Analysis of DNN Speech Signal Enhancement for Robust Speaker Recognition. (November 2019)

• Record Type: Journal Article
• Authors: Novotný, Ondřej; Plchot, Oldřich; Glembek, Ondřej; Černocký, Jan "Honza"; Burget, Lukáš
• Is Part Of: Computer speech & language. Volume 58(2019)
• Page Start: 403
• Page End: 421
• Publication Date: 2019-11
• Subjects: Speaker verification -- Signal enhancement -- Autoencoder -- Neural network -- Robustness -- Embedding; Speech processing systems -- Periodicals; Automatic speech recognition -- Periodicals; Computers -- Periodicals; Linguistics -- Periodicals; Speech-Language Pathology -- Periodicals; Traitement automatique de la parole -- Périodiques; Reconnaissance automatique de la parole -- Périodiques; Automatic speech recognition; Speech processing systems; Electronic journals; Periodicals; 006.454
• DOI: 10.1016/j.csl.2019.06.004
• Languages: English
• ISSNs: 0885-2308

Abstract

Highlights: This work presents an analysis of a DNN-based autoencoder for speech enhancement, dereverberation and denoising for robust speaker verification (SV). It includes a detailed performance analysis for different SV system paradigms (i-vectors vs. x-vectors), features and test conditions. Presented results are computed on the public and widely used NIST SRE 2010, 2016, PRISM and SITW. It also presents results achieved with pure autoencoder enhancement, multi-condition PLDA training and their simultaneous use. Best system performance is achieved by combining the DNN autoencoder with x-vectors and multi-condition training in PLDA. Abstract: In this work, we present an analysis of a DNN-based autoencoder for speech enhancement, dereverberation and denoising. The target application is a robust speaker verification (SV) system. We start our approach by carefully designing a data augmentation process to cover a wide range of acoustic conditions and to obtain rich training data for various components of our SV system. We augment several well-known databases used in SV with artificially noised and reverberated data and we use them to train a denoising autoencoder (mapping noisy and reverberated speech to its clean version) as well as an x-vector extractor which is currently considered as state-of-the-art in SV. Later, we use the autoencoder as a preprocessing step for a text-independent SV system. We compare results achieved with autoencoder enhancement, multi-condition PLDA training and their simultaneous use. We present a detailed analysis with various conditions of NIST SRE 2010, 2016, PRISM and with re-transmitted data. We conclude that the proposed preprocessing can significantly improve both i-vector and x-vector baselines and that this technique can be used to build a robust SV system for various target domains.