Update speaker_verification.py for better use of the onnxruntime

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

6.38k stars 784 forks source link

Update speaker_verification.py for better use of the onnxruntime #1693

Open CaioMizerkowski opened 7 months ago

CaioMizerkowski commented 7 months ago

Improve performance using onnx and remove the warming:

[W:onnxruntime:, session_state.cc:1166 VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.

hbredin commented 7 months ago

Thanks.

Comment says "improve performance using onnx". Could you please provide us with reproducible numbers?
Genuine question: can you please document what this change does exactly?