Consistent DNN Uncertainty Training and Decoding for Robust ASR

Karan Nathwani 1 Emmanuel Vincent 2 Irina Illina 2
2 MULTISPEECH - Speech Modeling for Facilitating Oral-Based Communication
Inria Nancy - Grand Est, LORIA - NLPKD - Department of Natural Language Processing & Knowledge Discovery
Abstract : We consider the problem of robust automatic speech recognition (ASR) in noisy conditions. The performance improvement brought by speech enhancement is often limited by residual distortions of the enhanced features, which can be seen as a form of statistical uncertainty. Uncertainty estimation and propagation methods have recently been proposed to improve the ASR performance with deep neural network (DNN) acoustic models. However, the performance is still limited due to the use of uncertainty only during decoding. In this paper, we propose a consistent approach to account for uncertainty in the enhanced features during both training and decoding. We estimate the variance of the distortions using a DNN uncertainty estimator that operates directly in the feature maximum likelihood linear regression (fMLLR) domain and we then sample the uncertain features using the unscented transform (UT). We report the resulting ASR performance on the CHiME-2 and CHiME-3 datasets for different uncertainty estimation/propagation techniques. The proposed DNN uncertainty training method brings 4% and 8% relative improvement on these two datasets, respectively, compared to a competitive fMLLR-domain DNN acoustic modeling baseline.
Type de document :
Communication dans un congrès
2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), Dec 2017, Okinawa, Japan
Liste complète des métadonnées


https://hal.inria.fr/hal-01585956
Contributeur : Karan Nathwani <>
Soumis le : mardi 12 septembre 2017 - 11:32:56
Dernière modification le : lundi 18 septembre 2017 - 01:07:53

Fichier

Camera_Ready_ASRU.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01585956, version 1

Citation

Karan Nathwani, Emmanuel Vincent, Irina Illina. Consistent DNN Uncertainty Training and Decoding for Robust ASR. 2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), Dec 2017, Okinawa, Japan. <hal-01585956>

Partager

Métriques

Consultations de
la notice

75

Téléchargements du document

20