Neural network based spectral mask estimation for acoustic beamforming

By Jahn Heymann, Lukas Drude, Reinhold Häb-Umbach. IEEE International Conference on Acoustics, Speech, and Signal Processing, 2016

514 citations recorded by the source corpus.

Subfields: Computational Mechanics, Signal Processing

View Paper: https://www.semanticscholar.org/paper/3818510b5c44f2c5dff5232ef93757c37ceab3cb

How researchers see this paper

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

In Computational Mechanics

In Signal Processing

Citations in the corpus

FaSNet: Low-Latency Adaptive Beamforming for Multi-Microphone Audio Processing (2019, background): “The second category, which we refer to as the masking-based (MB) beamforming, estimates the FaS beamforming filters in frequency domain by estimating time-frequency (T-F) masks for the sources of interest [10–23].”
Speaker-Aware Neural Network Based Beamformer for Speaker Extraction in Speech Mixtures (2017, methodology): “Following [7], we use a Generalized Eigenvector beamformer (GEV), whose filters are obtained as”
Neural Target Speech Extraction: An overview (2023, methodology): “We briefly review the mask-based beamforming approach, which was introduced initially for noise reduction and BSS [42], [43].”
Deep Learning for Environmentally Robust Speech Recognition (2017, methodology): “ch for each microphone channel. In this case, only simulated data is possible to be employed for the network training. To relax this requirement to some extent, a follow-up work has been presented in [123], where only the clean speech was employed for the mask estimation. This slight improvement ena”
Cracking the cocktail party problem by multi-beam deep attractor network (2017, methodology): “We used the implementation of [29] for OGEV and that of [30] for OMVDR.”
SpeechBrain: A General-Purpose Speech Toolkit (2021, methodology): “The SCMs for speech and noise can be obtained using time-frequency masks [56, 103, 104].”
Recent Advances in End-to-End Automatic Speech Recognition (2021, methodology): “The most popular beamforming technology for ASR is signal processing based superdirective beamforming, while there is a trend to replace it with the neural beamforming for joint optimization with the backend hybrid models [86, 189, 235, 307].”
Speech Processing for Digital Home Assistants: Combining signal processing with deep-learning techniques (2019, background): “Given a set of features extracted from the microphone signals at its input and the true class affiliations z , t f at the output, the network is trained to output the SPP , t c f [22], [23].”
Front-end processing for the CHiME-5 dinner party scenario (2018, background): “In particular, supervised learning approaches for time-frequency mask estimation employing a neural network [1, 6] with subsequent beamforming, achieved excellent re-sults.”
SpeakerBeam: Speaker Aware Neural Network for Target Speaker Extraction in Speech Mixtures (2019, methodology): “This procedure for neural network mask-based beamforming was proposed for speech denoising [44], [45] and has been shown to be very effective.”
Attention-based Neural Beamforming Layers for Multi-channel Speech Recognition (2021, methodology): “Various neural adaptive beamforming techniques were applied to estimate masks for speech and noise and also to predict beamforming filter coefficients [4, 5, 6].”
End-to-end Microphone Permutation and Number Invariant Multi-channel Speech Separation (2019, methodology): “A general pipeline in the design of many recent neural beamformers is to first perform pre-separation on each channel independently, and then apply conventional beamforming techniques such as minimum variance distortionless response beamforming (MVDR) or multi-channel Wiener filtering (MWF) based on”
The RWTH/UPB/FORTH System Combination for the 4th CHiME Challenge Evaluation (2016, methodology): “The spatial correlation matrices needed for the beamforming operation are estimated using time-frequency masks from a neural network [12][4].”
Supervised Speech Separation Based on Deep Learning: An Overview (2017, unknown): “[71] 1366 in the CHiME-3 challenge by using DNN estimated masks 1367”
Improved MVDR Beamforming Using Single-Channel Mask Prediction Networks (2016, methodology): “We became aware of [17] recently and we plan to quantitatively compare their approach with ours in the future.”