Modeling of Breathy Voice Quality Using Pitch-strength Estimates.
David A Eddins, Supraja Anand, Arturo Camacho and 1 others
PMID 26775221WHAT IT FOUND
A computer estimate of how strongly a voice sounds pitched predicted how breathy listeners judged voice samples.
It matched natural voices well; for synthetic voices it often underpredicted breathiness.
Key findings
01Pitch-strength estimates tracked listener breathiness judgments, with correlations from 0.82 to 0.97 in absolute value.
02A model built from 15 natural voices predicted breathiness ratings for another 15 natural voices, accounting for 88% of the spread in ratings.
03For the same perceived breathiness, pitch-strength estimates were higher for synthetic voices than for natural voices.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study tested laboratory voice samples and listener judgments, not patient outcomes or treatment decisions. The listeners were students with normal hearing, not necessarily experienced voice clinicians. The natural stimuli came from a database and may not represent the voices in a particular clinic. The model was built on natural voices and generalized less well to synthetic voices. For the same perceived breathiness, synthetic voices had higher pitch-strength estimates than natural voices. The Discussion notes the model underpredicted perceived breathiness of synthetic stimuli by up to 7 dB over much of the range.
The easy way to misread this
Do not conclude that this model is a validated clinical tool for assessing patient voice. The study tested laboratory voice samples and listener judgments, not patient care outcomes.