The corpus, named AlloSat, is composed of real-life call center conversations in French and is continuously annotated in frustration and satisfaction. This corpus has been set up to develop new systems able to model the continuous aspect of semantic and paralinguistic information at the conversation level.
The AlloSat corpus was collected in association with Allo-Media, a company specialized in the automatic analysis of phone conversations from call-centers.
- 303 audio files (wav)
- 303 transcripts partially checked by humans in STM format
- Continue annotations on the satisfaction/frustration axis by 3 annotators
- Discrete annotations on satisfaction/frustration axis and valence by 3 annotators.