An Investigation on Applying Acoustic Feature Conversion to ASR of Adult and Child Speech

TL;DR

This study explores various acoustic feature conversion methods to improve child speech recognition by reducing domain mismatch with adult speech, highlighting the effectiveness of F0 normalization and disentanglement-based auto-encoder approaches.

Contribution

It compares different adult-to-child acoustic feature conversion techniques, identifying effective methods like F0 normalization and DAE, to enhance ASR performance on child speech.

Findings

F0 normalization yields the best ASR performance improvement.

Statistic matching is ineffective for domain adaptation in this context.

Converted F0 distribution correlates with conversion quality.

Abstract

The performance of child speech recognition is generally less satisfactory compared to adult speech due to limited amount of training data. Significant performance degradation is expected when applying an automatic speech recognition (ASR) system trained on adult speech to child speech directly, as a result of domain mismatch. The present study is focused on adult-to-child acoustic feature conversion to alleviate this mismatch. Different acoustic feature conversion approaches, including deep neural network based and signal processing based, are investigated and compared under a fair experimental setting, in which converted acoustic features from the same amount of labeled adult speech are used to train the ASR models from scratch. Experimental results reveal that not all of the conversion methods lead to ASR performance gain. Specifically, as a classic unsupervised domain adaptation method, the statistic matching does not show an effectiveness. A disentanglement-based auto-encoder (DAE) conversion framework is found to be useful and the approach of F0 normalization achieves the best performance. It is noted that the F0 distribution of converted features is an important attribute to reflect the conversion quality, while utilizing an adult-child deep classification model to make judgment is shown to be inappropriate.