Table 3 Performance of Conv-TasNet and DPCCN trained with ground-truth labels on WHAMR! and Aishell2Mix test sets
From: Heterogeneous separation consistency training for adaptation of unsupervised speech separation
System | Dataset | SDRi (dB) | SI-SNRi (dB) |
|---|---|---|---|
Conv-TasNet | WHAMR! | 11.03 | 10.59 |
Aishell2Mix | 9.00 | 8.32 | |
DPCCN | WHAMR! | 11.01 | 10.56 |
Aishell2Mix | 8.86 | 8.14 |