[sdiy] Looking for Speech Synth Chips! Can you help me?

[jbv]

> Now that would be something I would be interested in...something like that
> would be a lot of fun to play with.
>
> At 10:10 PM 9/14/2003 +0000, jbv wrote:
> >That's why I started to work on an emulation of that
> >chip : a VC filer bank, driven by a uC with data about
> >speech formants (and formant transitions) extracted from
> >samples with a phase vocoder... When it's done (if it's
> >ever finished) the thing should be a programable analog
> >singing machine...

Sure...
I actually spent some time wondering what would be the best
choice of design for the VC BP filter bank (especially coz I'm
not an analog wizzard)... I finally decided that the first step
was to check if the concept would work, and decided to prototype
that thing with a set of CEM3350 chips.

In my design I set up a uC with a DAC with multiplexed S&H
driving a set of 3 to 5 VC-BPF + VCA in parallel. This part is
supposed to generated vowel sounds with their formants.
I also included an extra voice with VC-BPF + VCA + VC-EG
to generate consonant sounds such as SSS (line in "silk") or SH
(like in "shame"). Programation of control signals for this part
is much more complex than the vowel part, because transitions
between vowels is quite straightforward, while transitions between
vowels and consonants and/or between consonnants is much
more complex : sometimes the sound is slightly modified by
previous and/or next sound (like with "sh"), but sometimes it
isn't (like with "SSS"), or sometimes the attack-decay is
modified, etc.

For various reasons I had to put this on a shelf for a while.
And these days, I'm wondering if a formant-based design
is still a good idea (at least an analog formant-based design)
and if I should switch to a DSP design, or even to an additive
synthesis design...

JB