Voice cue perception and voice gender categorisation via a humanoid robot interface

Laura Rachman,Luke Meyer,Gloria Araiza-Illan,Etienne Gaudrain,Deniz Baskent
DOI: https://doi.org/10.1121/10.0019037
2023-03-01
The Journal of the Acoustical Society of America
Abstract:Two voice cues, fundamental frequency (F0) and vocal-tract length (VTL), are important for characterising voice gender. Due to their often repetitive nature, psychophysical tests evaluating perception of these vocal cues can at times cause individuals to lose engagement during the test. To help with this, we propose the use of an interactive humanoid NAO robot as an alternative to the conventionally used laptop interface. As a first step, we compare the performance of two psychophysical tests between the robot and laptop interfaces. Experiment I measured F0 and VTL cue perception through an adaptive test in just noticeable differences (JNDs). Experiment II measured voice gender categorisation using F0 and VTL manipulated stimuli. The robot implementation made use of the tactile sensors on the hands and head of the robot for response logging. Performance accuracy between the computer and robot interfaces was functionally similar, confirming data reliability. Test duration comparison showed that both experiments were longer on the NAO robot than the laptop. Despite potential design limitations of the robot interface, both interfaces showed that the F0 and VTL JNDs are very small in normal-hearing listeners, and that normal-hearing listeners make effective use of F0 and VTL cues for voice gender categorisation.
acoustics,audiology & speech-language pathology
What problem does this paper attempt to address?