Multi-Style Reverberation Models and Efficient Model Adaptation for Robust Distant-Talking Speech Recognition with REMOS

Maas R, Sehr A, Kellermann W (2010)

Publication Language: English

Publication Type: Conference contribution

Publication year: 2010

City/Town: Bochum, Germany

Pages Range: 28

Conference Proceedings Title: ITG Conference on Speech Communication

Event location: Bochum DE

Abstract

To further increase the flexibility of the REMOS (REverberation MOdeling for Speech recognition) concept for distant-talking speech recognition, multi-style reverberation models (RVMs) trained on data from different rooms as well as simplified RVMs are analyzed in this contribution. If the multi-style probability density functions (pdfs) used for score calculation are adapted to the reverberation conditions of the current room, a remarkable improvement in recognition performance can be achieved. Evaluations of a very efficient reverberation model adaptation scheme by connected digit recognition experiments show that REMOS can be adjusted to different reverberation conditions with minimal effort.

Authors with CRIS profile

Roland Maas Lehrstuhl für Multimediakommunikation und Signalverarbeitung (LMS) Walter Kellermann Professur für Signalverarbeitung

How to cite

APA:

Maas, R., Sehr, A., & Kellermann, W. (2010). Multi-Style Reverberation Models and Efficient Model Adaptation for Robust Distant-Talking Speech Recognition with REMOS. In ITG Conference on Speech Communication (pp. 28). Bochum, DE: Bochum, Germany.

MLA:

Maas, Roland, Armin Sehr, and Walter Kellermann. "Multi-Style Reverberation Models and Efficient Model Adaptation for Robust Distant-Talking Speech Recognition with REMOS." Proceedings of the ITG Conference on Speech Communication, Bochum Bochum, Germany, 2010. 28.

BibTeX: Download