This presentation focuses on the discovery of repeating patterns in musical data using the theory of mathematical morphology. We demonstrate how this theory, which is mainly applied to image processing, can be used to obtain fundamental mathematical results for musical pattern discovery. Using these results, we propose a method for discovering MTEC patterns, i.e. patterns to which it is not possible to add dots without decreasing their number of occurrences. This work provides insight into the importance of MTEC patterns in music data, which is a current problem in the music information retrieval community.
You can visit Paul's YouTube channel: Mathémusique.
For more details on the seminar, the SCRIME website.