Speech Enhancement On Deep Noise Suppression

評価指標

PESQ-WB

SI-SDR-WB

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

比較表

モデル名	PESQ-WB	SI-SDR-WB
high-fidelity-speech-enhancement-with-band	3.42	21.3
モデル 2	1.58	9.1
remixit-continual-self-training-of-speech	2.34	16.0
interactive-speech-and-noise-modeling-for-1	-	-
remixit-continual-self-training-of-speech	2.95	19.7
モデル 6	3.69	21.15
raw-speech-enhancement-with-deep-state-space	2.98	-
high-fidelity-speech-enhancement-with-band	-	21.4
real-time-monaural-speech-enhancement-with	2.77	-
a-mask-free-neural-network-for-monaural	3.43	20.31
high-fidelity-speech-enhancement-with-band	3.45	21.1
モデル 12	3.81	22.22
real-time-monaural-speech-enhancement-with	2.82	-
tf-locoformer-transformer-with-local-modeling	3.72	23.3
dccrn-deep-complex-convolution-recurrent-1	-	-
monaural-speech-enhancement-with-complex	3.23	-
dual-signal-transformation-lstm-network-for	-	16.34
fullsubnet-channel-attention-fullsubnet-with	3.218	16.81
real-time-monaural-speech-enhancement-with	2.82	-
continual-self-training-with-bootstrapped	2.69	18.6
fullsubnet-a-full-band-and-sub-band-fusion	2.777	17.29
real-time-monaural-speech-enhancement-with	2.79	-
high-fidelity-speech-enhancement-with-band	3.32	-
a-modulation-domain-loss-for-neural-network-1	2.75	-
cleanunet-2-a-hybrid-speech-denoising-model	3.262	-
high-fidelity-speech-enhancement-with-band	3.53	21.4
continual-self-training-with-bootstrapped	2.60	18.0
speech-denoising-in-the-waveform-domain-with	3.146	-
dccrn-deep-complex-convolution-recurrent-1	-	-
weighted-speech-distortion-losses-for-neural-1	2.65	-
phase-aware-single-stage-speech-denoising-and-1	-	16.22
exploring-the-best-loss-function-for-dnn-1	2.73	-
poconet-better-speech-enhancement-with	2.7885	-
explicit-estimation-of-magnitude-and-phase	3.62	21.03