Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation

Abstract

Single-channel, speaker-independent speech separation methods have recently seen great progress. However, the accuracy, latency, and computational cost of such methods remain insufficient. The majority of the previous methods have formulated the separation problem through the time-frequency representation of the mixed signal, which has several drawbacks, including the decoupling of the phase and magnitude of the signal, the suboptimality of time-frequency representation for speech separation, and the long latency in calculating the spectrograms. To address these shortcomings, we propose a fully-convolutional time-domain audio separation network (Conv-TasNet), a deep learning framework for end-to-end time-domain speech separation. Conv-TasNet uses a linear encoder to generate a representation of the speech waveform optimized for separating individual speakers. Speaker separation is achieved by applying a set of weighting functions (masks) to the encoder output. The modified encoder representations are then inverted back to the waveforms using a linear decoder. The masks are found using a temporal convolutional network (TCN) consisting of stacked 1-D dilated convolutional blocks, which allows the network to model the long-term dependencies of the speech signal while maintaining a small model size. The proposed Conv-TasNet system significantly outperforms previous time-frequency masking methods in separating two- and three-speaker mixtures. Additionally, Conv-TasNet surpasses several ideal time-frequency magnitude masks in two-speaker speech separation as evaluated by both objective distortion measures and subjective quality assessment by human listeners. Finally, Conv-TasNet has a significantly smaller model size and a shorter minimum latency, making it a suitable solution for both offline and real-time speech separation applications.

arXiv commentAccepted by IEEE/ACM Transactions on Audio, Speech and Language Processing. This version is the authors' version and may vary from the final publication in details

References in corpus (9)

• Adam: A Method for Stochastic Optimization
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling
• WaveNet: A Generative Model for Raw Audio
• Deep attractor network for single-microphone speaker separation
• Speaker-independent Speech Separation with Deep Attractor Network
• Depthwise Separable Convolutions for Neural Machine Translation
• Deep Clustering and Conventional Networks for Music Separation: Stronger Together
• End-to-End Speech Separation with Unfolded Iterative Phase Reconstruction

Cited by in corpus (299)

• SpeechBrain: A General-Purpose Speech Toolkit
• A Survey of Sound Source Localization with Deep Learning Methods
• Speech Enhancement and Dereverberation with Diffusion-based Generative Models
• Music Source Separation in the Waveform Domain
• LibriMix: An Open-Source Dataset for Generalizable Speech Separation
• Towards End-to-End Synthetic Speech Detection
• Dynamical Variational Autoencoders: A Comprehensive Review
• SpEx: Multi-Scale Time Domain Speaker Extraction Network
• On Loss Functions for Supervised Monaural Time-Domain Speech Enhancement
• Neural Target Speech Extraction: An Overview
• Voice Separation with an Unknown Number of Multiple Speakers
• Sudo rm -rf: Efficient Networks for Universal Audio Source Separation
• Multi-modal Multi-channel Target Speech Separation
• CMGAN: Conformer-Based Metric-GAN for Monaural Speech Enhancement
• On The Compensation Between Magnitude and Phase in Speech Separation
• Unsupervised Sound Separation Using Mixture Invariant Training
• Phasebook and Friends: Leveraging Discrete Representations for Source Separation
• Physics-Inspired Temporal Learning of Quadrotor Dynamics for Accurate Model Predictive Trajectory Tracking
• Learning with Learned Loss Function: Speech Enhancement with Quality-Net to Improve Perceptual Evaluation of Speech Quality
• End-to-End Multi-Channel Speech Separation
• Improving Universal Sound Separation Using Sound Classification
• FaceFilter: Audio-visual speech separation using still images
• Hybrid Spectrogram and Waveform Source Separation
• Two-Step Sound Source Separation: Training on Learned Latent Targets
• The INTERSPEECH 2020 Deep Noise Suppression Challenge: Datasets, Subjective Speech Quality and Testing Framework
• Audio-Visual Speech Separation and Dereverberation with a Two-Stage Multimodal Network
