eduzhai > Applied Sciences > Engineering >

CUCHILD A Large-Scale Cantonese Corpus of Child Speech for Phonology and Articulation Assessment

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: This paper describes the design and development of CUCHILD, a large-scaleCantonese corpus of child speech. The corpus contains spoken words collectedfrom 1,986 child speakers aged from 3 to 6 years old. The speech materialsinclude 130 words of 1 to 4 syllables in length. The speakers cover bothtypically developing (TD) children and children with speech disorder. Theintended use of the corpus is to support scientific and clinical research, aswell as technology development related to child speech assessment. The designof the corpus, including selection of words, participants recruitment, dataacquisition process, and data pre-processing are described in detail. Theresults of acoustical analysis are presented to illustrate the properties ofchild speech. Potential applications of the corpus in automatic speechrecognition, phonological error detection and speaker diarization are alsodiscussed.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×