| Sumario: | Three commercially available computer programs for voice analysis were used to compute measures of fundamental frequency (F0) from samples of the second sentence of the ?Rainbow Passage? in groups of normal speakers (adult males, adult females, and children). Results indicated that interprogram correlations for mean F0 estimation were strong (r?s > .90) and all of the programs produced mean F0 estimates within 5% of each other. However, based on the present data analyses, differences in the acoustic estimation of the voices of adult males suggest that care must be taken when analyzing and interpreting obtained fundamental frequency data. Significant differences in estimating F0 standard deviations were also observed when analyzing adult female voices. Correlations of equivalence among the programs for measures of F0 standard deviation were weaker in nature than those observed for estimates of mean speaking F0 and were nonsignificant for the analysis of adult male voices. Results indicated that measures associated with speaking F0 variability and range may be particularly influenced by gross F0 extraction errors. Details of computer program algorithms, description of possible sources of gross F0 extraction errors, and suggestions for recognizing and revising gross errors in F0 extraction are provided.
|
|---|