ufal / whisper_streaming

Whisper realtime streaming for long speech-to-text transcription and translation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

language detection

Gldkslfmsd opened this issue · comments

If language=None in faster-whisper, the language is detected from source:

https://github.com/SYSTRAN/faster-whisper/blob/f144e4c83d54f3c3304b6a75a3f563e5f84de6cf/faster_whisper/transcribe.py#L333C1-L346C42

It can be used in Whisper-Streaming -- first test it whether it's reliable, how much source second does it need, whether it's reasonable to run it once and keep the detected language, or rerun it every time...

not tested much, but seems fine.
It might have an impact on latency. There can be a rule to avoid repetitive detection.