Translate Reverberated Speech to Anechoic Ones: Speech Dereverberation with BERT

TL;DR

This paper introduces a novel speech dereverberation method using a BERT-based model combined with a neural vocoder, outperforming traditional approaches and matching state-of-the-art results on CHiME challenge data.

Contribution

It applies BERT, a NLP transformer model, to speech dereverberation, incorporating a pre-sequence network and neural vocoder for phase reconstruction, which is a new approach in this domain.

Findings

Outperforms traditional WPE dereverberation method.

Achieves comparable performance with BLSTM-based models.

Demonstrates effectiveness of BERT in speech signal processing.

Abstract

Single channel speech dereverberation is considered in this work. Inspired by the recent success of Bidirectional Encoder Representations from Transformers (BERT) model in the domain of Natural Language Processing (NLP), we investigate its applicability as backbone sequence model to enhance reverberated speech signal. We present a variation of the basic BERT model: a pre-sequence network, which extracts local spectral-temporal information and/or provides order information, before the backbone sequence model. In addition, we use pre-trained neural vocoder for implicit phase reconstruction. To evaluate our method, we used the data from the 3rd CHiME challenge, and compare our results with other methods. Experiments show that the proposed method outperforms traditional method WPE, and achieve comparable performance with state-of-the-art BLSTM-based sequence models.