Lantian Li 2015-09-21
1. Go on my three deep speaker embedding tasks:
1). large-scale deep speaker vector framework -- hold.
2). knowledge transfer for i-vector -- working with Zhiyuan Zhang.
The training loss does not converge.
3). derive binary i-vectors using Hamming distance learning
make summary and write paper for ICASSP.
4). Start my new task: metric learning.
The linear transform metric is better than non-linear one.
Besides, make summary and write paper for ICASSP.
1. Go on the 3) and 4) of the task 1.
2. Prepare paper for ICASSP.