felixbur / nkululeko

Machine learning speaker characteristics
MIT License
31 stars 5 forks source link

Add speaker embedding from speechbrain #78

Closed bagustris closed 1 year ago

bagustris commented 1 year ago

This PR related to #73 and #43.

Supported features= "spkrec-xvect-voxceleb", "spkrec-ecapa-voxceleb", "spkrec-resnet-voxceleb"

It works on existing experiments (such as in test; just change the feat with the feature above. But it doesn't work with target=speaker task. See comment in #73