ISCA Archive Interspeech 2021
ISCA Archive Interspeech 2021

Semantic Data Augmentation for End-to-End Mandarin Speech Recognition

Jianwei Sun, Zhiyuan Tang, Hengxin Yin, Wei Wang, Xi Zhao, Shuaijiang Zhao, Xiaoning Lei, Wei Zou, Xiangang Li

End-to-end models have gradually become the preferred option for automatic speech recognition (ASR) applications. During the training of end-to-end ASR, data augmentation is a quite effective technique for regularizing the neural networks. This paper proposes a novel data augmentation technique based on semantic transposition of the transcriptions via syntax rules for end-to-end Mandarin ASR. Specifically, we first segment the transcriptions based on part-of-speech tags. Then transposition strategies, such as placing the object in front of the subject or swapping the subject and the object, are applied on the segmented sentences. Finally, the acoustic features corresponding to the transposed transcription are reassembled based on the audio-to-text forced-alignment produced by a pre-trained ASR system. The combination of original data and augmented one is used for training a new ASR system. The experiments are conducted on the Transformer[2] and Conformer[3] based ASR. The results show that the proposed method can give consistent performance gain to the system. Augmentation related issues, such as comparison of different strategies and ratios for data combination are also investigated.


doi: 10.21437/Interspeech.2021-1162

Cite as: Sun, J., Tang, Z., Yin, H., Wang, W., Zhao, X., Zhao, S., Lei, X., Zou, W., Li, X. (2021) Semantic Data Augmentation for End-to-End Mandarin Speech Recognition. Proc. Interspeech 2021, 1269-1273, doi: 10.21437/Interspeech.2021-1162

@inproceedings{sun21b_interspeech,
  author={Jianwei Sun and Zhiyuan Tang and Hengxin Yin and Wei Wang and Xi Zhao and Shuaijiang Zhao and Xiaoning Lei and Wei Zou and Xiangang Li},
  title={{Semantic Data Augmentation for End-to-End Mandarin Speech Recognition}},
  year=2021,
  booktitle={Proc. Interspeech 2021},
  pages={1269--1273},
  doi={10.21437/Interspeech.2021-1162}
}