Dual-label Deep LSTM Dereverberation For Speaker Verification

TL;DR

This paper introduces a dual-label deep LSTM approach for dereverberation in speaker verification, using neural networks to map reverberant speech features to clean speech and auxiliary features, improving verification accuracy.

Contribution

It proposes a novel dual-label LSTM neural network model that simultaneously estimates clean speech features and auxiliary spectral information for dereverberation.

Findings

Reduced equal error rates in speaker verification

Effective mapping of reverberant to clean speech features

Improved robustness against reverberation distortions

Abstract

In this paper, we present a reverberation removal approach for speaker verification, utilizing dual-label deep neural networks (DNNs). The networks perform feature mapping between the spectral features of reverberant and clean speech. Long short term memory recurrent neural networks (LSTMs) are trained to map corrupted Mel filterbank (MFB) features to two sets of labels: i) the clean MFB features, and ii) either estimated pitch tracks or the fast Fourier transform (FFT) spectrogram of clean speech. The performance of reverberation removal is evaluated by equal error rates (EERs) of speaker verification experiments.