ElevenLabs doesn't work properly for Japanese
Opened this issue · 4 comments
blakedgross commented
Currently the ElevenLabs implementation does not seem to pass the correct parameters to have Japanese work correctly with their v2 API, where it works properly on the ElevenLabs site. I used this string 大きい地震である with the English, Female, Rachael multilingual and it ended up parsing it as a mix of Japanese and Mandarin.
luc-vocab commented
Do you get the proper, expected output when using the ElevenLabs website ?
blakedgross commented
Yes
…On Wed, Sep 13, 2023 at 7:29 AM Luc ***@***.***> wrote:
Do you get the proper, expected output when using the ElevenLabs website ?
—
Reply to this email directly, view it on GitHub
<#299 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABMKF246TRZEKBON4Q5Z5S3X2GKKBANCNFSM6AAAAAA4VMK6NI>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
luc-vocab commented
please give me more info so I can reproduce:
- input: 大きい地震である
- what you're getting right now (please romanize in a way i can understand)
- what you expect to get instead
blakedgross commented
What I'm getting now: da4 ki di4zhen4 de aru
What I expect to get: Ō kī jishin de aru
Here is a link to the mp3's of each: https://drive.google.com/drive/folders/1SFXxOoJbtqVkzDKSJvCkfENKS5pmNBsX?usp=share_link