Vocalisation Challenge Corpus (VOC-C)

2022 ComParE Vocalisations Sub-Challenge

This repository contains the Vocalisation Challenge Corpus (VOC-C) for the 2022 Vocalisations Sub-Challenge of the Computational Paralinguistics ChallengE (ComParE).

The corpus features human non-speech vocalisations, such as laughter, cries, moans, or screams, expressing different emotions, varying from mild to extreme affective intensity. The emotion classes include achievement/triumph, anger, fear, pain, sexual pleasure, and positive surprise. The data from the female speakers have been made openly available (see publication below); the male speakers are so far unseen. The data are partitioned into the female vocalisations in train (6 speakers, 625 items) and development (5 speakers, 460 items), and the male vocalisations (2 speakers, 276 items) in test. 

Please find information on the published female speaker data in Holz, N., Larrouy-Maestri, P., & Poeppel, D. (2022). The variably intense vocalizations of affect and emotion (VIVAE) corpus prompts new perspective on nonspeech perception. Emotion, 22(1), 213–225. 

Additional information on male speaker data: Male speakers were undergraduate students at the Berklee College of Music in Boston. None of them underwent formal acting training or had professional acting experience. In difference to the female speakers, none of them reported stage experience as singers. Recording procedures were the same across all speakers; no technical validation was performed on the male speaker data.