• Register
  • Help
Results 1 to 4 of 4

Topic: live performance by dead performers

  1. #1

    live performance by dead performers

    Zenph studios claims to have perfected a audio to midi system that can capture piano performances flawlessly from old scratchy recordings.

    They say they now have the software not only acurately identifying the polyphonic notes, but also providing precise velocity and timing of the notes.(I would think exact pedalling reproduction would be a challenge, although if one simply holds the notes the full length they sound and avoids pedal use it might be close enough)

    The midi is reperformed on a disclavier, thus allowing one to hear the original performance "live".

    They test the timing by playing the new recording at the same time as the original. If there are even slight timing differences they show up as echos.

    The technology will be unveiled at an upcoming recital:

    Here is an article from new scientist:

    I suppose that playing on a Yamaha might not be faithful to Glen Gould's original performances on a Heintzman. However with a digital piano one could match the character of response of his original piano (ha!, a new form of "original instruments" for critics to argue over). Of course we would all miss Gould's vocalizations( or would that be called singing??). However he claimed he always wanted them removed, but the technicians couldn't do it.

    I find it fascinating that as we move into the future more and more, the past can come into sharper focus. One can restore old media by removing scratches, compensating for artifacts from old recording equipment, etc. However this is a more powerful form of restoration: re-synthesis. It can be done with imagery as well as sound. One could model an actor and set and position to exactly match the original, yet render in high quality on IMAX. In the future this will be the vanguard of restoration.


  2. #2
    Join Date
    Sep 2004
    Poole, Dorset, UK

    Re: live performance by dead performers

    Fascinating! And for those with an interest in how they did it technically I found a short overview article:


    (Couldn't seem to connect to New Scientist site at all, so this may be duplicate information)

  3. #3

    Re: live performance by dead performers

    Yes, this is very interesting indeed. I use to listen to piano rolls converted to midi files. I use them with the best sampled piano libraries and the best convolution software for natural ambience. There are also some excellent recordings of specific Bösendorfer and Steinway concert grands, rebuilt with piano roll technology, performing the old piano rolls, in recordings such as the "Masters of the piano rolls" series (Debussy, Ravel, Gershwin, Mahler).

    But being able to restore old audio recording is something else and much more complicated. One disadvantage here is that they don't get the true original recording, only a midi transcription performed with another grand piano (Yamaha). Also I find it strange that they don't put any audio demos of their work. They just talk about what they can do, not demonstrating it.

    Do you think that a software such as Melodyne can do this also? (it splits up the incoming audio signals into fragments, reminding of midi)

  4. #4

    Re: live performance by dead performers

    That pragmatic programmer article was especially interesting. What a fun project to work on.

    I don't think Melodyn could do this sort of thing without a lot more development.. does it handle polyphonic input? There are several
    auto to midi programs that handle solo melody lines, but to my knowledge
    any that try to do polyphony usually do a bad job.

    I would think that the piano is quite a bit easier to handle than a full orchestra, or non-percussive polyphonic music. The sound of each note at a given velocity level is quite constant. One could first examine the entire recording to derive the tones for the different pitches. This would harder for instruments like the violin, which have a continuously varying output.

    Perhaps some day we will be able to analyze and resynthesize old Toscanini performances. Speech and singing might also be possible. The dialogue in old movies could sound as if it was just recorded. This may seem to be a bit out of reach today, but not that long ago something like the battle scenes in Lord of the Rings would have seemed out of reach. Once the software is written(the hard part.. I think computers are fully powerful enough today) then this will become easy to do, and hence commonplace.


Go Back to forum


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts