Is that possible to specify multiple languages in whisper transcriptions api?

I am using whisper model for recognition. I hope I can specify multiple languages in API for more accuracy and less prediction. I am sure the voice is English or Chinese but not other language. (if I don’t specify the language, sometimes whisper detected the language wrongly)

following is my code:
“”"
transcription = await self.openai_client.audio.transcriptions.create(
model=self.STT_MODLE,
file=(self._get_new_file_name(),audio_binary),
language=“zh,en”)
“”"
But I got error : Error code: 400 - {‘error’: {‘message’: “Invalid language ‘zh,en’. Language parameter must be specified in ISO-639-1 format.”

So I wonder if API support multiple languages specify? or it has to be one.
If it has to be one right now, any plan openai to support multiple languages in future?

1 Like