Automatic language identification using deep neural networks

López-Moreno, Ignacio; González-Domínguez, Javier; Oldrich, Plchot; Martínez, David R.; González Rodríguez, Joaquín

UAM_Biblioteca

Author

López-Moreno, Ignacio; González-Domínguez, Javier; Oldrich, Plchot; Martínez, David R.; González Rodríguez, Joaquín

Entity

UAM. Departamento de Tecnología Electrónica y de las Comunicaciones

Publisher

IEEE

Date

2014-01-01

Citation

IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), IEEE, 2014. 5337-5341

ISSN

1520-6149 (print)

DOI

10.1109/ICASSP.2014.6854622

Editor's Version

http://dx.doi.org/10.1109/ICASSP.2014.6854622

Subjects

Automatic Language Identification; DNNs; i-vectors; Telecomunicaciones

URI

http://hdl.handle.net/10486/666848

Note

Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. I. López-Moreno, J. González-Domínguez, P. Oldrich, D. R. Martínez, J. González-Rodríguez, "Automatic language identification using deep neural networks", IEEE International Conference on Acoustics, Speech, and Signal Processing ICASSP, Florence (Italy), 2014

Rights

Abstract

This work studies the use of deep neural networks (DNNs) to address automatic language identification (LID). Motivated by their recent success in acoustic modelling, we adapt DNNs to the problem of identifying the language of a given spoken utterance from short-term acoustic features. The proposed approach is compared to state-of-the-art i-vector based acoustic systems on two different datasets: Google 5M LID corpus and NIST LRE 2009. Results show how LID can largely benefit from using DNNs, especially when a large amount of training data is available. We found relative improvements up to 70%, in Cavg, over the baseline system.

Show full item record