• A Multi-Phase Gammatone Filterbank for Speech Separation via TasNet
• DCCRN: Deep Complex Convolution Recurrent Network for Phase-Aware Speech Enhancement
• Encoder-Decoder Based Attractors for End-to-End Neural Diarization
• Asteroid: the PyTorch-based audio source separation toolkit for researchers
• Dual-path RNN: efficient long sequence modeling for time-domain single-channel speech separation
• On End-to-end Multi-channel Time Domain Speech Separation in Reverberant Environments
• D3Net: Densely connected multidilated DenseNet for music source separation
• Compute and memory efficient universal sound source separation
• Jointly optimal denoising, dereverberation, and source separation
• L3DAS22 Challenge: Learning 3D Audio Sources in a Real Office Environment
• DeFT-AN: Dense Frequency-Time Attentive Network for Multichannel Speech Enhancement
• Multi-talker ASR for an unknown number of sources: Joint training of source counting, separation and ASR
• Exploring the Best Loss Function for DNN-Based Low-latency Speech Enhancement with Temporal Convolutional Networks
• End-to-end training of time domain audio separation and recognition
• ClearBuds: Wireless Binaural Earbuds for Learning-Based Speech Enhancement
• Noise Estimation for Generative Diffusion Models
• F-T-LSTM based Complex Network for Joint Acoustic Echo Cancellation and Speech Enhancement
• Time-domain speaker extraction network
• Dual-Path Transformer Network: Direct Context-Aware Modeling for End-to-End Monaural Speech Separation
• Space-based gravitational wave signal detection and extraction with deep neural network
• LEAF: A Learnable Frontend for Audio Classification
• FaSNet: Low-latency Adaptive Beamforming for Multi-microphone Audio Processing
• Semantic Hearing: Programming Acoustic Scenes with Binaural Hearables
• Multi-Domain Processing via Hybrid Denoising Networks for Speech Enhancement
• Domain Adaptation and Autoencoder Based Unsupervised Speech Enhancement
• Speaker Extraction with Co-Speech Gestures Cue
• Efficient Personalized Speech Enhancement through Self-Supervised Learning
• Learning to Enhance or Not: Neural Network-Based Switching of Enhanced and Observed Signals for Overlapping Speech Recognition
• Phase-Aware Deep Speech Enhancement: It's All About The Frame Length
• Group Communication with Context Codec for Lightweight Source Separation
• Blind and neural network-guided convolutional beamformer for joint denoising, dereverberation, and source separation
• Scalable and Efficient Neural Speech Coding: A Hybrid Design
• The Cone of Silence: Speech Separation by Localization
• PHASEN: A Phase-and-Harmonics-Aware Speech Enhancement Network
• Look Once to Hear: Target Speech Hearing with Noisy Examples
• VoiceFixer: Toward General Speech Restoration with Neural Vocoder
• NORESQA: A Framework for Speech Quality Assessment using Non-Matching References
• Time-Domain Audio Source Separation Based on Wave-U-Net Combined with Discrete Wavelet Transform
• Deep Multi-Frame MVDR Filtering for Single-Microphone Speech Enhancement
• Finding Strength in Weakness: Learning to Separate Sounds with Weak Supervision
• Listening to Sounds of Silence for Speech Denoising
• SAGRNN: Self-Attentive Gated RNN for Binaural Speaker Separation with Interaural Cue Preservation
• USTC-NELSLIP System Description for DIHARD-III Challenge
• Graph-PIT: Generalized permutation invariant training for continuous separation of arbitrary numbers of speakers
• Efficient Transformer-based Speech Enhancement Using Long Frames and STFT Magnitudes
• A Squeeze-and-Excitation and Transformer based Cross-task System for Environmental Sound Recognition
• Real-time Monaural Speech Enhancement With Short-time Discrete Cosine Transform
• Investigating the Design Space of Diffusion Models for Speech Enhancement
• SpEx+: A Complete Time Domain Speaker Extraction Network
• Conformer-based Target-Speaker Automatic Speech Recognition for Single-Channel Audio
• Guided Variational Autoencoder for Speech Enhancement With a Supervised Classifier
• Functional Split of In-Network Deep Learning for 6G: A Feasibility Study
• Assessing the Generalization Gap of Learning-Based Speech Enhancement Systems in Noisy and Reverberant Environments
• FurcaNet: An end-to-end deep gated convolutional, long short-term memory, deep neural networks for single channel speech separation
• PixIT: Joint Training of Speaker Diarization and Speech Separation from Real-world Multi-speaker Recordings
• A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models
