Cover Song Identification Using Compression-based Distance Measures

M.Sc. Teppo E. Ahonen will defend his doctoral thesis Cover Song Identification Using Compression-based Distance Measures on Friday the 1st of April 2016 at 12 o'clock in the University of Helsinki Exactum Building, Auditorium CK112 (Gustaf Hällströminkatu 2b).  His opponent is Academy Professor Petri Toiviainen (University of Jyväskylä) and custos Professor Esko Ukkonen (University of Helsinki). The defence will be held in Finnish.

Measuring similarity in music data is a problem with various potential applications. In recent years, the task known as cover song identification has gained widespread attention. In cover song identification, the purpose is to determine whether a piece of music is a different rendition of a previous version of the composition. The task is quite trivial for a human listener, but highly challenging for a computer.

 

This research approaches the problem from an information theoretic starting point. Assuming that cover versions share musical information with the original performance, we strive to measure the degree of this common information as the amount of computational resources needed to turn one version into another. Using a similarity measure known as normalized compression distance, we approximate the non-computable Kolmogorov complexity as the length of an object when compressed using a real-world data compression algorithm. If two pieces of music share musical information, we should be able to compress one using a model learned from the other.

In order to use compression-based similarity measuring, the meaningful musical information needs to be extracted from the raw audio signal data. The most commonly used representation for this task is known as chromagram: a sequence of real-valued vectors describing the temporal tonal content of the piece of music. Measuring the similarity between two chromagrams effectively with a data compression algorithm requires further processing to extract relevant features and find a more suitable discrete representation for them. Here, the challenge is to process the data without losing the distinguishing characteristics of the music.

In this research, we study the difficult nature of cover song identification and search for an effective compression-based system for the task. Harmonic and melodic features, different representations for them, commonly used data compression algorithms, and several other variables of the problem are addressed thoroughly. The research seeks to shed light on how different choices in the scheme attribute to the performance of the system. Additional attention is paid to combining different features, with several combination strategies studied. Extensive empirical evaluation of the identification system has been performed, using large sets of real-world music data.

Evaluations show that the compression-based similarity measuring performs relatively well but fails to achieve the accuracy of the existing solution that measures similarity by using common subsequences. The best compression-based results are obtained by a combination of distances based on two harmonic representations obtained from chromagrams using hidden Markov model chord estimation, and an octave-folded version of the extracted salient melody representation. The most distinct reason for the shortcoming of the compression performance is the scarce amount of data available for a single piece of music. This was partially overcome by internal data duplication. As a whole, the process is solid and provides a practical foundation for an information theoretic approach for cover song identification.

Availability of the dissertation

An electronic version of the doctoral dissertation is available on the e-thesis site of the University of Helsinki at http://urn.fi/URN:ISBN:978-951-51-2026-7.

Printed copies will be available on request from Teppo E. Ahonen: tel. 02941 51276 or teppo.ahonen@cs.helsinki.fi.

Created date

30.03.2016 - 14:53

Brain poetry

In the latest research result of the month section, we interview PhD student Jukka Toivanen about his recent work on brain poetry in the Discovery group led by professor Hannu Toivonen. How can humans and machines be creative together?

Kjell Lemström to be new Head of Studies

Since Jaakko Kurhila left the department to head the Open University, we had to find a new university lecturer to act as head of studies in short order. We received a total of 28 applications. Out of these, and after a preliminary qualification round, evaluations, interviews, and a department council hearing, Kjell Lemström (KL) was elected for the post. He started working as the department's Head of Studies on 2 March 2015, so the Head of the department (JP) conducted the following induction interview that very week.

This is by no means the first time Kjell has been employed by the department. He defended his thesis on ‘String Matching Techniques for Music Retrieval’ in 2000, and has held numerous teaching and research positions both before and after that, until he transferred to the Laurea University of Applied Sciences in 2011 (luckily, that was only temporary).

Head of Studies Jaakko Kurhila to head Open University

The Head of Studies at the department, University Lecturer Jaakko Kurhila, has been elected to the post of director of the Open University at the University of Helsinki. It was a tough race: all in all, 39 applicants sought the post, some of them through the Mercuri Urval headhunting process. After a consultant evaluation, interviews, and aptitude assessments, the preparatory committee for the post, the steering committee for the Open University, and the rector of the university came to a unanimous decision to select Jaakko, and the contract is already being drawn up.

Being selected from this prestigious group of applicants, and after such a thorough process, is indisputable proof of the qualifications of Jaakko and the high esteem the academic community has for him. The department extends its warmest congratulations to Jaakko for this career development and is proud of the success of its protégé.

Bridging the Gap Between Research and Standardization

In the fourth research result of the month, we report a joint work between the UH NODES group and the Cambridge NetOS group, lead by Prof. Sasu Tarkoma and Prof. Jon Crowcroft, respectively. Their work recently received the best paper award "Best of CCR" from ACM SIGCOMM.