Unable to use ARPABET phonetics input

#52
by TaskeenJafri - opened

I am unable to use input like:
{F AY1 R F AY2 T ER0} {AA1 R} {AO1 S AH0 M}
Is there any way to do this on the gradio, or will I need to run it locally?

I don't think that is possible with the model

In the official documentation by Microsoft,
https://www.microsoft.com/en-us/research/project/e2-tts/
it says that they allow ARPABET input.

Hi,
The E2 model is not the official E2 model by Microsoft, but rather a model trained using E2's architecture. Not all features are supported.
See https://github.com/SWivid/F5-TTS/issues/241 for more details.

mrfakename changed discussion status to closed

Sign up or log in to comment