Where can I get some help with formant-based speech synthesis?

[Tony]

Hi
I'd like to embed some basic speech synthesis in a project. I know it's been done with
PICs and AVRs, but I need to shoe-horn it into an MSP430 with an already-fixed audio
system, so I need to understand it all. It'll be reciting fixed phases, so I'd like some
simple open source Windows-based preprocessor (to phonemes or allophones or whatever), and
just render the result in the MCU.
Any ideas?
Cheers,
Tony

 

[Cem Uzunoglu]

Hi
I'd like to embed some basic speech synthesis in a project. I know it's been done with
PICs and AVRs, but I need to shoe-horn it into an MSP430 with an already-fixed audio
system, so I need to understand it all. It'll be reciting fixed phases, so I'd like some
simple open source Windows-based preprocessor (to phonemes or allophones or whatever), and
just render the result in the MCU.
Any ideas?
Cheers,
Tony

Here is the deal with the vowels: Every vowel has distinct frequencies
called formants. You just have to find a formant table for the language
you wish to implement and put at least 3 high Q band-pass filter in
parallel with center frequencies set to formant frequencies in front of
a complex waveform oscillator.

Each consonant seems to have a different synthesis method though. You
have to look it up for the language of your choice. The one I
implemented was Turkish.

Cheers,
Cem