Mood Cloud : A Real-Time Music Mood Visualization Tool

We present Mood Cloud, an application of automatic music mood prediction from audio content. While playing a song, we visualize in real-time the prediction probabilities of five mood categories : “happy”, “sad”, “aggressive”, “relax” and “party”. Each mood is represented by a colored bar graph with text, dynamically resized according to the mood probability. The resulting application is a dynamic visualization of the mood predictions, demonstrating the performance of current state of the art techniques in Music Information Retrieval and especially in automatic mood classification.