limitations of MD

[P L Patrick]

This is a very helpful discussion to me, and perhaps for those
interested. But can anyone reconcile what Celso reported favorably
about MD compression with what Alexander said (and I understood to be
true)? I believe it isn't the dynamic range, rather the selective
distortion of the signal, that is the problem with MD. But we have
reached the limits of my knowledge...

On Mon, 3 Apr 2000 09:49:18 -0500 (EST) Alexandre Enkerli
<aenkerli at indiana.edu> wrote:

[MD &] MP3s..
> both use psychoacoustical phenomena to simply get rid of the
> parts of the sound that can't be heard (due to masking effects and
> such). This means that you really lose some of the information that's
> in the signal but that this loss shouldn't affect the perceived
result.

	True. But linguists cannot trust the music industry to
determine what is "perceived" or "can be heard". Of course, there is no
such universal standard of perception: my two-year-old is constantly
asking me to identify sounds which I am unaware of until he asks,
having processed them out of my awareness.
	In a pinch I prefer to add noise (with analog recording) rather
than to subtract elements of the authentic speech signal (with MD).
	Lots of people believe anything is OK for "just interviews".
But it's not possible to know what Qs you may want to ask in 5 or 10
years' time; nor to know whether your data will prove historically
useful for someone else, asking Qs of a different sort. Why run the
risk of distorting data at the outset? You can never get it back.

> In ethnomusicology where both sound quality & archiving are important
> factors, the recommendations are still in favor of analog. DAT tapes
> have not been tested for longevity and DAT recorders are usually very
> fragile for most field situations. Which doesn't mean that digitizing
> is out of the question, just that digital documents are not very good
> for archival purposes.

	DAT tapes are modern tapes, however, without the vulnerability
to magnetic fields that magnetic tapes have. Magnetic tapes are indeed
the longest-lasting format we have so far, and have been recovered from
undersea and found to still contain information. The last 15 years have
seen a lot of knowledge in tape construction, so DAT tapes are probably
a pretty good bet too.
	The best bet for archiving, however, is always multiple copies
(regardless of medium) preserved in independent locations. You can't
beat digital formats for ease and fidelity of copying -- you just have
to be prepared to transfer every time the formats change. (When was the
last time you accessed a 5.25" floppy disk?!)
	--peter
	
Prof. Peter L. Patrick
Dept. of Language & Linguistics
University of Essex
Wivenhoe Park
COLCHESTER CO4 3SQ
U.K.

Tel: (from within UK) 01206.87.2088
    (from outside UK) +44.1206.87.2088
Fax: (as above)           1206.87.2198
Email: patrickp at essex.ac.uk
Web: http://privatewww.essex.ac.uk/~patrickp