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
5.95k stars 756 forks source link

Loss is not decreasing. A warning message is showing continuesly. #764

Closed Kashif-Inam closed 2 years ago

Kashif-Inam commented 2 years ago

Hi, I am trying to fine-tune the speaker diarization pipeline as the instructions given. But my loss is starting from 0.8579 and remains contant for next 600 iterations. Also continuesly showing a warning message that:

/home/ai/env/transformers/lib/python3.8/site-packages/sklearn/cluster/_affinity_propagation.py:144: UserWarning: All samples have mutually equal similarities. Returning arbitrary cluster center(s).
  warnings.warn(

My training .rttm file is look like:

SPEAKER audio01 1 0.56 2.9600000000000004 <NA> <NA> agent <NA> <NA>
SPEAKER audio01 1 3.5559999847412116 3.2500000000000004 <NA> <NA> customer <NA> <NA>

My training .uem file is look like:

audio01 1 0.56 94.10799804687501

And my training .lst file is look like:

audio01

I am not sure that something is wrong with my dataset or maybe I have missed something before training. Kindly help me out.

stale[bot] commented 2 years ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.