An extended experimental investigation of DNN uncertainty propagation for noise robust ASR - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

An extended experimental investigation of DNN uncertainty propagation for noise robust ASR

Résumé

Automatic speech recognition (ASR) in noisy environments remains a challenging goal. Recently, the idea of estimating the uncertainty about the features obtained after speech enhancement and propagating it to dynamically adapt deep neural network (DNN) based acoustic models has raised some interest. However, the results in the literature were reported on simulated noisy datasets for a limited variety of uncertainty estimators. We found that they vary significantly in different conditions. Hence, the main contribution of this work is to assess DNN uncertainty decoding performance for different data conditions and different uncertainty estimation/propagation techniques. In addition, we propose a neural network based uncertainty estima-tor and compare it with other uncertainty estimators. We report detailed ASR results on the CHiME-2 and CHiME-3 datasets. We find that, on average, uncertainty propagation provides similar relative improvement on real and simulated data and that the proposed uncertainty estimator performs significantly better than the one in [1]. We also find that the improvement is consistent, but it depends on the signal-to-noise ratio (SNR) and the noise environment.
Fichier principal
Vignette du fichier
nathwani_HSCMA17.pdf (185.75 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01446441 , version 1 (25-01-2017)

Identifiants

  • HAL Id : hal-01446441 , version 1

Citer

Karan Nathwani, Juan A Morales-Cordovilla, Sunit Sivasankaran, Irina Illina, Emmanuel Vincent. An extended experimental investigation of DNN uncertainty propagation for noise robust ASR. 5th Joint Workshop on Hands-free Speech Communication and Microphone Arrays (HSCMA 2017), Mar 2017, San Francisco, United States. ⟨hal-01446441⟩
482 Consultations
440 Téléchargements

Partager

Gmail Facebook X LinkedIn More