Speech Enhancement on VoiceBank + DEMAND

42개 결과 · ⬇ CSV · JSON

PESQ (wb)

Rank	Model	PESQ (wb)	CBAK	COVL	CSIG	STOI	ESTOI	SSNR	SI-SDR	Extra Training Data	Paper	Code	Year
21	MANNER	3.21	3.65	3.91	4.53	95	–	–	–		MANNER: Multi-view Attention Network for Noise Erasure	winddori2002/MANNER	2022
22	BSSE-SE	3.20	3.58	3.88	4.52	95.7	–	–	–		Boosting Self-Supervised Embeddings for Speech Enhancement	khhungg/BSSE-SE	2022
23	DeepFilterNet3	3.17	3.61	3.77	4.34	0.944	–	–	–		DeepFilterNet: Perceptually Motivated Real-Time Speech Enhancement	rikorose/deepfilternet	2023
23	PERL-AE	3.17	3.53	3.83	4.43	–	–	–	–	✓	Perceptual Loss based Speech Denoising with an ensemble of Audio Pattern Recognition and Self-Supervised Models	saurabh-kataria/PERL-samples	2020
25	PFPL	3.15	3.60	3.67	4.18	–	–	–	–		Improving Perceptual Quality by Phone-Fortified Perceptual Loss using Wasserstein Distance for Speech Enhancement	aleXiehta/PhoneFortifiedPerceptualLoss	2020
25	MetricGAN+	3.15	3.16	3.64	4.14	–	–	–	–		MetricGAN+: An Improved Version of MetricGAN for Speech Enhancement	speechbrain/speechbrain · JasonSWFu/MetricGAN · wooseok-shin/MetricGAN-plus-pytorch	2021
27	MANNER-S + MV-AT (8.1GF)	3.12	3.61	3.82	4.45	95	–	–	–		Multi-View Attention Transfer for Efficient Speech Enhancement		2022
27	MetricGAN-OKD (Causal Arch.)	3.12	3.13	3.64	4.17	–	–	–	–		MetricGAN-OKD: Multi-Metric Optimization of MetricGAN via Online Knowledge Distillation for Speech Enhancement	wooseok-shin/MetricGAN-OKD	2023
29	SGMSE+	3.11	–	–	–	–	–	–	–		An Analysis of the Variance of Diffusion-based Speech Enhancement		2024
30	DEMUCS (H=64, S=2 ,U =2)	3.07	3.4	3.63	4.31	95	–	–	–		Real Time Speech Enhancement in the Waveform Domain	facebookresearch/denoiser · yunyangzeng/taploss · jonashaag/paperswithcode-speech-enhancement-audiosamples	2020
31	Dense-TSNet	3.05	3.58	3.86	4.51	–	–	–	–		Dense-TSNet: Dense Connected Two-Stage Structure for Ultra-Lightweight Speech Enhancement		2024
32	RDL-Net 3.91M (Deep Xi - MMSE-LSA)	3.02	3.43	3.72	4.38	–	–	–	–		Deep Residual-Dense Lattice Network for Speech Enhancement	nick-nikzad/RDL-SE · MindSpore-paper-code-2/code400	2020
33	ROSE	3.01	3.56	3.72	4.47	95	–	–	–		ROSE: A Recognition-Oriented Speech Enhancement Framework in Air Traffic Control Using Multi-Objective Learning	xcyu-0903/rose	2023
34	FSPEN	2.97	–	–	–	0.942	–	–	–		FSPEN: AN ULTRA-LIGHTWEIGHT NETWORK FOR REAL TIME SPEECH ENAHNCMENT	gitwukeyi/FSPEN	2024
35	RDL-Net 3.91M (Deep Xi - SRWF)	2.94	3.35	3.67	4.36	–	–	–	–		Deep Residual-Dense Lattice Network for Speech Enhancement	nick-nikzad/RDL-SE · MindSpore-paper-code-2/code400	2020
36	RDL-Net 1.87M (Deep Xi - MMSE-LSA)	2.93	3.32	3.62	4.29	–	–	–	–		Deep Residual-Dense Lattice Network for Speech Enhancement	nick-nikzad/RDL-SE · MindSpore-paper-code-2/code400	2020
36	Causal DEMUCS (H=48,S=4, U =4)	2.93	3.25	3.52	4.22	95	–	–	–		Real Time Speech Enhancement in the Waveform Domain	facebookresearch/denoiser · yunyangzeng/taploss · jonashaag/paperswithcode-speech-enhancement-audiosamples	2020
36	SGMSE+ (Diffusion Model)	2.93	–	–	–	–	–	–	–		Speech Enhancement and Dereverberation with Diffusion-based Generative Models	sp-uhh/sgmse	2022
39	MetricGAN	2.86	3.18	3.42	3.99	–	–	–	–		MetricGAN: Generative Adversarial Networks based Black-box Metric Scores Optimization for Speech Enhancement	anicolson/DeepXi · JasonSWFu/MetricGAN · nick-nikzad/RDL-SE · +2	2019
40	RDL-Net 1.87M (Deep Xi - SRWF)	2.84	3.23	3.56	4.27	–	–	–	–		Deep Residual-Dense Lattice Network for Speech Enhancement	nick-nikzad/RDL-SE · MindSpore-paper-code-2/code400	2020