(差异) ←上一版本 | 最后版本 (差异) | 下一版本→ (差异)
Last week:
1. with the help of Lantian, basically realised dnn/lstm for learning i-vector, while without much performance;
This week:
1. find out if it's possible for nnet to learn from i-vector;
2. or guide nnet directly with metric loss function.
