Awesome Speech Enhancement

This repository summarizes the papers, codes, and tools for single-/multi-channel speech enhancement/speech separation. Welcome to pull requests.

Contents

• Speech_Enhancement
• Dereverberation
• Speech_Seperation
• Array_Signal_Processing
• Tools
• Books
• Resources

Speech_Enhancement

Magnitude spectrogram

spectral masking

• 2014, On Training Targets for Supervised Speech Separation, Wang. [Paper]
• 2018, A Hybrid DSP/Deep Learning Approach to Real-Time Full-Band Speech Enhancement, Valin. [Paper] [RNNoise] [RNNoise16k]
• 2020, A Perceptually-Motivated Approach for Low-Complexity, Real-Time Enhancement of Fullband Speech, Valin. Paper [PercepNet]
• 2020, Online Monaural Speech Enhancement using Delayed Subband LSTM, Li. [Paper]
• 2020, FullSubNet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement, Hao. [Paper] [FullSubNet]
• 2020， Weighted Speech Distortion Losses for Neural-network-based Real-time Speech Enhancement, Xia. [Paper] [NSNet]
• 2020, RNNoise-like fixed-point model deployed on Microcontroller using NNoM inference framework [example] [NNoM]
• 2021, RNNoise-Ex: Hybrid Speech Enhancement System based on RNN and Spectral Features. [Paper] [RNNoise-Ex]
• Other IRM-based SE repositories: [IRM-SE-LSTM] [nn-irm] [rnn-se] [DL4SE]

spectral mapping

• 2014, An Experimental Study on Speech Enhancement Based on Deep Neural Networks, Xu. [Paper]

• 2014, A Regression Approach to Speech Enhancement Based on Deep Neural Networks, Xu. [Paper] [sednn] [DNN-SE-Xu] [DNN-SE-Li]

• Other DNN magnitude spectrum mapping-based SE repositories: [SE toolkit] [TensorFlow-SE] [UNetSE]

• 2015, Speech enhancement with LSTM recurrent neural networks and its application to noise-robust ASR, Weninger. [Paper]

• 2016, A Fully Convolutional Neural Network for Speech Enhancement, Park. [Paper] [CNN4SE]

• 2017, Long short-term memory for speaker generalizationin supervised speech separation, Chen. [Paper]

• 2018, A Convolutional Recurrent Neural Network for Real-Time Speech Enhancement, Tan. [Paper] [CRN-Tan]

• 2018, Convolutional-Recurrent Neural Networks for Speech Enhancement, Zhao. [Paper] [CRN-Hao]

Complex domain

• 2017, Complex spectrogram enhancement by convolutional neural network with multi-metrics learning, Fu. [Paper]

• 2017, Time-Frequency Masking in the Complex Domain for Speech Dereverberation and Denoising, Williamson. [Paper]

• 2019, PHASEN: A Phase-and-Harmonics-Aware Speech Enhancement Network, Yin. [Paper] [PHASEN]

• 2019, Phase-aware Speech Enhancement with Deep Complex U-Net, Choi. [Paper] [DC-UNet]

• 2020, Learning Complex Spectral Mapping With GatedConvolutional Recurrent Networks forMonaural Speech Enhancement, Tan. [Paper] [GCRN]

• 2020, DCCRN: Deep Complex Convolution Recurrent Network for Phase-AwareSpeech Enhancement, Hu. [Paper] [DCCRN]

• 2020, T-GSA: Transformer with Gaussian-Weighted Self-Attention for Speech Enhancement, Kim. [Paper]

• 2020, Phase-aware Single-stage Speech Denoising and Dereverberation with U-Net, Choi. [Paper]

• 2021, DPCRN: Dual-Path Convolution Recurrent Network for Single Channel Speech Enhancement, Le. [Paper] [DPCRN]

• 2021, Real-time denoising and dereverberation with tiny recurrent u-net, Choi. [Paper]

• 2021, DCCRN+: Channel-wise Subband DCCRN with SNR Estimation for Speech Enhancement, Lv [Paper]

• 2022, FullSubNet+: Channel Attention FullSubNet with Complex Spectrograms for Speech Enhancement, Chen [Paper] [FullSubNet+]

• 2022, Dual-branch Attention-In-Attention Transformer for single-channel speech enhancement, Yu [Paper]

