Issues with pip package (missing attribute detect_language_multi_segment)

[treya-lin]

Hi, thanks for your great work.

I'd like to report that when I am trying this package, I noticed a weird thing. When I install this package with pip install, the code it installed missed some attributes in WhisperModel, so I couldn't run language_info = model.detect_language_multi_segment(wavpath) because model has no attribute detect_language_multi_segment, it happened both with version 1.0.0 and the upgraded 1.0.3.

In the github repo I can see that it does have the attribute, but when I entered the package code installed by pip, it does not contain this attribute (and several other attributes are missing too)

Finally I uninstalled the pip installed package, pull the github repo and ran pip install -e ., then this time, it is correctly install and I can use the detect_language_multi_segment attribute.

I am not sure if it was because I was using a pip mirror and that mirror was not updated properly (but the version number is correct, so I think it is strange enough). Maybe the author could check if the pip package is updated with all the attributes? And if anyone run into the same issue, just install the package from this repo, it shoule be able to fix it.

[Jiltseb]

Its not an issue with the pip package, rather this change is yet to be released as a Pypi package. The best would be to install from source and use it:
pip install git+https://github.com/SYSTRAN/faster-whisper.git