Can dysphonia be mimicked authentically? An acoustic and perceptual evaluation of imitated dysphonic speech samples

The purpose of this study was to investigate whether a healthy speaker can authentically imitate dysphonia. Mimicked dysphonic speech samples were assessed acoustically and perceptually. Evaluation was based on recordings of sustained vowels and connected speech produced by a female voice therapist. The speaker first used her normal voice and then mimicked a dysphonic voice. Acoustic assessment was performed with Praat (AVQI, Jitter, Shimmer, and HNR). Perceptual assessment was conducted by five Speech Language Pathologists. After listening to the speaker’s normophonic and mimicked dysphonic samples, they performed GRBAS ratings as well as authenticity and consistency ratings using a four-point scale. Results showed a moderate to severe degree of voice impairment for the mimicked dysphonic voice, with an AVQI score of 6.8. The mimicked dysphonic voice was predominantly perceived as rough (Median = 3) and asthenic (Median = 3). Authenticity and consistency were rated as acceptable and high respectively. This indicates that the speaker’s imitation of dysphonic voice was realistic and that she successfully maintained the same voice quality throughout the recording. Regarding the applications, we argue that imitation of dysphonic voice performed by a voice expert may be a useful method for investigating the effects of dysphonic voice on listeners’ spoken language processing. Speech samples validated in this study shall be used for future listening experiments and may improve our understanding of how speakers’ voice quality affects the listening experience.