Learning Spectral Mapping for Speech Dereverberation and Denoising

By Kun Han, Yuxuan Wang, Deliang Wang, William S. Woods, Ivo Merks, Zhang Tao. 2015

243 citations recorded by the source corpus.

Subfields: Cognitive Neuroscience, Computational Mechanics, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Computational Mechanics

In Signal Processing

Citations in the corpus

Time-Frequency Masking in the Complex Domain for Speech Dereverberation and Denoising (2017, neutral): “[13] learn a spectral mapping from the log-”
End-to-End Speech Separation with Unfolded Iterative Phase Reconstruction (2018, methodology): “There are some previous attempts at naively applying such iterative algorithms as a post-processing step on the magnitudes produced by deep learning based speech enhancement and separation [18, 19, 20, 3].”
A Deep Ensemble Learning Method for Monaural Speech Separation (2016, unknown): “Han et al. [13], [11] used DNN to learn a mapping from reverberant and reverberant-noisy speech to anechoic speech.”
Deep neural network based spectral feature mapping for robust speech recognition (2015, methodology): “It is worth mentioning that all the proposed methods substantially outperform the spectral mapping method in [5], where the learned clean spectral magnitudes are used to resynthesize the timedomain waveforms for ASR modeling.”
Deep Learning Based Phase Reconstruction for Speaker Separation: A Trigonometric Perspective (2018, methodology): “One straightforward approach for phase reconstruction is to use estimated magnitudes to drive an iterative phase reconstruction algorithm [15], [16], [5], [6].”
Supervised Speech Separation Based on Deep Learning: An Overview (2017, unknown): “show an improvement in dereverberation performance over the DNN in [58].”
WHAMR!: Noisy and Reverberant Single-Channel Speech Separation (2019, background): “Two-stage approaches have previously been explored for denoising plus dereverberation [19,20], separation plus dereverberation [21], and denoising plus separation [11].”
A Reverberation-Time-Aware Approach to Speech Dereverberation Based on Deep Neural Networks (2017, methodology): “The generalization capabilities of the proposed DNN models are illustrated in Section V. Finally we summarize our findings in Section VI.”
Deep Learning for Environmentally Robust Speech Recognition (2017, background): “am,re-training BLSTM 41.42% [91] masking: IRM (MA, SA; Mel) LSTM 17.68 (SDR) [103] hybrid BLSTM 22.20% [72] mapping: log Mel re-training BLSTM 22.16% [94] masking: PSM (SA; log Mel) BLSTM 14.76 (SDR) [63] mapping: log mag hybrid DNN ≈25% [110] masking: IRM (SA; log Mel) MTL; hybrid BLSTM 16.04% [144”
PoCoNet: Better Speech Enhancement with Frequency-Positional Embeddings, Semi-Supervised Conversational Data, and Biased Loss (2020, background): “Neural network based approaches have greatly improved the output quality of speech enhancement systems [1, 2, 3, 4, 5].”
Alternative Objective Functions for Deep Clustering (2018, unknown): “There are some previous attempts at applying Grifﬁn-Lim to deep-learning-based speech enhancement [21, 22, 23].”
Real-time Single-channel Dereverberation and Separation with Time-domain Audio Separation Network (2018, methodology): “In recent years, deep learning systems have shown to have better generalization ability and higher performance in various conditions for both separation and dereverberation [1, 2, 3, 4, 5, 6, 7, 8, 9].”
Speech Enhancement of Noisy and Reverberant Speech for Text-to-Speech (2018, methodology): “More recently, authors in [21] made use of a feed-forward neural network with 11 frames input and one frame output to perform both denoising and reverberation.”
Investigating Generative Adversarial Networks based Speech Dereverberation for Robust Speech Recognition (2018, background): “, feedforward [13], recurrent [14] and convolutional [15], have been explored in the field.”
A New Framework for CNN-Based Speech Enhancement in the Time Domain (2019, unknown): “The combination of the noisy phase with the enhanced magnitude of STFT is unlikely a valid STFT, as recently demonstrated in [29].”
Speech Dereverberation Using Fully Convolutional Networks (2018, methodology): “[13] applied a DNN to map noisy and reverberant”