see end...----- Original Message -----
From: "Richard Warren" <[log in to unmask]>
> Dear ARSC List People,
> Karl and Mike have raised an interesting subject but one which needs lots
> of careful attention and discussion.
> "Do no harm" or "Do minimal harm" is a good first rule.
> The first level of proficiency suggested seems by far the most important:
> transcription, the aim of which some of us would probably agree is to
> capture the maximum possible of the signal from the original, in analogue,
> digital, or some other format (perhaps yet to be developed). That's audio
> ANYTHING done after that stage must be defined as editing. None of the
> processes mentioned can be called "restoration" unless those attributes
> the original performance can be fully and provably documented (that's
> difficult if not impossible to do, human audio memory being a fleeting
> quality -- people forget sonic features, even simple ones, very rapidly);
> and even the best recording engineers haven't time to document all
> attributes of what they are recording. In most cases with which I'm
> familiar, especially in commercial recording, documentation of what has
> been recorded has been minimal and usually relates to the status and
> settings of the equipment used.
> Denoising nearly always changes one or more qualities of the recorded
> sound, at least to a slight extent.
> Repitching is fine if one can do it, but how can one learn how an original
> was pitched ? Approximation is probably the best that can be expected in
> most cases, especially those involving speech; and one should certainly
> document what has been done in processing and the basis for doing it.
> The working lives of many of us would be much simpler if there were some
> way to learn what the original form of a signal was, but unfortunately
> is usually impossible, even in what may seem to be the most obvious cases.
> There are hundreds of reasons for and examples of this.
> It should be unnecessary to comment on "improvement" on the original;
> anyone who has purchased reissues of recordings has heard both reasonable
> and horrible results of such work.
> No one who wants to call him or herself a sound archivist should be
> concerned about more than 1) achieving the best possible transcriptions of
> originals, 2) the suggesting of possible approximate pitch corrections,
> 3) such noise reduction as does not cause significant changes to the
> Performing adjustments of the types in items 2 and 3 in that list should
> apply to those who need to prepare copies of these transcriptions for
> reference purposes (or at customers' requests for commercial purposes,
> as "pleasing" the listeners).
> Any work beyond item 1 in the previous paragraph is editing, is likely to
> be at least partly subjective, and should be, except for instances of 2 &
> 3, outside the scope of work of an audio archivist; and all of these ideas
> except for transcription follow from the original principle of doing
> minimal harm.
> Unfortunately I don't have time to explain this position fully in an
> message, but the subject is important enough that a brief attempt is
> With best wishes, Richard
> At 09:25 AM 7/20/2004 -0700, you wrote:
> >At 08:34 AM 7/20/2004 -0500, Karl Miller wrote:
> >>Which brings up the question (at least in my mind) what should be the
> >>criteria for certification (assuming this is a good idea) in audio
> >>preservation and restoration.
> >Mmm - a good one!
> >There is a first qualification: to do minimal harm. Equipment, methods
> >experience are such that one expects minimal damage to the source
> >So certification would surely have to be by medium. Unlike medicine,
> >a podiatrist may legally practice psychiatry in most states, in this
> >there would be qualifiers for LPs, "78s", transcription discs, cylinders,
> >wire, and so on.
> >There is a second parameter of qualification, proficiency. That would be
> >similar to Bachelor, Master and PhD levels of conventional academia. The
> >first level would entail transcription (presumably including digitizing)
> >without editing. The second would qualify based on elementary denoising,
> >repitching and otherwise processing to return the signal as nearly as
> >possible to its original form. The third would entail 'improvement' on
> >original: rebalancing, equalizing and so on to provide the preferred
> >listening experience to the customer.
> >The medium parameter can be assessed largely objectively. The first level
> >of proficiency is largely objective, the next is mixed, the last almost
> >entirely subjective. (E.g., from multiple-choice to essay exam. <G>)
What we often forget is that the digitization of sound has transformed
recordings, old or new, into a series of bytes...and that it is only
necessary to redefine selected bytes in that sequence to come up with
an entirely new sound file...which could be as simple as the recording
had it been recorded and played at 78.26rpm, or as complex (if not now,
then in the near future) as the recording had it featured Bix on cornet
and Jimi Hendrix on guitar! All we need to do...and this may be under way...
is to digitally analyze a waveform and see if we can identify typical
waveform elements which sonically define Bix!