GE2E for Speaker_Verification

• 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.

Result

• similarity matrix looks like

• use N=20 speakers and M=10 utterances for training
• my purpose is to get the speaker utterances' d-vector