Tiny Speech Synth by Stan 1901
Type: C++ subroutine Created by: Stepanov Andrey, 2008 ( ICQ: 129179794, e-mail: andrewstepanov@mail.ru ) Description =========== This is a simple speech synth subroutine, based on formant synthesis theory. Speech is synthesized by passing source excitation signal through set of formant one-pole filters. Excitation signal is a sawtooth or noise (depending on sound type), although you can try other signals. Some explanation on design ========================== It was surprisingly hard to find any information on formant synthesis. Of course, there is a lot of information on the Internet, but the biggest part of it is just a general words of theory, nothing practically usable. The biggest problem was to find concrete formant frequencies. After reading a lot of articles I found that I knew nothing new, compared to my knowledge before reading... So I did frequency analysis by myself. I recorded my own voice and analyzed it's FFT. Analysis is VERY approximate and incomplete, so it causes bad speech quality. Terms of use ============ This program is free for use, and you can use it for any purposes you want, as long as you specify my name in you program source code and description.
[ back to the prod ]