Comparative Intelligibility of Five Synthesized Voices

Stacie L. Rupprecht, David R. Beukelman, Holli Vrtiska

Research output: Contribution to journalArticlepeer-review

10 Scopus citations

Abstract

The purpose of this investigation was to compare the intelligibility of words within sentences produced by five synthesized voices: DECTalkTM Paul, DECTalk Betty, MacinTalk, MacinTalk ProTM Male, and MacinTalk ProTM Female. Thirty adult participants, ranging in age from 20 to 40 years, listened to the five synthesized voices and transcribed the last word in each of the Speech Perception in Noise (SPIN) Test sentences (50 sentences for each synthesizer). The data were analyzed using an analysis of variance with repeated measures and a post-hoc test (Greenhouse Geisser) to determine differences among the voices. The MacinTalk voice was significantly less intelligible than the other four voices. The mean intelligibility scores of the DECTalk Paul, DECTalk Betty, MacinTalk Pro Male, and MacinTalk Pro Female voices were not significantly different from each other. Results suggest the need for further research, especially with the MacinTalk Pro voices.

Original languageEnglish (US)
Pages (from-to)244-248
Number of pages5
JournalAugmentative and Alternative Communication
Volume11
Issue number4
DOIs
StatePublished - Jan 1995

Keywords

  • communication
  • speech
  • speech synthesis

ASJC Scopus subject areas

  • Rehabilitation
  • Speech and Hearing

Fingerprint

Dive into the research topics of 'Comparative Intelligibility of Five Synthesized Voices'. Together they form a unique fingerprint.

Cite this