Problem about Speaker Verification test batch size

I was using the speaker recognition scripts in the directory “speechbrain/recpies/VoxCeleb/SpeakRec”. I have trained the ECAPA-TDNN speakerRec model and try to test the performance of the model.

However, I surprisingly found that when i changed the batch size in yaml file, i got different scores of the same pair enrol.wav and test.wav in text.txt. What 's more, when i set only one pair in the test.txt, i always got the “-1.0” score using CosineSimilarity function.

Why did the scores of the same pair change with different batch size? It seems that the same audio file can get different embedding in speakerRec model, is it reasonable?

Wish for your result! i can’t thanks anymore!

Hi, we currently have an ongoing PR to fix this issue. I guess it should be addressed in the upcoming weeks :slight_smile:

Okay, wait for your good news! Thank you!

Was this fixed by now?

Hi, not yet. This is a very very very large Pull Request.