End-to-End Spelling Correction Conditioned on Acoustic Feature for Code-switching Speech Recognition

	End-to-End Spelling Correction Conditioned on Acoustic Feature for Code-switching Speech Recognition
	Shuai Zhang1,2
	2021-08-30
会议名称	INTERSPEECH 2021
会议日期	2021-8-30
会议地点	Brno, Czechia
摘要	In this work, we propose a new end-to-end (E2E) spelling cor- rection method for post-processing of code-switching automatic speech recognition (ASR). Existing E2E spelling correction models take the hypotheses of ASR as inputs and annotated text as the targets. Due to the powerful modeling capabilities of the E2E model, the training of the correction system is extremely prone to over-fitting. It usually requires sufficient data diver- sity for reliable training. Therefore, it is difficult to apply the E2E correction models to the code-switching ASR task because of the data shortage. In this paper, we introduce the acoustic features into the spelling correction model. Our method can al- leviate the problem of over-fitting and has better performance. Meanwhile, because the acoustic features are encode-free, our proposed model can be applied to the ASR model without sig- nificantly increasing the computational cost. The experimental results on ASRU 2019 Mandarin-English Code-switching Chal- lenge data set show that the proposed method achieves 11.14% relative error rate reduction compared with baseline
文献类型	会议论文
条目标识符	http://ir.ia.ac.cn/handle/173211/48819
专题	多模态人工智能系统全国重点实验室_智能交互
作者单位	1.School of Artificial Intelligence, University of Chinese Academy of Sciences, China 2.NLPR, Institute of Automation, Chinese Academy of Sciences, China
第一作者单位	模式识别国家重点实验室
推荐引用方式 GB/T 7714	Shuai Zhang. End-to-End Spelling Correction Conditioned on Acoustic Feature for Code-switching Speech Recognition[C],2021.

条目包含的文件		下载所有文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
zhang21d_interspeech（327KB）	会议论文		开放获取	CC BY-NC-SA	浏览下载