Measurement of vocal breathiness perception with a matching task for sustained phonation and running speech

Abstract

Laboratory measurements of the perception of vocal breathiness typically employ recordings of sustained vowel phonations. Sustained vowels are thought to provide a good representation of the underlying vocal disorder and are relatively easy to produce, analyze, and synthesize. Clinical assessments of voice quality are sometimes based on the production of speech utterances that are more ecologically valid than sustained vowel phonation and that capture the dynamics of laryngeal function, breath support, and a variety of complex cognitive and neuromuscular challenges typically involved in natural speech. As a result, voice quality associated with running speech may correspond more closely with perceived handicap and may represent more relevant treatment targets than sustained vowels. In a listening experiment with naive listeners, breathy voice quality was evaluated for exemplars of sustained /a/ production as well as read speech from a selection of talkers that vary widely in terms of their vocal breathiness. A single-variable matching task was used to index perceived breathiness for both stimulus types. Listener reliability was high, and matching thresholds for the sustained phonation and speech stimuli were highly correlated, demonstrating the efficacy of the matching task for measuring vocal breathiness from speech utterances.