Voice Morphing
Verena Gabriele Skuk, Hideki Kawahara · 2018
Voice morphing is a framework to generate a new sound which has the mixed attribute of given voice examples. It provides a flexible tool for investigating perceptual attributes in voice communication, especially for quantifying paralinguistic and extralinguistic cues. Recent advances in parametric representation of speech sounds have made the morphing-based approach a practical alternative or complementary strategy to procedures using speech-production models. Stimulus continuum spanning between two typical voice recordings having different perceptual attributes provides an external reference to quantify subjective responses. Generalized morphing, which enables extrapolation as well as interpolation of arbitrarily many numbers of voices, provides a unique strategy for investigating speaker identity. Its ability to selectively manipulate any combination of fundamental parameters facilitates the development of technology to intervene and mediate paralinguistic and extralinguistic communication channels.