akhbar

/

chatterbox-tts-norwegian

like 10

Hei.
Ja, jeg er Norsk. Hva kan jeg hjelpe deg med?

...

Hei akhbar. Testet denne finetunen nå og den hørtes veldig bra ut! Kan du dele noe om hvilken pipeline du bruker, og hvor mye data du har brukt for å komme lage den?

I'll answer in english in case non-Norwegians stumble in here.

I trained the model using one of the finetuning repositories available on github. If I remember correctly, there are two: One with "standard" finetuning, and one that uses GRPO. I used the one without GRPO.
The model was trained on roughly 6500 hours of data for one epoch.

Thanks for the quick answer!