• Online Spectrogram Inversion for Low-Latency Audio Source Separation
• Glance and Gaze: A Collaborative Learning Framework for Single-channel Speech Enhancement
• Target Speech Extraction: Independent Vector Extraction Guided by Supervised Speaker Identification
• Speaker Reinforcement Using Target Source Extraction for Robust Automatic Speech Recognition
• End-to-End Complex-Valued Multidilated Convolutional Neural Network for Joint Acoustic Echo Cancellation and Noise Suppression
• DeFTAN-II: Efficient Multichannel Speech Enhancement with Subgroup Processing
• FullSubNet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement
• Improving speaker discrimination of target speech extraction with time-domain SpeakerBeam
• Speech Separation with Pretrained Frontend to Minimize Domain Mismatch
• Multitask-Based Joint Learning Approach To Robust ASR For Radio Communication Speech
• Temporal-Spatial Neural Filter: Direction Informed End-to-End Multi-channel Target Speech Separation
• Towards Listening to 10 People Simultaneously: An Efficient Permutation Invariant Training of Audio Source Separation Using Sinkhorn's Algorithm
• Continuous speech separation: dataset and analysis
• Continuous Speech Separation with Conformer
• Diffiner: A Versatile Diffusion-based Generative Refiner for Speech Enhancement
• Microphone Array Signal Processing and Deep Learning for Speech Enhancement
• Attention is All You Need in Speech Separation
• Two Heads Are Better Than One: A Two-Stage Approach for Monaural Noise Reduction in the Complex Domain
• Stepwise-Refining Speech Separation Network via Fine-Grained Encoding in High-order Latent Domain
• Speech Denoising with Auditory Models
• Effective Low-Cost Time-Domain Audio Separation Using Globally Attentive Locally Recurrent Networks
• Robust Time-Frequency Reconstruction by Learning Structured Sparsity
• CatNet: music source separation system with mix-audio augmentation
• EchoFilter: End-to-End Neural Network for Acoustic Echo Cancellation
• Exploring the time-domain deep attractor network with two-stream architectures in a reverberant environment
• Integrating Statistical Uncertainty into Neural Network-Based Speech Enhancement
• SVSNet: An End-to-end Speaker Voice Similarity Assessment Model
• Joint Echo Cancellation and Noise Suppression based on Cascaded Magnitude and Complex Mask Estimation
• Time-Domain Speech Extraction with Spatial Information and Multi Speaker Conditioning Mechanism
• DPLM: A Deep Perceptual Spatial-Audio Localization Metric
• Points2Sound: From mono to binaural audio using 3D point cloud scenes
• EEG-based Auditory Attention Decoding: Towards Neuro-Steered Hearing Devices
• Improving noise robust automatic speech recognition with single-channel time-domain enhancement network
• Spatial and spectral deep attention fusion for multi-channel speech separation using deep embedding features
• Deep Learning Based Phase Reconstruction for Speaker Separation: A Trigonometric Perspective
• Music Source Separation with Generative Flow
• X-TaSNet: Robust and Accurate Time-Domain Speaker Extraction Network
• On Batching Variable Size Inputs for Training End-to-End Speech Enhancement Systems
• ADL-MVDR: All deep learning MVDR beamformer for target speech separation
• End-to-end Microphone Permutation and Number Invariant Multi-channel Speech Separation
• A comparison of handcrafted, parameterized, and learnable features for speech separation
• Noisy-to-Noisy Voice Conversion Framework with Denoising Model
• DPCRN: Dual-Path Convolution Recurrent Network for Single Channel Speech Enhancement
• A Simultaneous Denoising and Dereverberation Framework with Target Decoupling
• DCCRN+: Channel-wise Subband DCCRN with SNR Estimation for Speech Enhancement
• LaFurca: Iterative Refined Speech Separation Based on Context-Aware Dual-Path Parallel Bi-LSTM
• Sparsely Overlapped Speech Training in the Time Domain: Joint Learning of Target Speech Separation and Personal VAD Benefits
• The Effect of Training Dataset Size on Discriminative and Diffusion-Based Speech Enhancement Systems
• Exploiting Temporal Structures of Cyclostationary Signals for Data-Driven Single-Channel Source Separation
• On Neural Architectures for Deep Learning-based Source Separation of Co-Channel OFDM Signals
• Visually-Guided Sound Source Separation with Audio-Visual Predictive Coding
• Speeding Up Permutation Invariant Training for Source Separation
• Multi-Scale Attention Neural Network for Acoustic Echo Cancellation
