SAM v0.00001

After trying Ludojoey’s talking demo, I wondered whether we could implement speech synthesis on Micro-8.I have always been a big fan of the TI-99’s LPC speech synthesis, but LPC is unfortunately too demanding for Micro-8 ( and more complex ) 
then I remembered SAM, the formant-based speech synthesizer. Unfortunately, Micro-8 is too slow to generate the speech in real time, so we have to precalc it—and the available memory limits the length of the speech.
( to save memory, I generate the audio at 4 kHz and output each byte twice )ofc the original SAM is much more complex, but it is fascinating to hear how mixing just 2 sinus + 1 square with some random noise can produce a recognizable voice.

original SAM Source: github.com/s-macke/SAM

 


 

Would be a fun project to try doing it in realtime using the hardware. YM2612 for generating the sine waves and the built-in synth for squares and noise. 

Yes, the YM2612 have a special mode perfect for that, called Composite Sine Wave Modeling. 

some infos about it (page 33 ) :  SEGA FM DRIVE TECH MANUAL