The effect of speaking context on spectral- and cepstral-based acoustic features of normal voice

Soren Lowell, Jennifer A. Hylkema

Research output: Contribution to journalArticle

5 Scopus citations

Abstract

The effect of speaking context on four cepstral- and spectral-based acoustic measures was investigated in 20 participants with normal voice. Speakers produced three different continuous speaking tasks that varied in duration and phonemic content. Cepstral and spectral measures that can be validly derived from continuous speech were computed across the three speaking contexts. Cepstral peak prominence (CPP), low/high spectral ratio, and the standard deviation (SD) of the low/high spectral ratio did not significantly differ across speaking contexts, and correlations for the first two measures were strong among the three speaking tasks. The SD of the CPP showed significant task differences, and relationships between the speaking contexts were generally moderate. These findings suggest that in speakers with normal voice, the differing phonemic content across several frequently used speaking stimuli minimally impacted group means for three clinically relevant cepstral- and spectral-based acoustic measures.

Original languageEnglish (US)
Pages (from-to)1-11
Number of pages11
JournalClinical Linguistics and Phonetics
Volume30
Issue number1
DOIs
StatePublished - Jan 2 2016

    Fingerprint

Keywords

  • Acoustic
  • cepstral
  • context
  • continuous
  • phonemic
  • spectral
  • speech
  • voice

ASJC Scopus subject areas

  • Speech and Hearing
  • Linguistics and Language
  • Language and Linguistics

Cite this