• DBNET: DOA-driven beamforming network for end-to-end farfield sound source separation
• Content based singing voice source separation via strong conditioning using aligned phonemes
• Blind Source Separation of Radar Signals in Time Domain Using Deep Learning
• Tdcgan: Temporal Dilated Convolutional Generative Adversarial Network for End-to-end Speech Enhancement
• The 2020 ESPnet update: new features, broadened applications, performance improvements, and future plans
• Continuous Speech Separation Using Speaker Inventory for Long Multi-talker Recording
• Diffusion-Based Speech Enhancement in Matched and Mismatched Conditions Using a Heun-Based Sampler
• SEFGAN: Harvesting the Power of Normalizing Flows and GANs for Efficient High-Quality Speech Enhancement
• An Investigation of the Effectiveness of Phase for Audio Classification
• Scalable Speech Enhancement with Dynamic Channel Pruning
• MetricNet: Towards Improved Modeling For Non-Intrusive Speech Quality Assessment
• Multi-channel Multi-frame ADL-MVDR for Target Speech Separation
• Improving Frame-Online Neural Speech Enhancement with Overlapped-Frame Prediction
• End-to-End Integration of Speech Separation and Voice Activity Detection for Low-Latency Diarization of Telephone Conversations
• Multi-stage Speaker Extraction with Utterance and Frame-Level Reference Signals
• Distributed speech separation in spatially unconstrained microphone arrays
• Single-Channel Speech Enhancement with Deep Complex U-Networks and Probabilistic Latent Space Models
• ESPnet-se: end-to-end speech enhancement and separation toolkit designed for asr integration
• Ultra Dual-Path Compression For Joint Echo Cancellation And Noise Suppression
• Onssen: an open-source speech separation and enhancement library
• Divide and Conquer: A Deep CASA Approach to Talker-independent Monaural Speaker Separation
• A Generalized Bandsplit Neural Network for Cinematic Audio Source Separation
• Sequential Multi-Frame Neural Beamforming for Speech Separation and Enhancement
• Deep Attention Fusion Feature for Speech Separation with End-to-End Post-filter Method
• Separation Guided Speaker Diarization in Realistic Mismatched Conditions
• Atss-Net: Target Speaker Separation via Attention-based Neural Network
• Generic Speech Enhancement with Self-Supervised Representation Space Loss
• Deep Harmonic Finesse: Signal Separation in Wearable Systems with Limited Data
• Improving Choral Music Separation through Expressive Synthesized Data from Sampled Instruments
• Quantifying Spatial Audio Quality Impairment
• A Unified Framework for Speech Separation
• Two-stage model and optimal SI-SNR for monaural multi-speaker speech separation in noisy environment
• An End-to-end Architecture of Online Multi-channel Speech Separation
• Multichannel Speech Enhancement by Raw Waveform-mapping using Fully Convolutional Networks
• Sandglasset: A Light Multi-Granularity Self-attentive Network For Time-Domain Speech Separation
• Real-time binaural speech separation with preserved spatial cues
• Listen to What You Want: Neural Network-based Universal Sound Selector
• Sequence to Multi-Sequence Learning via Conditional Chain Mapping for Mixture Signals
• Speaker-Conditional Chain Model for Speech Separation and Extraction
• Speech enhancement aided end-to-end multi-task learning for voice activity detection
• L3DAS21 Challenge: Machine Learning for 3D Audio Signal Processing
• Enhancing End-to-End Multi-channel Speech Separation via Spatial Feature Learning
• Time-domain Speech Enhancement with Generative Adversarial Learning
• One Shot Learning for Speech Separation
• Monaural Speech Enhancement with Complex Convolutional Block Attention Module and Joint Time Frequency Losses
• A 1.6-mW Sparse Deep Learning Accelerator for Speech Separation
• dCoNNear: An Artifact-Free Neural Network Architecture for Closed-loop Audio Signal Processing
• Neural Spatio-Temporal Beamformer for Target Speech Separation
• Efficient Integration of Multi-channel Information for Speaker-independent Speech Separation
• Revisiting Representation Learning for Singing Voice Separation with Sinkhorn Distances
• TDFNet: An Efficient Audio-Visual Speech Separation Model with Top-down Fusion
• Convolutive Prediction for Monaural Speech Dereverberation and Noisy-Reverberant Speaker Separation
• Multi-path RNN for hierarchical modeling of long sequential data and its application to speaker stream separation
• Don't shoot butterfly with rifles: Multi-channel Continuous Speech Separation with Early Exit Transformer
