voxsrc20_ask_00-How to understand intra-class and inter-class distance in speaker recoginition?

ID = voxsrc20_ask_00

Status: closed

Question

How to understand these two distances?

Since open-set speaker recognition is essentially a metric learning problem, the key is to learn features that
have small intra-class and large inter-class distance.[1]

Answer

1. Each speaker is a class.
2. Each utterance is a object
3. Small intra-class distance: all utterances of a speaker forms a speaker’s space,
   this space should be impact, small and discriminative.
4. Large inter-class distance: different speaker spaces should not be overlapped,
   this inter-class distance should be large

Reference

[1] In defence of metric learning for speaker recognition, arXiv:2003.11982