Project Details

Discriminative training of speaker-normalized models for automatic speech recognition

Project Period: 7. 1. 2011 – 7. 1. 2013

Project Type: grant

Code: SIGA890

Agency: Evropská unie

Program: Seventh Research Framework Programme

Czech title

Diskriminativní trénování modelů normalizovaných na mluvčího pro automatické rozpoznávání řeči

Type

grant

Keywords

speech recognition

Abstract

The proposed project deals with automatic speech recognition. It focuses on
investigation of discriminative training of speaker normalized models allowing
for building more accurate speech recognition systems that are more adapted to
the target user. Particular attention is devoted to the application of
discriminatively trained speaker adaptations in recently proposed sub-space
acoustic modeling of speech.

Team members

Rath Shakti Prasad – research leader

Publications

2013

RATH, S.; BURGET, L.; KARAFIÁT, M.; GLEMBEK, O.; ČERNOCKÝ, J. A Region-specific Feature-space Transformation for Speaker Adaptation and Singularity Analysis of Jacobian Matrix. Proceedings of Interspeeech 2013. Proceedings of the 14th Annual Conference of the International Speech Communication Association (Interspeech 2013). Lyon: International Speech Communication Association, 2013. p. 1228-1232. ISBN: 978-1-62993-443-3. ISSN: 2308-457X. Detail
RATH, S.; POVEY, D.; VESELÝ, K.; ČERNOCKÝ, J. Improved Feature Processing for Deep Neural Networks. Proceedings of Interspeech 2013. Proceedings of the 14th Annual Conference of the International Speech Communication Association (Interspeech 2013). Lyon: International Speech Communication Association, 2013. p. 109-113. ISBN: 978-1-62993-443-3. ISSN: 2308-457X. Detail

2012

RATH, S.; KARAFIÁT, M.; GLEMBEK, O.; ČERNOCKÝ, J. A factorized representation of FMLLR transform based on QR-decomposition. Proceedings of Interspeech 2012. Proceedings of Interspeech. Portland, Oregon: International Speech Communication Association, 2012. p. 1-4. ISBN: 978-1-62276-759-5. ISSN: 1990-9772. Detail