I submitted this interesting question to my old buddy Tony Lauck, who has
become a bit of an expert on such topics, and here is his (lengthy) and IMHO
well-considered reply:
There are two problems that have to be considered: (1) what to put into the
archive and (2) how is someone going to pull the item out of the archive at
some future date. (There is little point in putting something in an archive
if it will be useless when it comes time to take it out.)
If the material was received in a digital format, then you want to preserve
every bit of information that was received. If you don't do that, then the
copy will be unnecessarily inferior and this may be significant at some time
in the future. With audio files encoded in some file format there are two
ways of doing this, (a) by copying the actual bits in the file format as
received, or (b) transforming them to an equivalent format. What does this
mean, "equivalent"? This means that one can take the equivalent format, put
it through a reverse process and get back exactly the original bits. If one
can do this, then one can be certain that no information has been lost.
However, one should be skeptical about this and not believe software
manufacturer's claims.
I will give two examples. If the file is a WAV file, it can be compressed
with ZIP compression. This will save a certain amount of storage space.
However, the ZIP file can be expanded into an identical copy of the original
file. (Assuming that the ZIP software is of good quality, you did test this,
didn't you?) Or one could compress the WAV file with a lossless audio
compression program, like FLAC. If one believes the claims, the FLAC
compression can be restored and the audio file will have the identical
samples -- it would sound the same. However, the file will be slightly
different, because the header formats are different. So it won't be easy to
convince oneself that the FLAC software is reliable. It probably is, if you
have a good version. Or you could do a file compare of the decoded results
and if they differ, verify that it was only the header that changed. (This
might require using an audio editor.) So you see that there would be some
work involved if you make a format change -- work that insures that your
format change was of archival quality. If storage space is not at a premium
and your time is, then your best bet would be to preserve the file in its
original WAV format.
On to the second example. In the second example, the file is a lossy MP3
file, for example. Now some information has already been lost, but if this
is all you have, too bad, you are going to have to work with it. You want to
preserve everything that you do have. Unfortunately, if you convert the MP3
file to WAV you may not be able to get back the exact bits should you
recompress it to MP3. (This may depend on the particular MP3 encoder and
decoder.) The MP3 standard is supposed to tell you how to decode the MP3
file to WAV, but the standard does NOT tell you how to encode a WAV file to
MP3. Different MP3 encoders will add different distortions to an audio
signal and it is possible that multiple passes through MP3 to WAV to MP3 to
WAV will continue to deteriorate the sound quality. If you preserve the WAV
file, then some information will be lost -- and although this information
may not be valuable today, in the future new software may come up with a
better way of decoding your old MP3 file, so you had best preserve the
original input to the library, as it arrived. The good news, is that the MP3
format will be smaller than the WAV anyway, so this will be economical of
your time and of archival storage.
Now, I can't speak about Ogg Vorbis in any detail, because I don't have any
experience with files in this format. So this raises a bit of a red flag.
It gets me thinking. What happens if 100 years from now somebody pulls a
file out of the archive and tries to play it. Will they have Ogg Vorbis
software? Will their Ogg Vorbis software work the same way as the older Ogg
Vorbis software? I think it is highly likely that he will have software to
play MP3 files and WAV files, because these are widespread. But will he
have software for Ogg Vorbis? What to do? The same problem arises with
other formats, such as lossless formats such as FLAC, Apple Lossless,
Windows Media Lossless, etc. Will these formats be available in 100 years.
I am more included to go with FLAC, because it is an open format, but I am
not so certain about it. MP3 is an international standard and I feel more
comfortable about it.
What I would recommend doing, in any event, would be to do ALL of the
following in a belt and suspenders type of approach:
(1) preserve the original format as received unchanged
(2) also preserve a decoded copy(e.g. WAV) of the original media as
insurance that the decoding software may not be able to be run some time
into the future
(3) preserve a copy of the decoding software somewhere in the archive.
If it is only possible to keep one version, then I would preserve the file
format as it was received and take care that the necessary decoding software
is also preserved.
AGL
--------------------------------------------------------------
Also I note an article in this morning's Boston Globe called The Departed:
http://www.boston.com/business/technology/articles/2008/04/07/the_departed/
(The writer makes an amusingly erroneous assertion about 78s, but let it
pass, let it pass.)
clark