Scholarlink

In the context of an aging society, the research field of Chinese speech recognition technology for the elderly faces a significant challenge: data scarcity. To address this, this paper constructs the Elderly Chinese Speech Corpus (Elderly_CSC), which collects 11122 Chinese audio data items from 120 elderly individuals aged between 65 and 85. Firstly, the constructed elderly Chinese speech data were preprocessed through filtering and de-weighting. Then, the Elderly Chinese Speech Corpus was formed by manually annotating the data from four dimensions: the number of speakers, gender, age, and audio content. This targeted annotation ensured sample balance. Finally, speech recognition experiments were conducted on four representative mainstream speech models using this corpus. The experiments aim to verify the effectiveness of the corpus in optimizing the performance of speech recognition technology for the elderly.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Study on Corpus Construction and Recognition Methods for Elderly Chinese Speech Recognition

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Dongyang Wu,
  • Yukun Lv,
  • Honghui Yang

摘要

In the context of an aging society, the research field of Chinese speech recognition technology for the elderly faces a significant challenge: data scarcity. To address this, this paper constructs the Elderly Chinese Speech Corpus (Elderly_CSC), which collects 11122 Chinese audio data items from 120 elderly individuals aged between 65 and 85. Firstly, the constructed elderly Chinese speech data were preprocessed through filtering and de-weighting. Then, the Elderly Chinese Speech Corpus was formed by manually annotating the data from four dimensions: the number of speakers, gender, age, and audio content. This targeted annotation ensured sample balance. Finally, speech recognition experiments were conducted on four representative mainstream speech models using this corpus. The experiments aim to verify the effectiveness of the corpus in optimizing the performance of speech recognition technology for the elderly.