- only text dependent speaker vertification
- code reference from
https://github.com/JanhHyun/Speaker_Verification - use datasets LibriSpeech (http://www.openslr.org/11)
- there maybe exists errors, hope you can help to improve it.
- similarity matrix looks like
- use N=20 speakers and M=10 utterances for training
- my purpose is to get the speaker utterances' d-vector