Skip to Main content Skip to Navigation
Preprints, Working Papers, ...

Narrow-band Deep Filtering for Multichannel Speech Enhancement

Xiaofei Li 1, 2 Radu Horaud 1
1 PERCEPTION [2016-2019] - Interpretation and Modelling of Images and Videos [2016-2019]
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann , Grenoble INP [2007-2019] - Institut polytechnique de Grenoble - Grenoble Institute of Technology [2007-2019]
Abstract : In this paper, we address the problem of multichannel speech enhancement in the short-time Fourier transform (STFT) domain. A long short-time memory (LSTM) network takes as input a sequence of STFT coefficients associated with a frequency bin of multichannel noisy-speech signals. The network's output is the corresponding sequence of single-channel cleaned speech. We propose several clean-speech network targets, namely, the magnitude ratio mask, the complex STFT coefficients and the (smoothed) spatial filter. A prominent feature of the proposed model is that the same LSTM architecture, with identical parameters, is trained across frequency bins. The proposed method is referred to as narrow-band deep filtering. This choice stays in contrast with traditional wideband speech enhancement methods. The proposed deep filtering is able to discriminate between speech and noise by exploiting their different temporal and spatial characteristics: speech is non-stationary and spatially coherent while noise is relatively stationary and weakly correlated across channels. This is similar in spirit with unsupervised techniques, such as spectral subtraction and beamforming. We describe extensive experiments with both mixed signals (noise is added to clean speech) and real signals (live recordings). We empirically evaluate the proposed architecture variants using speech enhancement and speech recognition metrics, and we compare our results with the results obtained with several state of the art methods. In the light of these experiments we conclude that narrow-band deep filtering has very good speech enhancement and speech recognition performance, and excellent generalization capabilities in terms of speaker variability and noise type.
Complete list of metadatas

Cited literature [41 references]  Display  Hide  Download

https://hal.inria.fr/hal-02378413
Contributor : Team Perception <>
Submitted on : Wednesday, September 23, 2020 - 10:49:12 AM
Last modification on : Tuesday, October 6, 2020 - 12:44:47 PM

File

multichannel_lstm.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02378413, version 2
  • ARXIV : 1911.10791

Collections

Citation

Xiaofei Li, Radu Horaud. Narrow-band Deep Filtering for Multichannel Speech Enhancement. 2020. ⟨hal-02378413v2⟩

Share

Metrics

Record views

49

Files downloads

174