Time domain

• 2018, Improved Speech Enhancement with the Wave-U-Net, Macartney. [Paper] [WaveUNet]
• 2019, A New Framework for CNN-Based Speech Enhancement in the Time Domain, Pandey. [Paper]
• 2019, TCNN: Temporal Convolutional Neural Network for Real-time Speech Enhancement in the Time Domain, Pandey. [Paper]
• 2020, Real Time Speech Enhancement in the Waveform Domain, Defossez. [Paper] [facebookDenoiser]
• 2020, Monaural speech enhancement through deep wave-U-net, Guimarães. [Paper] [SEWUNet]
• 2020, Speech Enhancement Using Dilated Wave-U-Net: an Experimental Analysis, Ali. [Paper]
• 2020, Densely Connected Neural Network with Dilated Convolutions for Real-Time Speech Enhancement in the Time Domain, Pandey. [Paper] [DDAEC]
• 2021, Dense CNN With Self-Attention for Time-Domain Speech Enhancement, Pandey. [Paper]
• 2021, Dual-path Self-Attention RNN for Real-Time Speech Enhancement, Pandey. [Paper]
• 2022, Speech Denoising in the Waveform Domain with Self-Attention, Kong. [Paper]

Generative Model

GAN

• 2017, SEGAN: Speech Enhancement Generative Adversarial Network, Pascual. [Paper] [SEGAN]
• 2019, SERGAN: Speech enhancement using relativistic generative adversarial networks with gradient penalty, Deepak Baby. [Paper] [SERGAN]
• 2019, MetricGAN: Generative Adversarial Networks based Black-box Metric Scores Optimization for Speech Enhancement, Fu. [Paper] [MetricGAN]
• 2019, MetricGAN+: An Improved Version of MetricGAN for Speech Enhancement, Fu. [Paper] [MetricGAN+]
• 2020, HiFi-GAN: High-Fidelity Denoising and Dereverberation Based on Speech Deep Features in Adversarial Networks, Su. [Paper] [HifiGAN]
• 2022, CMGAN: Conformer-Based Metric GAN for Monaural Speech Enhancement, Abdulatif, Cao & Yang. [Paper] [CMGAN]

Flow

• 2021, A Flow-based Neural Network for Time Domain Speech Enhanceent, Strauss & Edler. [Paper]

VAE

• 2018, A variance modeling framework based on variational autoencoders for speech enhancement, [Leglaive]. [Paper] [mlsp]
• 2020, Speech Enhancement with Stochastic Temporal Convolutional Networks, Richter. [Paper] [STCN-NMF]

Diffusion Model

• 2022, Speech Enhancement with Score-Based Generative Models in the Complex STFT Domain, Welker. [Paper] [SGMSE]
• 2022, StoRM: A Stochastic Regeneration Model for Speech Enhancement And Dereverberation, Lemercier. [Paper] [StoRM]
• 2022, Conditional Diffusion Probabilistic Model for Speech Enhancement, Lu. [Paper] [CDiffuSE]
• 2023, Speech Enhancement and Dereverberation with Diffusion-Based Generative Models, Richter. [Paper] [SGMSE]

Hybrid SE

• 2019, Deep Xi as a Front-End for Robust Automatic Speech Recognition, Nicolson. [Paper] [DeepXi]
• 2019, Using Generalized Gaussian Distributions to Improve Regression Error Modeling for Deep-Learning-Based Speech Enhancement, Li. [Paper] [SE-MLC]
• 2020, Deep Residual-Dense Lattice Network for Speech Enhancement, Nikzad. [Paper] [RDL-SE]
• 2020, DeepMMSE: A Deep Learning Approach to MMSE-based Noise Power Spectral Density Estimation, Zhang. [Paper]
• 2020, Speech Enhancement Using a DNN-Augmented Colored-Noise Kalman Filter, Yu. [Paper] [DNN-Kalman]

