Automatic Genre Detection for Intelligent Audio Tools

MixGenius works on automated musical production by real-time musical genre detection. High accuracy of genre detection for a variety of sub-genres is required for quality production and has yet to be achieved. The project involves researching available real-time high-accuracy musical genre detection methods and improving upon them, extracting each genre’s audio features, calibrating the algorithm with a database of songs of different genres, testing them with other songs of the same genres and repeating the process as needed to improve detection accuracy. The partner organization will benefit from insight into academia and access to state of the art genre detection research, new software modeled and created, superior results from music engine product gained by knowledge of genre, and implementation that is reusable and expandable for future development.

Faculty Supervisor:

Paul Charbonneau

Student:

Kim Thibault

Partner:

Mixgenius

Discipline:

Physics / Astronomy

Sector:

Information and communications technologies

University:

Université de Montréal