change language_detection_threshold type to float

[extrange]

language_detection_threshold can't be supplied as None as detect_language compares it with a float.

Change parameter type in WhisperModel to reflect this.

Added test to confirm this behavior.

[MahmoudAshraf97]

Hello, what is the use case where it needs to be supplied as None?

[extrange]

There's no use case for None, so the Optional has been removed. Furthermore, supplying None raises a TypeError as the test shows. So, we shouldn't allow None to be supplied for type safety.

[MahmoudAshraf97]

I see your point, thanks for noticing
for consistency can you also review other occurrences of this case? these are some examples

faster-whisper/faster_whisper/transcribe.py

Line 298 in 8327d8c

language_detection_threshold: Optional[float] = 0.5,

also I don't think the test is needed, the number of invalid input combos is too large to have tests for all of them, what do you think?

[extrange]

Sure, I'll have a look in a bit!

…

On Wed, 4 Dec 2024 at 19:00, Mahmoud Ashraf ***@***.***> wrote: I see your point, thanks for noticing for consistency can you also review other occurrences of this case? these are some examples https://github.com/SYSTRAN/faster-whisper/blob/8327d8cc647266ed66f6cd878cf97eccface7351/faster_whisper/transcribe.py#L298 also I don't think the test is needed, the number of invalid input combos is too large to have tests for all of them, what do you think? — Reply to this email directly, view it on GitHub <#1188 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AG7SUH3PYPH6G46ADVZIDIT2D3OGDAVCNFSM6AAAAABS7VYY6WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMJWHE4DENRRGU> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[kenho211]

Hello, what is the use case where it needs to be supplied as None?

From openai/whisper#676, the use case of None should be to conduct a majority vote for however many segments allowed to determine the top language.
For the current implementation, we can use language_detection_threshold = 1 to do the same.