|
Digital Drive Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it. |
|
|
|
Digital Drive Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it. |
|
In Reply to: That graph posted by JimF on January 2, 2003 at 08:39:00:
. . . what "resolution enhancement" consists of,
it would be the following. Take it with a shaker full
of salt, because I have only an audiophile's
(better than a layman's, maybe, but far less than a
mathematician's or DSP expert's) understanding of the
issues involved.In a Usenet thread from 1996 about Audio Alchemy's "EDR*S"
system, somebody wrote (see link):> > You can't increase the resolution of a given recording.
>
> This is not strictly true. If you know the signal follows a
> certain set of rules, you can theoretically reconstruct it.
> A bunch of people in the 70s did this for pictures, and
> you can also ask some math friends about analytic
> continuation.If you were to capture a window of 16-bit-digitized music
and perform a fast-Fourier transform on it, you would convert
the time-doman data into a set of frequency-doman components
(each representing a sine wave). There would be a lot
of these; too many to process all of them. If however,
you drew upon a body of knowledge about psychoacoustics
to identify the "really important" components (based on
their frequency and level), you could use the above-mentioned
"analytic continuation" to "clean up" just these components,
to add bits to the original 16-bit data "as if" those
components had been digitized at greater than 16-bit
resolution (and maybe even cleaned up better than they
could ever have been digitized by any real-world
A-to-D converter).I think something like this might be going on. I find
it particularly suggestive, for example, that the
graph supplied for the DTI Pro review in _Stereophile_
shows the biggest effect (up to 12 additional bits of
resolution, tantamount to 28-bit A/D conversion) in
a narrow region of frequencies around 5 kHz, just where
the human ear happens to be most sensitive.Some people have commented that it would be far better
to have a "real" 24-bit digitization, such as that on
the high-resolution PCM track of a DVD-Audio disk, rather
than a 24-bit datastream synthesized from a 16-bit CD.
Maybe this is true, on purist philosophical grounds, but
it wouldn't surprise me at all if sophisticated
psychoacoustic processing can generate a "prettier",
more palatable, result even than an unprocessed 24-bit
digital recording. Of course, there's no reason you
couldn't "gild the lily" of even a 24-bit recording to
further enhance its audiophile appeal using similar
processing.I find the thought rather heartening, rather than disgusting.
After all, audio has always contained a large component
of "smoke and mirrors" as well as straightforward, by-the-book
science and engineering. Back when digital audio was first
introduced, many people boasted (or lamented) that the
techniques would rid audio of the magic and mysticism
inherent in LP playback via tweak turntables and hand-built-
by-Japanese-artisan moving-coil cartridges made out of
exotic and mysterious materials. Well, it may not be
so after all! This kind of DSP processing may well offer
a window of opportunity for the same kind of application
of artistic discernment and cottage craftsmanship. Who
knows, maybe Keith Allsop is the Yoshiaki Sugano of the
digital world! ;->Jim F.
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- If I had to guess. . . - JimF 01/2/0309:53:06 01/2/03 (1)
- Masking and psychoacoustics. . . - JimF 11:47:22 01/2/03 (0)
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]