eduzhai > Applied Sciences > Engineering >

Aphasic Speech Recognition using a Mixture of Speech Intelligibility Experts

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: Robust speech recognition is a key prerequisite for semantic featureextraction in automatic aphasic speech analysis. However, standardone-size-fits-all automatic speech recognition models perform poorly whenapplied to aphasic speech. One reason for this is the wide range of speechintelligibility due to different levels of severity (i.e., higher severitylends itself to less intelligible speech). To address this, we propose a novelacoustic model based on a mixture of experts (MoE), which handles the varyingintelligibility stages present in aphasic speech by explicitly definingseverity-based experts. At test time, the contribution of each expert isdecided by estimating speech intelligibility with a speech intelligibilitydetector (SID). We show that our proposed approach significantly reduces phoneerror rates across all severity stages in aphasic speech compared to a baselineapproach that does not incorporate severity information into the modelingprocess.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×