ANGUS: Real-time manipulation of vocal roughness for emotional speech transformations
Vocal arousal, the non-linear acoustic features taken on by human and animal vocalizations when highly aroused, has an important communicative function because it signals aversive states such as fear, pain or distress. In this work, we present a computationally-efficient, real-time voice transformation algorithm, ANGUS, which uses amplitude modulation and time-domain filtering to simulate roughness, an important component of vocal arousal, in arbitrary voice recordings. In a series of 4 studies, we show that ANGUS allows parametric control over the spectral features of roughness like the presence of sub-harmonics and noise; that ANGUS increases the emotional negativity perceived by listeners, to a comparable level as a non-real-time analysis/resynthesis algorithm from the state-of-the-art; that listeners cannot distinguish transformed and non-transformed sounds above chance level; and that ANGUS has a similar emotional effect on animal vocalizations and musical instrument sounds than on human vocalizations. A real-time implementation of ANGUS is made available as open-source software, for use in experimental emotion reseach and affective computing.