Re: The people behind TTS voices
Hi everyone, I am going to revive this thread to hopefully get some sort of clarification on a few things. I'm also working on something a little exciting, I hope.
I saw a PDF file that contained instructions on how to build a speech synthesiser using a Moog Synthesiser module and various electronic manipulatives. Still, I wonder what people do exactly to model someone's voice. For instance, is eSpeak modelled after someone, or is it purely formant and not based on any model? Is there an in-between category between formant and concatenative? For instance, I heard that Eloquence and Microsoft Sam were primarily formant, but they also had some concatenative elements that were processed with digital signal processing as well.
The Wikipedia article indicated that there was a third type of speech synthesis called articulatory speech synthesis, but it didn't describe hot it was made, or at least, it didn't delve into the details of the processes needed to model someone's vocal tract to make a speech synthesiser out of it.
@FlybyChow, you said before you were interested in making your own speech synthesiser using concatenation. I.E. Using small segments of your voice along with some digital signal processing, or have someone model your vocal tract by recording different things to make an analysis. I don't know if there are accessible spectograms or whatever it is they use to replicate someone's voice using tones and noise. Anyhow, look at posts 52 and 57 of this topic. http://forum.audiogames.net/viewtopic.php?id=8877
I found these recordings in the Festvox demo. http://www.festvox.org/history/klatt.html
_______________________________________________ Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com https://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
