The SSPNet Speaker Personality Corpus includes 640 speech clips (10 seconds each) for a total of 322 subjects. Each clip was assessed by 11 raters in terms of the Big Five Personality traits (the assessments were performed using the BFI-10 questionnaire), namely Extraversion, Agreeableness, Conscientiousness, Neuroticism, Openness. The corpus includes not only the speech clips, but also the raw personality questionnaires, the overall personality scores and metadata associated to each clip: speaker gender, speaker status (journalist or non-journalist) and speaker ID.
The SSPNet Speaker Personality Corpus was used in the Speaker Trait Challenge held at Interspeech 2012.
The corpus is freely available, but users are kindly requested to cite the following paper in case they use the data:
G.Mohammadi and A.Vinciarelli, “Automatic Personality Perception: Prediction of Trait Attribution Based on Prosodic Features“, IEEE Transactions on Affective Computing, Vol. 3, no. 3, pp. 273-284, 2012.
- year: 2013
- url: http://www.dcs.gla.ac.uk/vincia/?p=307
- main_author: Gelareh Mohammadi and Alessandro Vinciarelli
- subjects: 322
- recordings: 640
- duration: 10 seconds
- naturality: natural
- media: Audio