At 04:36 PM 6/8/2005 -0500, Karl Miller wrote:
>Does anyone know any of the technical side of this. I recall having a
>discussion once with John Eargle. He thought it would be possible to take
>the fundamentals of each key stroke on a piano recording, measure the
>relative amplitudes, adjusting for their frequency and then using that
>information to create a midi you could have a pianist playing on
>whatever keyboard which could be controlled with the file created.
>Certainly the dynamic range of the recording would be a limiting factor.
>Then would Ansorge (a Liszt pupil whose rolls we are working on...who also
>made disc recordings) play it the same way if he was at a Steinway, versus
>a Yamaha, versus...
>As for Glenn Gould without the singing...would it be Glenn Gould without
>the singing?

My interpretation of the limited information is that the parameters used in
a Vorsetzer (hopefully, extended and refined) are extracted from the audio
recording. Those parameters are used to drive a piano. Correcting dynamic
range should be far less of a problem than determining pedaling and other
degrees of freedom whose effects may be too subtle for the noise level of
the source.

Your point about the choice of instrument is certainly valid. After all, a
knowledgable listener can easily distinguish a Boesendorfer from a Steinway
from a Yamaha - surely a pianist would adapt his performance to the tool at

As for Gould without the vocalism, it will be a fascinating experiment when
run. I'd like to watch the reaction of a knowledgable audience exposed to
one of his recordings re-created in this fashion. Would the pianist be
recognized? Would the specific performance be recognized?

[log in to unmask]