On 07/10/2008 23:27:07, Grammostola Rosea wrote:
I have not had occasion to use any form of music synthesiser that is a
realistic approximation of the human voice.
There is an addional complexity to the voice in that it usually has
words to sing so for a synthesiser to sing it has to do the usual thing
of interpreting MIDI to pick the right sample and make sure it plays at
the right pitch and volume with the right envelope while at the same
time doing the speech synthesiser thing of chosing the correct vowel
formants and the right kinds of noise for the consonants. Even then it
would sound very artificial.
Given the intention of this being for practice I'd forget about using a
realistic voice sound. Many singers are quite used to being played
their part on a piano so the tone of that would be familiar. If you
want a closer approximation and, in particularly something that lacks
the attack of a piano and has more sustain, like the voice, one of the
woodwind family would be good.
Linux-audio-user mailing list