How do we alter our utterances when a speech recognition system fails?

DOI

Bibliographic Information

Other Title
  • 音声認識装置が音声を正しく認識しない事態における人間の発声の変化

Abstract

In speech recognition systems speakers may consider their speech as an input to recognition devices. However, from the view of human-machine interaction, human vocal behaviors may also be an output affected by the performance of the recognition system.<br>Subjects were requested to speak out color names of rectangles painted with either blue, green, red, white, or yellow on a CRT. A part of the vocal responses of the subjects in the experimental condition were intentionally misrecognized and rejected after a series of successful recognition. The subjects' vocal responses became significantly louder and longer after misrecognition compared with the voice emitted just before the rejection. Female subjects altered their speech style more promptly than males did. Male subjects changed their vocal fundamental frequency. The subjects in the control group, where it is so designed that speech was always successfully recognized, did not show specific loudness changes.

Journal

Keywords

Details 詳細情報について

  • CRID
    1390282679508358784
  • NII Article ID
    130003871135
  • DOI
    10.5100/jje.31.287
  • ISSN
    18842844
    05494974
  • Text Lang
    ja
  • Data Source
    • JaLC
    • Crossref
    • CiNii Articles
  • Abstract License Flag
    Disallowed

Report a problem

Back to top