51–60 of 383 posts
Hmm the quality is not so impressive. I'm looking for a really naturally sounding model. Not very happy with piper/kokoro, XTTS was a bit complex to set up. For STT whisper is really amazing. But I miss a good TTS. And I don't mind throwing GPU power at it. But anyway. this isn't it either, this sounds worse than kokoro.
Foundational tools like this open up the possiblity of one-time payment or even free tools.
Web version: https://clowerweb.github.io/kitten-tts-web-demo/ It sounds ok, but impressive for the size.
> Error generating speech: failed to call OrtRun(). ERROR_CODE: 2, ERROR_MESSAGE: Non-zero status code returned while running Expand node. Name:'/bert/Expand' Status Message: invalid expand shape Doesn't seem to work with thai.
Earlier quoted context omitted.
If you're looking for other languages, Piper has been around in this scene for much longer and they have open-source training code and a lot of models (they're ~60MB instead of 25MB but whatever...) https://huggingface.co/rhasspy/piper-voices/tree/main
[flagged]
Earlier quoted context omitted.
Does anybody find it funny that sci-fi movies have to heavily distort "robot voices" to make them sound "convincingly robotic"? A robotic, explicitly non-natural voice would be perfectly acceptable, and even desirable, in many situations. I don't expect a smart toaster to talk like a BBC host; it'd be enough is the speech if easy to recognize.
This one is at least an interesting idea: https://genderlessvoice.com/
say is only 193K on MacOS ls -lah /usr/bin/say -rwxr-xr-x 1 root wheel 193K 15 Nov 2024 /usr/bin/say Usage: M1-Mac-mini ~ % say "hello world this is the kitten TTS model speaking"
That’s not a far comparison. Say just calls the speech synthesis APIs that have been around since at least Mac OS 8. That being said, the ‘classical’ (pre-AI) speech synthesisers are much smaller than kitten, so you’re not wrong per se, just for the wrong reason.
Web version: https://clowerweb.github.io/kitten-tts-web-demo/ It sounds ok, but impressive for the size.
I tried to replicate their demo text but it doesn't sound as good for some reason. If anyone else wants to try: > Kitten TTS is an open-source series of tiny and expressive text-to-speech models for on-device applications. Our smallest model is less than 25 megabytes.