elevenlabs / elevenlabs-python

The official Python API for ElevenLabs Text to Speech.
https://elevenlabs.io/docs/api-reference/getting-started
MIT License
2.04k stars 232 forks source link

Add Phoneme support to new models #308

Open bsullins opened 2 months ago

bsullins commented 2 months ago

https://elevenlabs.io/docs/api-reference/how-to-use-pronunciation-dictionaries

"Alias tags are supported by all models. Phoneme tags only work with the models eleven_turbo_v2 and eleven_monolingual_v1. If you use phoneme tags with other models, they will silently skip the word."

With monolingual_v2 and multilingual_v2 models I'm seeing better (more human-like) results than turbo_v2, however for my use case I have many words which require custom pronunciations. Currently this does not work.

As a user I would like to use a custom IPA dictionary with the 'eleven_monolingual_v2' and 'eleven_multilingual_v2' libraries so I can produce more human-like audio for my app.

for example:

<lexeme>
    <grapheme>Kia</grapheme>
    <phoneme>/ˈkiːɑ/</phoneme>
</lexeme>

θæŋks!

bsullins commented 2 months ago

lharries Any thoughts on this one?