Update Huggingface Hub version

[philschmid]

What does this PR do?

This PR updates the huggingface_hub package to avoid conflicts with libraries that use >0.9

[Majdoddin]

This is also important to resolve a version conflict with openai/whisper reported here

I doesn't seem that your commit is enough to resolve the issue, I just got an error:

!pip install git+https://github.com/pyannote/pyannote-audio.git@refs/pull/1098/merge

from pyannote.audio import Pipeline
pipeline = Pipeline.from_pretrained('pyannote/speaker-diarization')

---

ValueError Traceback (most recent call last)

in
1 from pyannote.audio import Pipeline
2
----> 3 pipeline = Pipeline.from_pretrained('pyannote/speaker-diarization')

1 frames

/usr/local/lib/python3.7/dist-packages/pyannote/audio/core/pipeline.py in from_pretrained(cls, checkpoint_path, hparams_file, use_auth_token, cache_dir)
104
105 if "params" in config:
--> 106 pipeline.instantiate(config["params"])
107
108 if hparams_file is not None:

/usr/local/lib/python3.7/dist-packages/pyannote/pipeline/pipeline.py in instantiate(self, params)
442
443 msg = f"parameter '{name}' does not exist"
--> 444 raise ValueError(msg)
445
446 self.initialize()

ValueError: parameter 'segmentation_onset' does not exist

[codecov]

Codecov Report

Base: 30.20% // Head: 30.20% // No change to project coverage 👍

Coverage data is based on head (79e1df6) compared to base (b3d9372).
Patch has no changes to coverable lines.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #1098   +/-   ##
========================================
  Coverage    30.20%   30.20%           
========================================
  Files           61       61           
  Lines         3751     3751           
========================================
  Hits          1133     1133           
  Misses        2618     2618           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[crazy4pi314]

Is there something blocking the macos build? This may be related to an issue downstream in openai/whisper: openai/whisper#264

[hbredin]

Will be part of next release. No ETA, though.

[philschmid]

I created a fork on the 2.0.1 release, which patches the huggingface_hub version: https://github.com/philschmid/pyannote-audio/tree/hub0.10

you can install it with pip install git+https://github.com/philschmid/pyannote-audio.git@hub0.10 --upgrade

[lukas681]

Thanks @philschmid! I can confirm that this solves the entire issue.

One addendum from my side:
@Majdoddin: I was facing this problem as well when I tried this simple fix here and installed it via pip directly from the commit. So there should not have been any differences between the hard fork of @philschmid. After I first got this segmentation_onset stuff I tried to fix it in the source local code directly as another workaround by deleting this specific property, but it not really worked. Creating a new venv and installing the correct versions for my CUDA device (especially pytorch + CUDA level 11+/RTX 3070) from scratch did allow me to have the coexistence of whisper and pyannote-audio without the workaround of postponing the installation. So maybe deleting all dependencies could also fix it on your side.
This is my environment, maybe it helps:
requirements.txt:

spacy
whisper
spacy-sentiws
numpy
tqdm
more-itertools
transformers>=4.19.0
ffmpeg-python==0.2.0
# pyannote.audio Using development branch
webvtt-py
pandas
pandas-compat

environment creation (Make):

venv/bin/activate: requirements.txt
    python3 -m venv venv
    venv/bin/pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116
    ./venv/bin/pip install git+https://github.com/philschmid/pyannote-audio.git@hub0.10  --upgrade
    ./venv/bin/pip install -r requirements.txt

[Majdoddin]

Thank you for the clean solution @lukas681