MossFormer2

Reported on 4 benchmarks across 1 task · 1 paper · 1 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Speech4 results

Speech SeparationonWSJ0-2mix-16k
SI-SDRi· uses extra data· 2023-02-23
20.5
SOTA
MossFormer: Pushing the Performance Limit of Monaural Speech Separation using Gated Single-Head Transformer with Convolution-Augmented Joint Self-Attentions arXiv:2302.11824
Speech SeparationonWHAMR!
SI-SDRi
17
best: 18.5 (TF-Locoformer (M))
Speech SeparationonWSJ0-3mix
SI-SDRi
22.2
best: 23.7 (SepTDA)
Speech SeparationonWHAM!
SI-SDRi
18.1
best: 18.4 (SepReformer-L + DM)