Single-channel Speech Separation on WSJ0-3mix v1 (test)
14.2P-SI-SNR (Pref=-30dB)Attractor Network
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Attractor Network2020.11 | 14.2 | 14.3 | |
| Multi-Decoder DPRNNPer-speaker training=false2020.11 | 14 | 14 | |
| Model-Select(Mulcat)Per-speaker training=true, Backbone=Mulcat2020.11 | 13.2 | 14.2 | |
| OR-PITRecall=95.7%, Aggregation=Average of 2 and 3 speakers2020.11 | 13.1 | 13.4 | |
| Model-Select(DPRNN)Per-speaker training=true, Backbone=DPRNN2020.11 | 10.7 | 12.1 |