Hacker News new | past | comments | ask | show | jobs | submit login

Since I've been playing around with Piper Text-to-Speech & the associated LibriTTS voice model, I couldn't resist:

* https://rancidbacon.gitlab.io/piper-tts-demos/#various_radio...

If the six speakers I selected for the demo don't match your taste in DJs, there's around 900 more in that voice model to try... :D

(The 3 audio players differ only in file format & whether I ran the output through normalization.)

Also, I was pretty impressed/surprised at the quality of their pronunciation of the meta-syntactic variables. :)




They all sound out of breath. Not sure what causes that effect/perception. Listening to it again, it seems like the plosives (like d, or t) have no pop like there was no air pressure behind them.


Ha, that’s awesome. Several of the voices sounded just right for radio. I see the lib is limited to 24khz, maybe why it has an almost AM radio sound?




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: