The Yamaha AN1x Synthesizer mailing list

> Here Giblet has gone and poked his face right out of the window into the
> fog.
> 
> "You" (who? scientists maybe?) can make "anything" (what? methane, a Casio
> watch, maybe your car horn?) with a programmable 'beep' (like your car?)
> simulate speech (what language?).

'You' == anyone who knows how to modulate speech with an oscillator. There
are lots of white-papers and books on the subject. A PC speaker is driven by
a square wave generator whose only programmable parameters are duty cycle
and frequency. If one can make THAT talk, then one can make a dog's
squeak-toy talk if one is persistent enough.

No algorithm is required at the an1x end -- only at the PC end.

While the interface to an AN1x's oscillators is indirect and primitive
compared with the PC's directly-programmable oscillator, it can't be all
THAT difficult to synthesize short waveform sequences on a PC and transfer
those to the synth. I've played with speech synthesis under DOS (I *am* a
scientist) on a PC but maybe I should shut up until I actually DO it on an
an1x... Don't hold your breath on that - it just isn't worth the effort.

There are several TTS examples available on the web, in assembly and C
languages. One can use them to observe the waveforms of those words one
wishes to duplicate on the an1x and 'simply' duplicate the waveform. (What
could be simpler, eh?)

This would be an interesting exercise, but a total waste of time. A Triton,
or any other sampler, can talk its fool head off right out of the box.