Towards Debugging Deep Neural Networks by Generating Speech Utterances
This work addresses the interpretability issue in speech classification DNNs, offering a method to understand model decisions, though it is incremental as it adapts an existing technique from image to speech domains.
The paper tackled the problem of debugging deep neural networks for speech classification by applying activation maximization to generate speech utterances that are classified into specific classes, and found that combining it with a natural speech prior enables generation of class examples, as validated by objective measurements and human evaluations.
Deep neural networks (DNN) are able to successfully process and classify speech utterances. However, understanding the reason behind a classification by DNN is difficult. One such debugging method used with image classification DNNs is activation maximization, which generates example-images that are classified as one of the classes. In this work, we evaluate applicability of this method to speech utterance classifiers as the means to understanding what DNN "listens to". We trained a classifier using the speech command corpus and then use activation maximization to pull samples from the trained model. Then we synthesize audio from features using WaveNet vocoder for subjective analysis. We measure the quality of generated samples by objective measurements and crowd-sourced human evaluations. Results show that when combined with the prior of natural speech, activation maximization can be used to generate examples of different classes. Based on these results, activation maximization can be used to start opening up the DNN black-box in speech tasks.