Decoupling-style

• 2020, A Recursive Network with Dynamic Attention for Monaural Speech Enhancement, Li. [Paper] [DARCN]
• 2020, Masking and Inpainting: A Two-Stage Speech Enhancement Approach for Low SNR and Non-Stationary Noise, Hao. [Paper]
• 2020, A Joint Framework of Denoising Autoencoder and Generative Vocoder for Monaural Speech Enhancement, Du. [Paper]
• 2020, Dual-Signal Transformation LSTM Network for Real-Time Noise Suppression, Westhausen. [Paper] [DTLN]
• 2020, Listening to Sounds of Silence for Speech Denoising, Xu. [Paper] [LSS]
• 2021, ICASSP 2021 Deep Noise Suppression Challenge: Decoupling Magnitude and Phase Optimization with a Two-Stage Deep Network, Li. [Paper]
• 2022, Glance and Gaze: A Collaborative Learning Framework for Single-channel Speech Enhancement, Li [Paper]
• 2022, HGCN : harmonic gated compensation network for speech enhancement, Wang. [Paper]
• 2022, Uformer: A Unet based dilated complex & real dual-path conformer network for simultaneous speech enhancement and dereverberation, Fu. [Paper] [Uformer]
• 2022， DeepFilterNet2: Towards Real-Time Speech Enhancement on Embedded Devices for Full-Band Audio, Schröter. [Paper] [DeepFilterNet]
• 2021, Multi-Task Audio Source Separation, Zhang. [Paper] [Code]

Data collection

• Kashyap([Noise2Noise])

Loss

• [Quality-Net]

Challenge

• DNS Challenge [DNS Interspeech2020] [DNS ICASSP2021] [DNS Interspeech2021]

Other repositories

• Collection of papers, datasets and tools on the topic of Speech Dereverberation and Speech Enhancement [Link]
• nanahou's awesome speech enhancement [Link]

Dereverberation

Traditional method

• SPENDRED [Paper] [SPENDRED]
• WPE(MCLP) [Paper][nara-WPE]
• GWPE [Code]
• LP Residual [Paper] [LP_residual]
• dereverberate [Paper] [Code]
• NMF [Paper] [NMF]

Hybrid method

• DNN_WPE [Paper] [Code]

NN-based Derev

• Dereverberation-toolkit-for-REVERB-challenge [Code]
• SkipConvNet [Paper] [Code]

Speech Separation (single channel)

• Tutorial speech separation, like awesome series [Link]

NN-based separation

• 2015, Deep-Clustering:Discriminative embeddings for segmentation and separation, Hershey and Chen.[Paper] [Code] [Code] [Code]
• 2016, DANet:Deep Attractor Network (DANet) for single-channel speech separation, Chen.[Paper] [Code]
• 2017, Multitalker speech separation with utterance-level permutation invariant training of deep recurrent, Yu.[Paper] [Code]
• 2018, LSTM_PIT_Speech_Separation [Code]
• 2018, Tasnet: time-domain audio separation network for real-time, single-channel speech separation, Luo.[Paper] [Code]
• 2019, Conv-TasNet: Surpassing Ideal Time-Frequency Masking for Speech Separation, Luo.(Paper) [Code]
• 2019, Dual-path RNN: efficient long sequence modeling for time-domain single-channel speech separation, Luo.[Paper] [Code1] [Code2]
• 2019, TAC end-to-end microphone permutation and number invariant multi-channel speech separation, Luo.[Paper] [Code]
• 2020, Continuous Speech Separation with Conformer, Chen.[Paper] [Code]
• 2020, Dual-Path Transformer Network: Direct Context-Aware Modeling for End-to-End Monaural Speech Separation, Chen.[Paper] [Code]
• 2020, Wavesplit: End-to-End Speech Separation by Speaker Clustering, Zeghidour.[Paper]
• 2021, Attention is All You Need in Speech Separation, Subakan.[Paper] [Code]
• 2021, Ultra Fast Speech Separation Model with Teacher Student Learning, Chen.[Paper]
• sound separation(Google) [Code]
• sound separation: Deep learning based speech source separation using Pytorch [Code]
• music-source-separation [Code]
• Singing-Voice-Separation [Code]
• Comparison-of-Blind-Source-Separation-techniques[Code]

BSS/ICA method

• FastICA[Code]
• A localisation- and precedence-based binaural separation algorithm[Download]
• Convolutive Transfer Function Invariant SDR [Code]

Array Signal Processing

• MASP:Microphone Array Speech Processing [Code]
• BeamformingSpeechEnhancer [Code]
• TSENet [Code]
• steernet [Code]
• DNN_Localization_And_Separation [Code]
• nn-gev:Neural network supported GEV beamformer CHiME3 [Code]
• chime4-nn-mask:Implementation of NN based mask estimator in pytorch（reuse some programming from nn-gev）[Code]
• beamformit_matlab:A MATLAB implementation of CHiME4 baseline Beamformit [Code]
• pb_chime5:Speech enhancement system for the CHiME-5 dinner party scenario [Code]
• beamformit:麦克风阵列算法 [Code]
• Beamforming-for-speech-enhancement [Code]
• deepBeam [Code]
• NN_MASK [Code]
• Cone-of-Silence [Code]

