A new uncertainty decoding scheme for DNN-HMM hybrid systems with multichannel speech enhancement

Hümmer C, Schwarz A, Maas R, Barfuß H, Astudillo RF, Kellermann W (2016)

Publication Language: English

Publication Status: Published

Publication Type: Conference contribution, Conference Contribution

Publication year: 2016

Publisher: Institute of Electrical and Electronics Engineers Inc.

Pages Range: 5760-5764

Article Number: 7472781

Event location: Shanghai CN

ISBN: 9781479999880

DOI: 10.1109/ICASSP.2016.7472781

Abstract

Uncertainty decoding combines a probabilistic feature description with the acoustic model of a speech recognition system. For DNN-HMM hybrid systems, this can be realized by averaging the DNN outputs produced by a finite set of feature samples (drawn from an estimated probability distribution). In this article, we employ this sampling approach in combination with a multi-microphone speech enhancement system. We propose a new strategy for generating feature samples from multichannel signals, based on modeling the spatial coherence estimates between different microphone pairs as realizations of a latent random variable. From each coherence estimate, a spectral enhancement gain is computed and an enhanced feature vector is obtained, thus producing a finite set of feature samples, of which we average the respective DNN outputs. In the experimental part, this new uncertainty decoding strategy is shown to consistently improve the recognition accuracy of a DNN-HMM hybrid system for the 8-channel REVERB Challenge task.

Authors with CRIS profile

Christian Hümmer Professur für Signalverarbeitung Andreas Schwarz Professur für Signalverarbeitung Roland Maas Lehrstuhl für Multimediakommunikation und Signalverarbeitung (LMS) Hendrik Barfuß Professur für Signalverarbeitung Walter Kellermann Professur für Signalverarbeitung

Involved external institutions

Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa (INESC-ID) PT Portugal (PT)

How to cite

APA:

Hümmer, C., Schwarz, A., Maas, R., Barfuß, H., Astudillo, R.F., & Kellermann, W. (2016). A new uncertainty decoding scheme for DNN-HMM hybrid systems with multichannel speech enhancement. In Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 5760-5764). Shanghai, CN: Institute of Electrical and Electronics Engineers Inc..

MLA:

Hümmer, Christian, et al. "A new uncertainty decoding scheme for DNN-HMM hybrid systems with multichannel speech enhancement." Proceedings of the International Conference on Acoustics, Speech and Signal Processing (ICASSP), Shanghai Institute of Electrical and Electronics Engineers Inc., 2016. 5760-5764.

BibTeX: Download