An Investigation Of Deep Neural Networks For Noise Robust Speech Recognition

Recently, a new acoustic model based on deep neural networks (DNN) has been introduced. While the DNN has generated signifi- cant improvements over GMM-based systems on several tasks, there has been no evaluation of the robustness of such systems to environ- mental distortion. In this paper, we investigate the noise robustness of DNN-based acoustic models and find that they can match state- of-the-art performance on the Aurora 4 task without any explicit noise compensation. This performance can be further improved by incorporating information about the environment into DNN training using a new method called noise-aware training. When combined with the recently proposed dropout training technique, a 7.5% rela- tive improvement over the previously best published result on this task is achieved using only a single decoding pass an

0007398.pdf
PDF file

In  ICASSP 2013

Publisher  IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

Details

TypeInproceedings
> Publications > An Investigation Of Deep Neural Networks For Noise Robust Speech Recognition