How to set whisper language

[behzadCube]

Hi
I trying to lunch jitsi meet with live transcription with whisper (skynet)
I set whisper model to the "large-v3" (supported for Persian Language )
but don't know who to set participant meet language to the "fa" Persian (in skynet) ?
is there any environment variable ?

[rpurdel]

Hi Behzad,

The language code needs to be provided in the audio chunks sent to the server, for each participant. Have a look at the data format here https://github.com/jitsi/skynet/blob/master/docs/streaming_whisper_module.md#data-format and at the code examples below.

Also I would suggest using the large turbo model as it greatly speeds up transcription.