• Improved Speech Separation with Time-and-Frequency Cross-domain Joint Embedding and Clustering
• Adversarial attacks on audio source separation
• Parallel and Flexible Sampling from Autoregressive Models via Langevin Dynamics
• The Whole Is Greater than the Sum of Its Parts: Improving Music Source Separation by Bridging Network
• Algorithms of Sampling-Frequency-Independent Layers for Non-integer Strides
• TENET: A Time-reversal Enhancement Network for Noise-robust ASR
• Improving Channel Decorrelation for Multi-Channel Target Speech Extraction
• SMRU: Split-and-Merge Recurrent-based UNet for Acoustic Echo Cancellation and Noise Suppression
• A Study of Transfer Learning in Music Source Separation
• Closing the Training/Inference Gap for Deep Attractor Networks
• Rethinking the Separation Layers in Speech Separation Networks
• Ultra-Lightweight Speech Separation via Group Communication
• Audio-visual Speech Separation with Adversarially Disentangled Visual Representation
• Online Speaker Diarization with Relation Network
• CleanUMamba: A Compact Mamba Network for Speech Denoising using Channel Pruning
• MITAS: A Compressed Time-Domain Audio Separation Network with Parameter Sharing
• RAVSS: Robust Audio-Visual Speech Separation in Multi-Speaker Scenarios with Missing Visual Cues
• TransMask: A Compact and Fast Speech Separation Model Based on Transformer
• Speech enhancement with weakly labelled data from AudioSet
• Classification of Audio Segments in Call Center Recordings using Convolutional Recurrent Neural Networks
• Speaker Separation Using Speaker Inventories and Estimated Speech
• Teacher-Student MixIT for Unsupervised and Semi-supervised Speech Separation
• Audio-visual Multi-channel Integration and Recognition of Overlapped Speech
• LiMuSE: Lightweight Multi-modal Speaker Extraction
• Multi-channel target speech extraction with channel decorrelation and target speaker adaptation
• Audio-Driven Reinforcement Learning for Head-Orientation in Naturalistic Environments
• Uformer: A Unet based dilated complex & real dual-path conformer network for simultaneous speech enhancement and dereverberation
• Target Speaker Verification with Selective Auditory Attention for Single and Multi-talker Speech
• Personalized Speech Enhancement through Self-Supervised Data Augmentation and Purification
• Combining TF-GridNet and Mixture Encoder for Continuous Speech Separation for Meeting Transcription
• Unified Gradient Reweighting for Model Biasing with Applications to Source Separation
• Muse: Multi-modal target speaker extraction with visual cues
• A Dual-Branch Parallel Network for Speech Enhancement and Restoration
• What is Learnt by the LEArnable Front-end (LEAF)? Adapting Per-Channel Energy Normalisation (PCEN) to Noisy Conditions
• The NTNU Taiwanese ASR System for Formosa Speech Recognition Challenge 2020
• Leveraging Sound Source Trajectories for Universal Sound Separation
• Design of a Speech Enhancement-Based Deep Semantic Autoencoder for Wireless Communications
• Memory Requirement Reduction of Deep Neural Networks Using Low-bit Quantization of Parameters
• Multi-channel Speech Separation Using Deep Embedding Model with Multilayer Bootstrap Networks
• Speech Separation Based on Multi-Stage Elaborated Dual-Path Deep BiLSTM with Auxiliary Identity Loss
• Distortionless Multi-Channel Target Speech Enhancement for Overlapped Speech Recognition
• Sparse Mixture of Local Experts for Efficient Speech Enhancement
• MIMO Self-attentive RNN Beamformer for Multi-speaker Speech Separation
• Speaker-conditioned Target Speaker Extraction based on Customized LSTM Cells
• On TasNet for Low-Latency Single-Speaker Speech Enhancement
• Multimodal Attention Fusion for Target Speaker Extraction
• Neural Network-based Virtual Microphone Estimator
• Distortion-controlled Training for End-to-end Reverberant Speech Separation with Auxiliary Autoencoding Loss
• Short-time deep-learning based source separation for speech enhancement in reverberant environments with beamforming
• Incorporating Real-world Noisy Speech in Neural-network-based Speech Enhancement Systems
• Towards Neural Diarization for Unlimited Numbers of Speakers Using Global and Local Attractors
• Basis-MelGAN: Efficient Neural Vocoder Based on Audio Decomposition
• Multi-Scale Temporal Convolution Network for Classroom Voice Detection
• PIAVE: A Pose-Invariant Audio-Visual Speaker Extraction Network
• Visual-Informed Speech Enhancement Using Attention-Based Beamforming