Tools

• APS:A workspace for single/multi-channel speech recognition & enhancement & separation. [Code]
• AKtools:the open software toolbox for signal acquisition, processing, and inspection in acoustics [SVN Code](username: aktools; password: ak)
• espnet [Code]
• asteroid:The PyTorch-based audio source separation toolkit for researchers[PDF][Code]
• pytorch_complex [Code]
• ONSSEN: An Open-source Speech Separation and Enhancement Library [Code]
• separation_data_preparation[Code]
• MatlabToolbox [Code]
• athena-signal [[Code]](https://github.com/athena-team/athena-signal）
• python_speech_features [Code]
• speechFeatures [Code]
• sap-voicebox [Code]
• Calculate-SNR-SDR [Code]
• RIR-Generator [Code]
• Signal-Generator (for moving sources or a moving array) [Code]
• Python library for Room Impulse Response (RIR) simulation with GPU acceleration [Code]
• ROOMSIM:binaural image source simulation [Code]
• binaural-image-source-model [Code]
• PESQ [Code]
• SETK: Speech Enhancement Tools integrated with Kaldi [Code]
• pb_chime5:Speech enhancement system for the CHiME-5 dinner party scenario [Code]

Books

• P. C.Loizou: Speech Enhancement: Theory and Practice
• J. Benesty, Y. Huang: Adaptive Signal Processing: Applications to Real-World Problems
• S. Haykin: Adaptive Filter Theory
• Eberhard Hansler, Gerhard Schmidt: Single-Channel Acoustic Echo Cancellation 和 Topics in Acoustic Echo and Noise Control
• J. Benesty, S. Makino, J. Chen: Speech Enhancement
• J. Benesty, M. M. Sondhi, Y. Huang: Handbook Of Speech Processing
• Ivan J. Tashev: Sound Capture and Processing: Practical Approaches
• I. Cohen, J. Benesty, S. Gannot: Speech Processing in Modern Communication
• E. Vincent, T. Virtanen, S. Gannot: Audio Source Separation and Speech Enhancement
• J. Benesty 等: A Perspective on Stereophonic Acoustic Echo Cancellation
• J. Benesty 等: Advances in Network and Acoustic Echo Cancellation
• T. F.Quatieri: Discrete-time speech signal processing: principles and practice
• 宋知用: MATLAB在语音信号分析与合成中的应用
• Harry L.Van Trees: Optimum Array Processing
• 王永良: 空间谱估计理论与算法
• 鄢社锋: 优化阵列信号处理
• 张小飞: 阵列信号处理及matlab实现
• 赵拥军: 宽带阵列信号波达方向估计理论与方法
• The-guidebook-of-speech-enhancement

Resources

• Sixty-years-of-frequency-domain-monaural-speech-enhancement [se_overview]
• Speech Signal Processing Course(ZH) [Link]
• Speech Algorithms(ZH) [Link]
• Speech Resources[Link]
• Sound capture and speech enhancement for speech-enabled devices [Link]
• CCF语音对话与听觉专业组语音对话与听觉前沿研讨会(ZH) [Link]

---

• binauralLocalization [Code]
• robotaudition_examples:Some Robot Audition simplified examples (sound source localization and separation), coded in Octave/Matlab [Code]
• WSCM-MUSIC [Code]
• doa-tools [Code]
• Regression and Classification for Direction-of-Arrival Estimation with Convolutional Recurrent Neural Networks [Code] [PDF]
• messl:Model-based EM Source Separation and Localization [Code]
• messlJsalt15:MESSL wrappers etc for JSALT 2015, including CHiME3 [Code]
• fast_sound_source_localization_using_TLSSC:Fast Sound Source Localization Using Two-Level Search Space Clustering [Code]
• Binaural-Auditory-Localization-System [Code]
• Binaural_Localization:ITD-based localization of sound sources in complex acoustic environments [Code]
• Dual_Channel_Beamformer_and_Postfilter [Code]
• 麦克风声源定位 [Code]
• RTF-based-LCMV-GSC [Code]
• DOA [Code]

Sound Event Detection

• sed_eval - Evaluation toolbox for Sound Event Detection [Code]
• Benchmark for sound event localization task of DCASE 2019 challenge [Code]
• sed-crnn DCASE 2017 real-life sound event detection winning method. [Code]
• seld-net [Code]