Thread 105560551 - /g/ [Archived: 1223 hours ago]

Anonymous
6/11/2025, 3:58:55 PM No.105560551
1749650315299
1749650315299
md5: b62a1d9e5a56f49d934eed00f07318a3🔍
I found this TTS program from 2003-2018
https://github.com/festvox/flite
It fast on a CPU, and it generates pretty good audio
https://voca.ro/16czryMpnexI

I'm wondering if anyone here uses other TTS programs.

I tried several models with https://github.com/coqui-ai/TTS and they were all less consistent than this humble little Flite program. Not to mention, the also consumed wayy more resources. Flite can work on mobile.
Replies: >>105560674 >>105560920 >>105560967 >>105561017
Anonymous
6/11/2025, 4:13:28 PM No.105560674
>>105560551 (OP)
>No build and requires outdated C
WHY
Replies: >>105560810 >>105565068
Anonymous
6/11/2025, 4:33:27 PM No.105560810
>>105560674
modern compilers can still compile 1980s C. there is no such thing as outdated C.
Replies: >>105561065
Anonymous
6/11/2025, 4:47:05 PM No.105560920
>>105560551 (OP)
This reminds me of studying vocal formants for speech synthesis a few years ago, the readme doesn't say much besides they used diphones
Anonymous
6/11/2025, 4:51:40 PM No.105560967
>>105560551 (OP)
>I'm wondering if anyone here uses other TTS programs.
dectalk
Anonymous
6/11/2025, 4:59:19 PM No.105561017
>>105560551 (OP)
piper is pretty much perfect, and fast even on cpu
kokoro is even better
Replies: >>105561104
Anonymous
6/11/2025, 5:03:12 PM No.105561049
aplikacja jest?
Anonymous
6/11/2025, 5:05:14 PM No.105561065
>>105560810
not by default
Anonymous
6/11/2025, 5:10:58 PM No.105561104
>>105561017
damn, Kokoro sounds great. I found this nice wrapper for it too
https://github.com/remsky/Kokoro-FastAPI

cheers
Anonymous
6/11/2025, 6:02:34 PM No.105561576
Neat thread compared to all the glASS threads. I need to leave this place, or update my post filters...
Anonymous
6/11/2025, 9:27:47 PM No.105563622
zimp
Anonymous
6/11/2025, 11:40:09 PM No.105565059
1749517594657998
1749517594657998
md5: 6291a5e10877393c140813a834429488🔍
I preferred speech synthesis before AI. We could create cool formant stuff.
Anonymous
6/11/2025, 11:41:08 PM No.105565068
>>105560674
RENT
FREE