Scales used for the voice ratings, translations, and terms used in the study with French-speaking raters.

The GVPS was used for the ratings in both corpora, but the procedures involved in collecting the ratings differed slightly. For the MUC corpus, the perceived voice ratings were collected by a stimulus ranking procedure of the emotion portrayals, separately for each speaker. On a computer screen, raters arranged icons representing the audio stimuli (which they could listen to repeatedly) recursively on a continuum that was consequently recoded to a score ranging from 0 to 100. For the corpus, a more conventional rating procedure was involved, with raters using a visual analogue scale on the screen immediately after listening to each portrayal (later recoded to a score ranging from 0 to 100). All participants provided ratings for all voice scales sequentially and in random order (stimuli were also presented in random order for assessment within each scale).Scales used for the voice ratings, translations, and terms used in the study with French-speaking raters.