Speech Synthesis on VCS

[Cootster]

Not the one I was thinking about, but this is cool, and there is probably an emu out there that can do mic or CD as cassette line in . . . Thanks . . . I'll have to mess around with this.

 

The one I used had a waveform table for each syllable in English and accepted text input. It was way cool . . .

[ZylonBane]

A while back, whilst pondering digital sound on the 2600, I had this thought--

 

What if, instead of the usual volume-only digitizing approach, you broke an audio sample down into 1/60th-second size chunks, then used an FFT to determine the dominant frequency and amplitude of each chunk? Then, once a frame the 2600 could grab a single byte containing packed volume and frequency data and stuff it in one of the sound registers.

 

Yes, the end result would be horribly indistinct and you'd practically have to already know what you're supposed to be hearing to understand it... but then, 2600 graphics are the same way.

[Bryan]

I'd thought about that too. Except it would probably work much better on Pokey. You could actually analyze the frequency spectrum for each of Pokey's distortion types, and include them too if they provide a better match than a pure tone. What you would be doing is something similar to the TMS5220 speech chip (used in arcade games like Gauntlet).

 

-Bry

[ZylonBane]

No matter what, it couldn't sound worse than the speech in the original Castle Wolfenstein game!

[Bryan]

Uh huh.

 

-Bry

[Thomas Jentzsch]

What if, instead of the usual volume-only digitizing approach, you broke an audio sample down into 1/60th-second size chunks, then used an FFT to determine the dominant frequency and amplitude of each chunk? Then, once a frame the 2600 could grab a single byte containing packed volume and frequency data and stuff it in one of the sound registers.

And you could mix both channels for better results.

 

Still, I don't think it will work and I never heared of any software working like this,