Exploring Relationships between Audio Features and Emotion in Music

In this paper, we present an analysis of theassociations between emotion categories and audio features automaticallyextracted from raw audio data. This work is based on 110 excerpts from film soundtracksevaluated by 116 listeners. This data is annotated with 5 basic emotions (fear,anger, happiness, sadness, tenderness) on a 7 points scale for each emotion.Exploiting state-of-the-art Music Information Retrieval (MIR) techniques, weextract audio features of different kind: timbral, rhythmic and tonal. Amongothers we also compute estimations of dissonance, mode, onset rate andloudness. We study statistical relations between audio descriptors and emotioncategories confirming results from psychological studies. We also usemachine-learning techniques to model the emotion ratings. We create regressionmodels using the Support Vector Regression algorithm that can estimate theratings with a correlation of 0.65 in average.