search

clovaai / voxceleb_trainer

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

Generate embeddings of a long audio file #43

Closed Joepetey closed 3 years ago

Joepetey commented 4 years ago

So I want to use this to generate embeddings of a long audio file, and was wondering what your recommendation for that is?

joonson commented 4 years ago

To get a single embedding for a long speech segment, use the argument --eval_frames 0.