English

Filter-based Discriminative Autoencoders for Children Speech Recognition

Computation and Language 2022-05-24 v2 Machine Learning Sound Audio and Speech Processing

Abstract

Children speech recognition is indispensable but challenging due to the diversity of children's speech. In this paper, we propose a filter-based discriminative autoencoder for acoustic modeling. To filter out the influence of various speaker types and pitches, auxiliary information of the speaker and pitch features is input into the encoder together with the acoustic features to generate phonetic embeddings. In the training phase, the decoder uses the auxiliary information and the phonetic embedding extracted by the encoder to reconstruct the input acoustic features. The autoencoder is trained by simultaneously minimizing the ASR loss and feature reconstruction error. The framework can make the phonetic embedding purer, resulting in more accurate senone (triphone-state) scores. Evaluated on the test set of the CMU Kids corpus, our system achieves a 7.8% relative WER reduction compared to the baseline system. In the domain adaptation experiment, our system also outperforms the baseline system on the British-accent PF-STAR task.

Keywords

Cite

@article{arxiv.2204.00164,
  title  = {Filter-based Discriminative Autoencoders for Children Speech Recognition},
  author = {Chiang-Lin Tai and Hung-Shin Lee and Yu Tsao and Hsin-Min Wang},
  journal= {arXiv preprint arXiv:2204.00164},
  year   = {2022}
}

Comments

Published in EUSIPCO 2022

R2 v1 2026-06-24T10:34:09.448Z