Imitating dysphonic voice: a suitable technique to create speech stimuli for spoken language processing tasks?

The aim of this study was to assess the suitability of imitated dysphonic voice samples for their application in listening tasks investigating the impact of speakers' voice quality on spoken language processing.

A female voice expert recorded speech samples (sustained vowels and connected speech) in her normal voice and while imitating a dysphonic voice. Voice characteristics, authenticity, and consistency of the two voice qualities were evaluated by means of acoustic measurements (Acoustic Voice Quality Index [AVQI], jitter, shimmer, harmonics-to-noise ratio [HNR]) and perceptual evaluation (GRBAS scale, consistency, and authenticity rated by five speech-language pathologists).

Based on acoustic and perceptual assessments, the degree of voice impairment for the imitated dysphonic voice was found to be moderate to severe. Roughness and asthenia were the predominant perceptual features. The perceptual rating indicated a high consistency and acceptable authenticity of the imitated dysphonic voice.

Results suggest that an imitation of dysphonic voice quality may resemble the voice characteristics typically found in dysphonic patients.

The voice samples validated here shall be applied in future listening tasks and may promote our understanding of how dysphonic speech is processed.