Extracting information from collections of musical performances using Dynamic Time Warping

Extracting information from collections of musical performances using Dynamic Time Warping

Peperkamp, J.B.

Liem, C.C.S. (mentor)
Hildebrandt, K.A. (mentor)

Electrical Engineering, Mathematics and Computer Science

Intelligent Systems

Media and Knowledge Engineering

2016-08-30

When performing a piece of music, there are certain ways in which performers turn the score into their own musical interpretation of it. The difference between what is written and what is played can be substantial, and can vary between performers and even individual performances. In this thesis we examine what information can be gained from analyzing the differences that can be found when inspecting a collection of performances of a piece by various performers. We develop several techniques for analyzing the alignment yielded by Dynamic Time Warping and also explore the possibility of employing techniques from computer graphics to solve the task of finding an alignment. We test the techniques on synthetic data first, where applicable. We also test the techniques on a real dataset, a collection of performances of Chopin's mazurkas. Finally, we give some indication of how the techniques may most effectively be put to use based on the results of the tests.

Dynamic Time Warping
musical performance analysis
musical structure analysis

http://resolver.tudelft.nl/uuid:ef93503c-ce97-41d4-baf3-c7e7c08ed355

Student theses

master thesis

(c) 2016 Peperkamp, J.B.