Classification

Classification of musical genres gives a useful measure of similarity and is often the most useful descriptor of a musical piece. Previous techniques to use hidden Markov models (HMMs) for automatic genre classification have used a single HMM to model an entire song or genre. This paper provides a framework to give finer segmentation of HMMs through acoustic segment modeling. Modeling each of these acoustic segments with an HMM builds a timbral dictionary in the same fashion that one would create a phonetic dictionary for speech. A symbolic transcription is created by finding the most likely sequence of symbols. These transcriptions then serve as inputs into an efficient text classifier utilized to provide a solution to the genre classification problem. This paper demonstrates that language-ignorant approaches provide results that are consistent with the current state-of-the-art for the genre classification problem. However, the finer segmentation potentially allows for “musical language”-based syntactic rules to enhance performance.

This paper presents first steps towards a simple, robust computational model of automatic melody identification. Based on results from music psychology that indicate a relationship between melodic complexity and a listener’s attention, we postulate a relationship between musical complexity and the probability of a musical line to be perceived as the melody. We introduce a simple measure of melodic complexity, present an algorithm for predicting the most likely melody note at any point in a piece, and show experimentally that this simple approach works surprisingly well in rather complex music. 1

Abstract. Retrieving music from large digital databases is a demanding computational task. The cost for indexing and searching depends not only on the computational effort of measuring musical similarity, but also heavily on the number and sizes of files in the database. One way to speed up music retrieval is to reduce the search space by removing redundant and uninteresting material in the database. We propose a simple measure of ‘interestingness ’ based on music complexity, and present a reduction algorithm for MIDI files based on this measure. It is evaluated by comparing reduction ratios and the correctness of retrieval results for a query by humming task before and after applying the reduction. 1

In this paper we present an analysis of the suitability of four different feature sets which are currently employed to represent music signals in the context of the automatic music genre classification. To such an aim, feature selection is carried out through genetic algorithms, and it is applied to multiple feature vectors generated from different segments of the music signal. The feature sets used in this paper, which encompass time-domain and frequency-domain characteristics of the music signal, comprise: short-time Fourier transform, Mel frequency cepstral coefficient, beat-related features, pitch-related features, inter-onset interval histogram coefficients, rhythm histograms and statistical spectrum descriptors. The classification is based on the use of multiple feature vectors and an ensemble approach, according to time and space decomposition strategies. Feature vectors are extracted from music segments from the beginning, middle and end parts of the music signal (time-decomposition). Despite music genre classification being a multi-class problem, we accomplish the task using a combination of binary classifiers, whose results are merged to produce the final music genre label (space decomposition). Experiments were carried out on two databases: the Latin Music Database, which contains 3,227

Abstract. In this paper, we will examine the importance of music complexity as a factor for melody recognition in multi-voiced popular music. The assumption is that the melody (or lead instrument) will contain the largest amount of information – that it will be the least redundant voice. Measures of melodic complexity calculated from pitch and timing information are proposed. We test the different complexity measures and different prediction strategies, and evaluate them on the task of predicting which track of a MIDI file contains the main melody. Filtering out melody tracks can be useful when searching large databases for similar songs. 108 melody track annotated pop songs were included in the experiment. 1

Berkeley, California. During a decade of conferences, ISM2008 has established itself as an international renowned forum for researchers and practitioners to exchange ideas, connect with colleagues, and advance the state of the art and practice of multimedia computing, as well as to identify emerging research topics and to define the future of this cross-disciplinary field. This year, the conference counted 155 paper submissions for the main conference from 36 countries. Every submission was reviewed by at least three experts from the technical program committee. Due to the large amount of top-quality submissions, only 37 full-length papers were accepted for publication. This represents an acceptance rate of 24%. In addition, these proceedings feature high-quality short papers as well as articles from five workshops and the demo session. All of these papers provide novel ideas, new results, and state-of-the-art techniques in the field. We are honored to have several of the world’s leading experts in the field to join us as distinguished speakers. They are: Ruzena Bajcsy (University of California, Berkeley), Alex (Sandy) Pentland (MIT Media Lab), and Herbert Freeman (Rutgers, The State University of New Jersey). Altogether, we are proud to be able to present you a rich program that contains a variety of top-notch research. Of course, the technical program of ISM2008 would not have been possible without the exceptional

Modelling image complexity by independent

large music collections

We present an approach for cover version identification which is based on combining differentdiscretized features derived from the chromagram vectors extracted from the audio data. For measuring similarity between features, we use a parameter-free quasi-universal similarity metric which utilizes data compression. Evaluation proves that combined feature distances increase the accuracy in cover version identification. 1.