Speech Separation On Wsj0 2Mix

Metrics

Number of parameters (M)

SDRi

SI-SDRi

Results

Performance results of various models on this benchmark

Model Name	Number of parameters (M)	SDRi	SI-SDRi	Paper Title	Repository
TF-Locoformer (S) + DM	5.0	23	22.8	TF-Locoformer: Transformer with Local Modeling by Convolution for Speech Separation and Enhancement
TD-Conformer (XL) + DM	-	-	21.2	On Time Domain Conformer Models for Monaural Speech Separation in Noisy Reverberant Acoustic Environments
Conv-TasNet	5.1	-	15.3	Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation
Two-step Conv-TasNet	-	-	16.1	Two-Step Sound Source Separation: Training on Learned Latent Targets
TF-Locoformer (M)	15.0	23.8	23.6	TF-Locoformer: Transformer with Local Modeling by Convolution for Speech Separation and Enhancement
DPTNet	-	-	20.2	Dual-Path Transformer Network: Direct Context-Aware Modeling for End-to-End Monaural Speech Separation
SepReformer-L	59.4	25.2	25.1	Separate and Reconstruct: Asymmetric Encoder-Decoder for Speech Separation
Wavesplit v2	-	22.3	22.2	Wavesplit: End-to-End Speech Separation by Speaker Clustering	-
Deformable TCN + Dynamic Mixing	3.6	17.4	17.2	Deformable Temporal Convolutional Networks for Monaural Noisy Reverberant Speech Separation
Separate And Diffuse	-	-	23.9	Separate And Diffuse: Using a Pretrained Diffusion Model for Improving Source Separation	-
TasNet	-	-	10.8	TasNet: time-domain audio separation network for real-time, single-channel speech separation
Sandglasset	-	-	21.0	Sandglasset: A Light Multi-Granularity Self-attentive Network For Time-Domain Speech Separation
SPGM + DM	26.2	-	22.7	SPGM: Prioritizing Local Features for enhanced speech separation performance
Deep Clustering ++	-	-	10.8	Deep clustering: Discriminative embeddings for segmentation and separation
SPGM	26.2	-	22.1	SPGM: Prioritizing Local Features for enhanced speech separation performance
Sudo rm -rf (U=36)	-	-	19.5	Compute and memory efficient universal sound source separation
SepTDA (L=12)	-	-	24.0	Boosting Unknown-number Speaker Separation with Transformer Decoder-based Attractor	-
SepIt	-	-	22.4	SepIt: Approaching a Single Channel Speech Separation Bound	-
MossFormer (L) + DM	42.1	-	22.8	MossFormer: Pushing the Performance Limit of Monaural Speech Separation using Gated Single-Head Transformer with Convolution-Augmented Joint Self-Attentions
Gated DualPathRNN	-	-	20.12	Voice Separation with an Unknown Number of Multiple Speakers

0 of 38 row(s) selected.