Interactive Biofeedback to Improve Voice Masculinity in Transmasculine Speakers
SPRING 2020 RESEARCH INCUBATION AWARDEES
What is the Challenge?
Acoustic parameters can target voice modification sessions to alter the perception of gender of transmasculine speakers meaningfully. Additionally, novel speech resynthesis techniques can be used to modulate speech parameters of transmasculine speakers before and following voice masculinization.
What is the Solution?
To address this problem, PI Cara Stepp, alongside her CO-PI, Lauren Tracy plan to create a set of voice samples with specific modulated speech characteristics. These characteristics can be used in an auditory-perceptual experiment amidst unfamiliar listeners to identify which speech parameters are most important for gender perception. The results of the auditory-perceptual experiment can be used to develop an online voice modification module that transmasculine speakers can use to complete targeted voice modification sessions in the comfort and privacy of their homes through instantaneous biofeedback.
What is the Process?
The team uses speech resynthesis techniques to create modulated transmasculine voice samples. Pre- and post-voice masculinization samples will be decomposed into quantitative features (i.e., fundamental frequency, formant locations, and pitch contour). Using the algorithmic implementation of STRAIGHT Speech Synthesis, the features of the pre-modification signal would then be changed to reflect the features of the post-modification signal. These would be re-synthesized into various combinations to create new acoustic signals. Additionally, an auditory-perceptual experiment acts to determine which acoustic features are most important to the perception of gender. Twenty listeners are recruited to listen to modulated speech samples and rate each speech sample in terms of gender perception. Using these results of the auditory-perceptual experiment, an interactive module can be developed that transmasculine speakers may use to monitor targeted acoustic parameters. With this module, clinicians would be able to set target values for each acoustic parameter, and transmasculine speakers would be able to receive real-time biofeedback as they complete voice modification sessions.