X-SepFormer
Paper
X-SepFormer: End-to-end Speaker Extraction Network with Explicit Optimization on Speaker Confusion
Introduction
This is a demo for our paper X-SepFormer: End-to-end Speaker Extraction Network with Explicit Optimization on Speaker Confusion.
We list some speech examples of our baseline system, and compare our proposed methods with the baseline.
If you have any questions or suggestions, please contact liukai89@huawei.com
Examples
We divided examples into three categories:
-
Female-Male Mixtures
-
Male-Male Mixtures
-
Female-Female Mixtures
For better display, we divided each category into 2 groups.
1. Female - Male Mixtures (Group 1 & 2)
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
2. Male - Male Mixtures (Group 1 & 2)
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
3. Female - Female Mixtures (Group 1 & 2)
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |