The similarity measure between musical contours is a challenge since the algorithms available in the literature are not completely efficient. The ACMEMB algorithm (Marvin 1987) is incapable of dealing with large contours of 15 or more elements, and the algorithm FOURIER (Schmuckler 1999, 2010), unable to handle small contours with less than 5 elements. The main goal of this project is to propose new similarity algorithms to handle contours of all sizes. This project was financially supported by CNPq.
The main results of this project are the algorithms (awaiting publication) and the systems developed:
- Contour Metrics. Web based System for Calculating Contours and Comparing Algorithms
- Pocket Contour. Contour calculation application for Android.
In this project I performed the following tasks:
- Literature review of contour, musical similarity, and algorithm analysis
- Review of similarity algorithms of other areas, (such as string matching, geometry, etc.)
- Algorithms Benchmarking system implementation
- Development of new algorithms
- Analysis of melodies with the implemented algorithms
- Contour calculator implementation on Android