Running without "wespeaker-voxceleb-resnet34-LM"

[doublex]

Is it possible to run version 3 without wespeaker-voxceleb-resnet34-LM?

[github-actions]

Thank you for your issue.You might want to check the FAQ if you haven't done so already.

Feel free to close this issue if you found an answer in the FAQ.

If your issue is a feature request, please read this first and update your request accordingly, if needed.

If your issue is a bug report, please provide a minimum reproducible example as a link to a self-contained Google Colab notebook containing everthing needed to reproduce the bug:

• installation
• data preparation
• model download
• etc.

Providing an MRE will increase your chance of getting an answer from the community (either maintainers or other power users).

Companies relying on pyannote.audio in production may contact me via email regarding:

• paid scientific consulting around speaker diarization and speech processing in general;
• custom models and tailored features (via the local tech transfer office).

This is an automated reply, generated by FAQtory

[hbredin]

You can use any supported speaker embedding in place of hbredin/wespeaker-voxceleb-resnet34-LM.

See pyannote.audio.pipelines.speaker_verification.PretrainedSpeakerEmbedding for a list of supported models.

It should be as simple as downloading this configuration file, updating this line and that line, and then load it with

from pyannote.audio import Pipeline
pipeline = Pipeline.from_pretrained("config.yml")

Note however that the reason why I chose this speaker embedding for version 3.0.0 is that it is the one that gave me the best result. I'd love to hear the reason why you would like to change it? Research? Any other reasons?

[doublex]

@hbredin
Thanks for your answer.
onnxruntime is slow (runs on the CPU). There is a dropin-replacement onnxruntime-gpu which uses CUDA - but using onnxruntime is cumbersome....

[hbredin]

Would switching to onnxruntime-gpu solve the issue?
Does it also support CPU?

[doublex]

Yes - but it does not support running on the CPU - 😠😠😠😠
My knowledge is limited.

[hbredin]

Could you try with this and let me know if that allows to run on GPU on your side?

pip install https://github.com/pyannote/pyannote-audio/archive/refs/heads/fix/onnxruntime-gpu.zip

All it does is switch from onnxruntime to onnxruntime-gpu which does seem to also support CPU

[guilhermehge]

@hbredin how would I know which threshold to use for each model?

[hbredin]

You would not know. You have to tune this threshold.
You could also use pyannote/speaker-diarization-2.1 configuration file as a reasonable starting point.

[hbredin]

Closing as latest version no longer relies on ONNX hbredin/wespeaker-voxceleb-resnet34-LM and this solves the original issue.

Please update to pyannote.audio 3.1 and pyannote/speaker-diarization-3.1 (and open new issues if needed).