a computer.

-- 
R Pickett           The people that once bestowed commands, consulships,
Hayseed Networks    legions, and all else, now meddles no more and longs
emerson_at_hayseed.net eagerly for just two things  --  bread and circuses.

-- This is the Newtontalk mailinglist - http://www.newtontalk.net To unsubscribe or manage: visit the above link or mailto:newtontalk-request_at_newtontalk.net?Subject=unsubscribe



This archive was generated by hypermail 2.1.2 : Thu Nov 01 2001 - 10:02:58 EST