#websockets API for multilingual model not working?

1 messages · Page 1 of 1 (latest)

sage dust
#

Hi there,

It seems that the websocket API does not actually respect the model_id param?

I am trying to generate voices in languages other than english but its clearly just the monolingual english model generating those and it sounds terrible.

URI used:

wss://api.elevenlabs.io/v1/text-to-speech/<VOICE_ID>/stream-input?model_type=eleven_multilingual_v2&optimize_streaming_latency=4&output_format=ulaw_8000

can you please confirm how to connect to a websocket with eleven_multilingual_v2 since this does not work or confirm if the multilingual model only works with /stream endpoint? thanks

#

to clarify this works to get output but in German and French for example, its CLEARLY generated by an English monolingual model