HyperAI超神経
ホームニュース最新論文チュートリアルデータセット百科事典SOTALLMモデルGPU ランキング学会
検索
サイトについて
HyperAI超神経

Speech Separation On Libri2Mix

評価指標

SI-SDRi

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

モデル名
SI-SDRi
Paper TitleRepository
TDANet16.9An efficient encoder-decoder architecture with top-down attention for speech separation
MossFormer2 (w/o DM)21.7MossFormer2: Combining Transformer and RNN-Free Recurrent Network for Enhanced Time-Domain Monaural Speech Separation
MossFormer2 (w speed perturb)22.2MossFormer2: Combining Transformer and RNN-Free Recurrent Network for Enhanced Time-Domain Monaural Speech Separation
Conv-Tasnet (Libri1Mix speech enhancement pre-trained)14.1Stabilizing Label Assignment for Speech Separation by Self-supervised Pre-training
TF-Locoformer (M)22.1TF-Locoformer: Transformer with Local Modeling by Convolution for Speech Separation and Enhancement
WHYV17.5An alternative Approach in Voice Extraction-
TDANet Large17.4An efficient encoder-decoder architecture with top-down attention for speech separation
Conv-Tasnet (Libri1Mix speech enhancement multi-task)13.7Stabilizing Label Assignment for Speech Separation by Self-supervised Pre-training
Conv-Tasnet13.2Stabilizing Label Assignment for Speech Separation by Self-supervised Pre-training
Separate And Diffuse21.5Separate And Diffuse: Using a Pretrained Diffusion Model for Improving Source Separation-
0 of 10 row(s) selected.