Large scale evaluation of importance maps in automatic speech recognition
This work addresses the evaluation of interpretability methods for speech recognition systems, though it appears incremental as it builds on their previous bubble noise technique.
The authors tackled the problem of evaluating importance maps in automatic speech recognition by proposing a structured saliency benchmark (SSBM) metric, and found that their bubble noise approach outperformed a baseline method on 100 sentences from the AMI corpus.
In this paper, we propose a metric that we call the structured saliency benchmark (SSBM) to evaluate importance maps computed for automatic speech recognizers on individual utterances. These maps indicate time-frequency points of the utterance that are most important for correct recognition of a target word. Our evaluation technique is not only suitable for standard classification tasks, but is also appropriate for structured prediction tasks like sequence-to-sequence models. Additionally, we use this approach to perform a large scale comparison of the importance maps created by our previously introduced technique using "bubble noise" to identify important points through correlation with a baseline approach based on smoothed speech energy and forced alignment. Our results show that the bubble analysis approach is better at identifying important speech regions than this baseline on 100 sentences from the AMI corpus.