[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ARSCLIST] Digital (was 78) Listening tests, was Pristine Audio and the Milllennials . . .
I think the first pass of ingestion SHOULD be 96/24, under almost all circumstances. Why?
Digi-processing seems to work better under most conditions with higher resolution. Less "guesswork"
by the software. As for the final ARCHIVED format, that's a tricky question. I'm not sure archiving
is akin to mastering, so you might want to save at least your unprocessed first pass and your final
result audio, whereas in mastering, the final result is what makes the money (and, in field
recording and earlier-era music recording in the analog world, the first pass WAS the final result).
I think storage is dirt-cheap now and gets dirt-cheaper and this will be the least of the budgetary
concern for a properly-funded transfer/archive project.
Where the point I think Goran was making comes into play is in the playback and public-access
formats. You can save server horsepower and playback equipment costs by keeping this format
appropriate to the content. 192kbps MP3 are more than adequate for fidelity-challenged or simply not
wide-frequency-range recordings. Spoken word, for example, can be superbly recorded and still sound
just fine even down at 128kbps (where this doesn't work is spoken word with lots of ambient noise
behind it because the high-frequency part of the ambient noise can easily turn to digi-swish and be
super-annoying if the MP3 bitrate is too low). Another known space-saver: if you are making WAV
format available, keep mono source material as mono WAV files because they are smaller file-size
than duplicate-channel "stereo" WAV files.
I standardized on 96/24 a couple of years ago, and I've been grateful because there have been cases
where I've learned a new technique and gone back and had a high-resolution transfer to work with and
gotten better final results than the first time through. Also, at least in the case of Sony
Soundforge, I firmly believe that hiss-reduction works better at higher resolution (better = it
takes out more hiss without adding digital artifacts, whereas at lower resolution it adds artifacts
at almost any setting that actually audibly removes hiss).
A final consideration is that, yes, on a well-designed D-A chain, 44.1/16 should sound darn close to
the source for 90+% of people and probably indistinguishable in a blind test on a system in a room
that the listeners are not intimately familiar with (a trained ear is most atuned when listening to
their own system and in their own room because they have learned to filter out some or most of the
inherent imperfections and colorations). But, on most real-world playback system, the CD audio
format suffers from filter artifacts and/or inferior D-A conversion and/or inferior mechanics of the
drive and/or heavily-compromised media requiring more-than-desireable error correction if not
error-correction-failure, all of which case various imperfections and artifacts and compromise sound
quality. The higher-resolution formats sound better on most "universal" players because there are
few if any brickwall filter artifacts, among other things. Just to be clear -- I'm a big fan of the
CD format and I think, indeed I know, that great-sounding CD's can be made, CD's that sound to the
vast majority of ears and through the vast majority of systems just like the input of the transfer
chain would sound under that same condition (notice I'm NOT saying "like the master tape" because
who knows what sort of playback system might be used in a different situation from the CD mastering
chain). That's why it's so disappointing that so many CD's sound so bad.
-- Tom Fine
----- Original Message -----
From: "Marcos Sueiro Bal" <mls2137@xxxxxxxxxxxx>
To: <ARSCLIST@xxxxxxxxxxxxxxxx>
Sent: Monday, December 10, 2007 2:42 PM
Subject: Re: [ARSCLIST] Digital (was 78) Listening tests, was Pristine Audio and the Milllennials .
. .
I think I prefer the freedom to use 24 bits for live recording as I don't feel as compelled to
risk clipping in order to optimize level settings.
I prefer 24-bit as well to record, for the same reason as Richard, as well as to avoid possible
effective bit reduction in further processing.
I hope that these tests bring about a comfort level in the archival world with digitization at
less than 96/24 and hopefully we don't see specifications for digitizing oral history cassettes
at 96/24. While bits are cheap for storage, they aren't free. I know IASA says 48/24 for
everything with 96/24 preferred.
This is an uphill battle with many administrators, who naturally prefer one standard across the
board, and if the "official" document says 96/24 is *preferred*, one cannot blame them for
choosing that standard just to be on the safe side. It is not what I would choose for oral history
cassettes either, but...
Marcos