IEMOCAP

The IEMOCAP, or Interactive Emotional Dyadic Motion Capture database, is a widely used dataset in affective computing and speech emotion recognition research. It was created by researchers at the University of Southern California and comprises audio-visual recordings of 10 actors performing 10 different scripted scenarios and 10 improvisational scenarios. The goal of the dataset was to capture a range of emotional expressions in a dyadic (two-person) interaction context.

The recordings were made with high-quality motion capture technology, allowing for the capture of detailed facial and body movements. Each interaction was annotated by the actors themselves and by independent raters for several emotional categories, including happiness, sadness, anger, excitement, frustration, and neutral. The dataset also includes transcripts of the spoken dialogue.

IEMOCAP has been instrumental in advancing research in areas such as automatic emotion recognition from speech and facial expressions, multimodal emotion detection, and understanding human-human interaction. Its rich annotation and diverse nature make it a valuable resource for training and evaluating machine learning models designed to understand and generate emotional responses. The dataset is publicly available for research purposes, contributing to its widespread adoption in the scientific community.