Unsupervised adaptation of a denoising autoencoder by Bayesian Feature Enhancement for reverberant asr under mismatch conditions

J. Heymann, R. Haeb-Umbach, P. Golik, R. Schlueter, in: Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference On, 2015, pp. 5053–5057.

Conference Paper | English
Author
Abstract
The parametric Bayesian Feature Enhancement (BFE) and a datadriven Denoising Autoencoder (DA) both bring performance gains in severe single-channel speech recognition conditions. The first can be adjusted to different conditions by an appropriate parameter setting, while the latter needs to be trained on conditions similar to the ones expected at decoding time, making it vulnerable to a mismatch between training and test conditions. We use a DNN backend and study reverberant ASR under three types of mismatch conditions: different room reverberation times, different speaker to microphone distances and the difference between artificially reverberated data and the recordings in a reverberant environment. We show that for these mismatch conditions BFE can provide the targets for a DA. This unsupervised adaptation provides a performance gain over the direct use of BFE and even enables to compensate for the mismatch of real and simulated reverberant data.
Publishing Year
Proceedings Title
Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on
Page
5053-5057
LibreCat-ID

Cite this

Heymann J, Haeb-Umbach R, Golik P, Schlueter R. Unsupervised adaptation of a denoising autoencoder by Bayesian Feature Enhancement for reverberant asr under mismatch conditions. In: Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference On. ; 2015:5053-5057. doi:10.1109/ICASSP.2015.7178933
Heymann, J., Haeb-Umbach, R., Golik, P., & Schlueter, R. (2015). Unsupervised adaptation of a denoising autoencoder by Bayesian Feature Enhancement for reverberant asr under mismatch conditions. In Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on (pp. 5053–5057). https://doi.org/10.1109/ICASSP.2015.7178933
@inproceedings{Heymann_Haeb-Umbach_Golik_Schlueter_2015, title={Unsupervised adaptation of a denoising autoencoder by Bayesian Feature Enhancement for reverberant asr under mismatch conditions}, DOI={10.1109/ICASSP.2015.7178933}, booktitle={Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on}, author={Heymann, Jahn and Haeb-Umbach, Reinhold and Golik, P. and Schlueter, R.}, year={2015}, pages={5053–5057} }
Heymann, Jahn, Reinhold Haeb-Umbach, P. Golik, and R. Schlueter. “Unsupervised Adaptation of a Denoising Autoencoder by Bayesian Feature Enhancement for Reverberant Asr under Mismatch Conditions.” In Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference On, 5053–57, 2015. https://doi.org/10.1109/ICASSP.2015.7178933.
J. Heymann, R. Haeb-Umbach, P. Golik, and R. Schlueter, “Unsupervised adaptation of a denoising autoencoder by Bayesian Feature Enhancement for reverberant asr under mismatch conditions,” in Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference on, 2015, pp. 5053–5057.
Heymann, Jahn, et al. “Unsupervised Adaptation of a Denoising Autoencoder by Bayesian Feature Enhancement for Reverberant Asr under Mismatch Conditions.” Acoustics, Speech and Signal Processing (ICASSP), 2015 IEEE International Conference On, 2015, pp. 5053–57, doi:10.1109/ICASSP.2015.7178933.

Link(s) to Main File(s)
Access Level
Restricted Closed Access

Export

Marked Publications

Open Data LibreCat

Search this title in

Google Scholar