search

pyannote / pyannote-audio

Neural building blocks for speaker diarization: speech activity detection, speaker change detection, overlapped speech detection, speaker embedding
http://pyannote.github.io
MIT License
6.41k stars 789 forks source link

Can not reproduce "adapting_pretrained_pipeline.ipynb" on local machine #1698

Closed jyhan03 closed 7 months ago

jyhan03 commented 7 months ago

Tested versions

System information

CentOS 7.9 - pyannote.audio=2.1.1

Issue description

Hi, thanks for the great work! When I follow https://github.com/pyannote/pyannote-audio/blob/develop/tutorials/adapting_pretrained_pipeline.ipynb and run code on Colab, everything is fine. But when I try to run the same code on my local machine, there is always an issue from pyannote-database info AMI-SDM.SpeakerDiarization.mini : UnknownExtra: typer 0.12.3 has no such extra feature 'all'.

I tried to install the same version of python/pip/pyannote as the Colab, but the problem is still there. Any suggestions to solve it? Thanks.

Minimal reproduction example (MRE)

https://github.com/pyannote/pyannote-audio/blob/develop/tutorials/adapting_pretrained_pipeline.ipynb

hbredin commented 7 months ago

Try updating pyannote.database to 5.1.0.

jyhan03 commented 7 months ago

Try updating pyannote.database to 5.1.0.

Solved. Many thanks.

jyhan03 commented 7 months ago

Sorry to bother you again. I found the tutorial works when I use a single GPU, but it fails when running with 2 GPUs: _pickle.PicklingError: Can't pickle <class 'pyannote.database.registry.AMI-SDM'>: attribute lookup AMI-SDM on pyannote.database.registry failed

Do you have any suggestions?

jyhan03 commented 7 months ago

Maybe there are some incompatible issues. Multi-GPU works for pyannote3.1.