The Eat at Joe's Kawai K5000 Message Board Digest
Separating Formants from Harmonics


------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------


Harmonics and formants ....
Sunday, 12-Apr-98 05:09:08 

     194.134.10.25 writes:

     I own my K5Ks for a while now, but I haven't been able to spend to much time
     creating my own sufficiently good patches.
     But i did think about how I would do a re-syntesis, but came up with some
     questions and remarks:

     In case you are using sounddivers fourier import function on a single sample,
     no distinction is made between variable harmonic properties which shift with
     frequency and formant-like properties witch can be stationary (i.e. not
     shifting with frequency).
     As I observed, there will be the munchkinisation effect as if you would use
     a sampler without multi-sampling.

     Maybe it is possible to mimic multi-sampling on a K5K by using more additive
     waveforms and assign them to key ranges. But this would be a quick and dirty
     work-arround, not a good solution.

     The best thing to do however would be to analyse several samples of an instrument
     and determine what is the variable harmonic contents and what amplitude
     effects are the result of the stationary part.
     I realise this is easier said then done .... ;-)

     The approach of Leiter in the "Xenopluk" patch is probably a step ahead of this
     which resembles virtual synthesis.


     Does anyone know some literature or have ideas how to deal with these kind of
     things?


     Bart

     worp@euronet.nl 

------------------------------------------------------------------------------------------

Re: Harmonics and formants ....
Sunday, 12-Apr-98 13:46:49 

     192.28.2.19 writes:

     Apparently finding formants is difficult even for the academics in this field of
     study. I think you get a publication out of it when you find one, or something.

     From what I've looked at so far, tho, it seems like real instruments tend to
     have a few typical waveform generators; square or pulse, triangle or uneven
     triangle, or triangular pulse; coupled with a resonant body that has a few
     sharp peaks in the low range and closer and shallower peaks as you go up. Of
     course, a generic theory gets you a generic instrument, but that's a start.

     Fletcher and Rossing, "The Physics of Musical Instruments" is full of good
     information, but it requires some physics background. It can take some hunting and
     head-scratching to get an answer out of it.

     There is a database of instrument frequency profiles at:
     http://sparky.parmly.luc.edu/sandell/sharc/SHARC.homepage.html
     I think this has some average frequency profiles that can be helpful in setting
     the formant filter.

     leiter@skypoint.com 


------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------