A Combined Approach for Estimating a Feature-Domain Reverberation Model in Non-diffuse Environments

Conference contribution


Publication Details

Author(s): Sehr A, Wen JYC, Kellermann W, Naylor PA
Publication year: 2008
Conference Proceedings Title: Proc. Int. Workshop on Acoustic Echo and Noise Control (IWAENC)
Language: English


Abstract

A combined approach for estimating a feature-domain reverbera-tion model suitable for the robust distant-talking automatic speech recognition concept REMOS (REverberation MOdeling for Speech recognition) [1] is proposed. Based on a few calibration utterances recorded in the target environment, the combined approach employs ML estimation and blind estimation of the reverberation time to de-termine a two-slope reverberation model. Since measurements of room impulse responses become unnecessary, the effort for training is greatly reduced compared to [1] and compared to training HMMs on artificially reverberated data. Connected digit recognition exper-iments show that the proposed reverberation models in connection with the REMOS concept significantly outperform HMM-based rec-ognizers trained on reverberant data.


FAU Authors / FAU Editors

Kellermann, Walter Prof. Dr.-Ing.
Professur für Nachrichtentechnik
Sehr, Armin Dr.-Ing.
Professur für Nachrichtentechnik


How to cite

APA:
Sehr, A., Wen, J.Y.C., Kellermann, W., & Naylor, P.A. (2008). A Combined Approach for Estimating a Feature-Domain Reverberation Model in Non-diffuse Environments. In Proc. Int. Workshop on Acoustic Echo and Noise Control (IWAENC). Seattle, Washington, US.

MLA:
Sehr, Armin, et al. "A Combined Approach for Estimating a Feature-Domain Reverberation Model in Non-diffuse Environments." Proceedings of the Int. Workshop on Acoustic Echo and Noise Control (IWAENC), Seattle, Washington 2008.

BibTeX: 

Last updated on 2019-10-05 at 17:23