Time: 2:00 - 3:00pm
Venue: Maths 103
Title: Music Structure Analysis: Paths like Blocks
Abstract: In music structure analysis the two principles of repetition and homogeneity are fundamental for partitioning a given audio recording into musically meaningful structural elements. When converting the audio recording into a suitable self-similarity matrix (SSM), repetitions typically lead to path structures, whereas homogeneous regions yield block structures. In previous research, handling both structural elements at the same time has turned out to be a challenging task.
In this talk, I will show that paths and blocks may like each other. After giving an illustrative introduction to various segmentation principles, I will present an idea for converting path structures into block structures. The main technical step consists in an eigenvalue decomposition of an SSM in combination with suitable clustering techniques. I will discuss the theoretical background by considering typical examples. Then, I demonstrate the effectiveness of the proposed conversion procedure by showing that algorithms previously designed for homogeneity-based structure analysis can now be applied for repetition-based structure analysis.
Bio: Harald Grohganz recieved the M.Sc. (Diplom) degree in mathematics from Bonn University in 2010. Since October 2010, he is pursuing a Ph.D. in the Multimedia Information Retrieval and Signal Processing group at Bonn University under the joint supervision of Michael Clausen (Bonn University) and Meinard Müller (International Audio Laboratories Erlangen). Working in the field of music signal processing and music information retrieval, his research interests cover music structure analysis, music segmentation, key and timbre estimation, rhythm transcription, and user interface design.