search

clovaai / voxceleb_trainer

In defence of metric learning for speaker recognition
MIT License
1.03k stars 272 forks source link

New to this field #72

Closed sbkim052 closed 3 years ago

sbkim052 commented 3 years ago

I am new to this field and have a simple question. When testing the pre-trained model with my own data of the input format (label(0/1), wav1, wav2), how can I get the predicted label?

By the way, this work is quite interesting and impressive.

Shane-pe commented 3 years ago

The output is not label but negative Euclidean distance, e.g., -0.52.