Set language when using model.transcribe rather than whisper.decode

[Zoobdude]

As the first 30 seconds of my audio file the auto detected language is incorrect therefore I need a way to set it manually. I understand I can achieve this if I use the more low level whisper.decode. However, my audio file is far longer than 30s. Is their a way I can either use model.transcribe and specify the language as an argument. Or can I use whisper.decode on files longer than 30s. Thanks.

[glangford]

Is their a way I can either use model.transcribe and specify the language as an argument.

result = model.transcribe(your_file, language="en", verbose=False) # for example

See also

• Specifying the language within Python #745

[Zoobdude]

Thanks very much.

[chanheehi]

thanks for your comment