Hi,
very cool!
I don´t have a MBFM and dunno if it is already/at all possible, but what might be an option would be to have a dedicated formant sequence storage, that stores the center frequencies and other mbfm synth operator parameters and then is played back over time, just like the formant sequences of the FS1R from Yamaha (which has 128 to 512 playback steps per formant sequence, each step contains center frequency, volume, and formant-shape which equals the "width" of the spectrum emitted).
As data entry is not viable or really a lot of work, one could maybe use an existing formant converter, this one is in flash:
http://blog.zacharch...ditor-in-flash/
"well, excuse me, princess" <- very nice :-)
The converter is nice, because it also visualizes the formants and so explains a difficult procedure.
When looped, these sequences are able to create completely new (and evolving) un-fm-ish sounds, have a look at the "p11 Uuul-ha" preset for example.
The flash tool generates files for the FS1R (8 "voiced" + 8 "unvoiced" operators), but in my tests it already sounded quite good using only 3 or 4 "voiced" operators (all additive, flat fm algorithm).
I guess the generated sysex files may be parsed to strip excess operators and unavailable fm operator parameters for the opl3. If this would be realized, the very nice and professional formant sequences created by yamaha 12 years ago would be available for MBFM users...
Best regards,
Peter