• Unifying Diarization, Separation, and ASR with Multi-Speaker Encoder
• GALD-SE: Guided Anisotropic Lightweight Diffusion for Efficient Speech Enhancement
• Conv-NILM-Net, a causal and multi-appliance model for energy source separation
• U-shaped Transformer with Frequency-Band Aware Attention for Speech Enhancement
• Inter-channel Conv-TasNet for multichannel speech enhancement
• SNRi Target Training for Joint Speech Enhancement and Recognition
• Source Mixing and Separation Robust Audio Steganography
• USEV: Universal Speaker Extraction with Visual Cue
• Should We Always Separate?: Switching Between Enhanced and Observed Signals for Overlapping Speech Recognition
• Audio-visual Multi-channel Recognition of Overlapped Speech
• Tackling real noisy reverberant meetings with all-neural source separation, counting, and diarization system
• A Dual-Staged Context Aggregation Method Towards Efficient End-To-End Speech Enhancement
• Deep Ad-hoc Beamforming
• BERT for Joint Multichannel Speech Dereverberation with Spatial-aware Tasks
• Source-Aware Neural Speech Coding for Noisy Speech Compression
• Toward Speech Separation in The Pre-Cocktail Party Problem with TasTas
• Informed Source Extraction With Application to Acoustic Echo Reduction
• CSLNSpeech: solving extended speech separation problem with the help of Chinese sign language
• Beam-Guided TasNet: An Iterative Speech Separation Framework with Multi-Channel Output
• Multi-Metric Optimization using Generative Adversarial Networks for Near-End Speech Intelligibility Enhancement
• MULTIMODAL ANALYSIS: Informed content estimation and audio source separation
• Speech Enhancement using Separable Polling Attention and Global Layer Normalization followed with PReLU
• Manifold-Aware Deep Clustering: Maximizing Angles between Embedding Vectors Based on Regular Simplex
• Active Restoration of Lost Audio Signals Using Machine Learning and Latent Information
• Correlating Subword Articulation with Lip Shapes for Embedding Aware Audio-Visual Speech Enhancement
• Mixed Precision DNN Qunatization for Overlapped Speech Separation and Recognition
• A Study of the Scale Invariant Signal to Distortion Ratio in Speech Separation with Noisy References
• Overlapped speech recognition from a jointly learned multi-channel neural speech extraction and representation
• Dissecting Performance Degradation in Audio Source Separation under Sampling Frequency Mismatch
• Binaural Speech Enhancement Using Complex Convolutional Recurrent Networks
• Refining DNN-based Mask Estimation using CGMM-based EM Algorithm for Multi-channel Noise Reduction
• Multi-accent Speech Separation with One Shot Learning
• Lightweight Dual-channel Target Speaker Separation for Mobile Voice Communication
• WASE: Learning When to Attend for Speaker Extraction in Cocktail Party Environments
• Few-shot learning of new sound classes for target sound extraction
• Multi-channel Speech Enhancement with 2-D Convolutional Time-frequency Domain Features and a Pre-trained Acoustic Model
• Raw Differentiable Architecture Search for Speech Deepfake and Spoofing Detection
• Investigation of Practical Aspects of Single Channel Speech Separation for ASR
• Online Self-Attentive Gated RNNs for Real-Time Speaker Separation
• Cross-domain Single-channel Speech Enhancement Model with Bi-projection Fusion Module for Noise-robust ASR
• BeamTransformer: Microphone Array-based Overlapping Speech Detection
• Improving RNN Transducer With Target Speaker Extraction and Neural Uncertainty Estimation
• WPD++: An Improved Neural Beamformer for Simultaneous Speech Separation and Dereverberation
• On permutation invariant training for speech source separation
• Speaker and Direction Inferred Dual-channel Speech Separation
• Dual-Path Modeling for Long Recording Speech Separation in Meetings
• Tune-In: Training Under Negative Environments with Interference for Attention Networks Simulating Cocktail Party Effect
• Efficient Independent Vector Extraction of Dominant Target Speech
• CochleaNet: A Robust Language-independent Audio-Visual Model for Speech Enhancement
• End-to-end Music-mixed Speech Recognition
• Time Domain Audio Visual Speech Separation
• Phase recovery with Bregman divergences for audio source separation
• Multi-microphone Complex Spectral Mapping for Utterance-wise and Continuous Speech Separation
• Stabilizing Label Assignment for Speech Separation by Self-supervised Pre-training
• Attention-based scaling adaptation for target speech extraction