This study investigates listening efforts in speech communication for deaf and hard-of-hearing (DHH) under multispeaker environments. When DHH individuals participate in speech-oriented group communication, other speakers disrupt their target speech. These interferences make it difficult for those with hearing aids to listen to the target speech. While speech enhancement can extract the target speech, most previous works have only focused on speech quality rather than the listening effort. Therefore, we contribute to developing speech enhancement technology for the DHH people by systematically evaluating the listening efforts. The subjective evaluation reveals that DHH participants acquire the extracted target speech with less listening effort than the multispeaker mixed speech under a low signal-to-noise ratio (SNR) condition. When analyzing correlations with objective evaluation measures for speech quality, the correlation deviated considerably among the participants, ranging from those with no correlation to a strong correlation.

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

Empirical Study on Listening Effort of Deaf and Hard-of-Hearing for Enhanced Speech in Multispeaker Environments

  • Akio Kobayashi,
  • Shota Fujie,
  • Keiichi Yasu

摘要

This study investigates listening efforts in speech communication for deaf and hard-of-hearing (DHH) under multispeaker environments. When DHH individuals participate in speech-oriented group communication, other speakers disrupt their target speech. These interferences make it difficult for those with hearing aids to listen to the target speech. While speech enhancement can extract the target speech, most previous works have only focused on speech quality rather than the listening effort. Therefore, we contribute to developing speech enhancement technology for the DHH people by systematically evaluating the listening efforts. The subjective evaluation reveals that DHH participants acquire the extracted target speech with less listening effort than the multispeaker mixed speech under a low signal-to-noise ratio (SNR) condition. When analyzing correlations with objective evaluation measures for speech quality, the correlation deviated considerably among the participants, ranging from those with no correlation to a strong correlation.