Feature Learning in Deep Neural Networks - Studies on Speech Recognition Tasks
This addresses the challenge of variability in speech signals for speech recognition systems, offering a method that matches or exceeds state-of-the-art systems without incremental improvements.
The paper tackles the problem of improving speech recognition accuracy by investigating deep neural networks (DNNs), showing that they outperform shallow networks and Gaussian mixture models (GMMs) due to their ability to learn robust internal representations, leading to better performance with deeper networks without needing explicit adaptation.
Recent studies have shown that deep neural networks (DNNs) perform significantly better than shallow networks and Gaussian mixture models (GMMs) on large vocabulary speech recognition tasks. In this paper, we argue that the improved accuracy achieved by the DNNs is the result of their ability to extract discriminative internal representations that are robust to the many sources of variability in speech signals. We show that these representations become increasingly insensitive to small perturbations in the input with increasing network depth, which leads to better speech recognition performance with deeper networks. We also show that DNNs cannot extrapolate to test samples that are substantially different from the training examples. If the training data are sufficiently representative, however, internal features learned by the DNN are relatively stable with respect to speaker differences, bandwidth differences, and environment distortion. This enables DNN-based recognizers to perform as well or better than state-of-the-art systems based on GMMs or shallow networks without the need for explicit model adaptation or feature normalization.