Abstract. We report on experiments done in an attempt to improve the performance of a music similarity measure which we introduced earlier. The technique aims at comparing music titles on the basis of their global “timbre”, which has many applications in the field of Music Information Retrieval. Such measures of timbre similarity have seen a growing interest lately, and every contribution (including ours) is yet another instantiation of the same basic pattern recognition architecture, only with different algorithm variants and parameters. Most give encouraging results with a little effort, and imply that near-perfect results would just extrapolate by fine-tuning the algorithms ’ parameters. However, such systematic testing over large, interdependent parameter spaces is both difficult and costly, as it requires to work on a whole general meta-database architecture. This paper contributes in two ways to the current state of the art. We report on extensive tests over very many parameters and algorithmic variants, either already envisioned in the literature or not. This leads to an improvement over existing algorithms of about 15 % R-precision. But most importantly, we describe many variants that surprisingly do not lead to any substancial improvement. Moreover, our simulations suggest the existence of a “glass ceiling ” at R-precision about 65 % which cannot probably be overcome by pursuing such variations on the same theme.

Manifold approaches exist for organization of music by genre and/or style. In this paper we propose the use of text categorization techniques to classify artists present on the Internet. In particular, we retrieve and analyze webpages ranked by search engines to describe artists in terms of word occurrences on related pages. To classify artists we primarily use support vector machines. We present 3 experiments in which we address the following issues. First, we study the performance of our approach compared to previous work. Second, we investigate how daily fluctuations in the Internet affect our approach. Third, on a set of 224 artists from 14 genres we study (a) how many artists are necessary to define the concept of a genre, (b) which search engines perform best, (c) how to formulate search queries best, (d) which overall performance we can expect for classification, and finally (e) how our approach is suited as a similarity measure for artists.

Our approach to generate recommendations for similar artists follows a recent tradition of authors tackling the problem not with content-based audio analysis. Following this novel procedure we rely on the acquisition, filtering and condensing of unstructured text-based information that can be found in the web. The beauty of this approach lies in the possibility to access so-called cultural metadata that is the agglomeration of several independent-originally subjective- perspectives about music. 1.

Large music collections require new ways to let users interact with their music. The concept of finding ‘similar’ songs, albums, or artists provides handles to users for easy navigation and instant retrieval. This paper presents the realization and user evaluation of a music retrieval music that sorts songs on the basis of similarity to a given seed song. Similarity is based on a userweighted combination of timbre, genre, tempo, year, and mood. A conclusive user evaluation assessed the usability of the system in comparison to two control systems in which the user control of defining the similarity measure was diminished.

We present the query-by-description (QBD) component of "Kandem," a time-aware music retrieval system. The QBD system we describe learns a relation between descriptive text concerning a musical artist and their actual acoustic output, making such queries as "Play me something loud with an electronic beat" possible by merely analyzing the audio content of a database. We show a novel machine learning technique based on Regularized Least-Squares Classification (RLSC) that can quickly and efficiently learn the non-linear relation between descriptive language and audio features by treating the problem as a large number of possible output classes linked to the same set of input features. We show how the RLSC training can easily eliminate irrelevant labels. I.

Abstract. As digital music collections grow, so does the need to organizing them automatically. In this paper we present an approach to hierarchically organize music collections at the artist level. Artists are grouped according to similarity which is computed using a web search engine and standard text retrieval techniques. The groups are described by words found on the webpages using term selection techniques and domain knowledge. We compare different term selection techniques, present a simple demonstration, and discuss our findings. 1

Recently there has been an increasing amount of work in the area of automatic genre classification of music in audio format. Such systems can be used as a way to evaluate features describing musical content as well as a way to structure large collections of music. However the evaluation and comparison of genre classification systems is hindered by the subjective perception of genre definitions by users. In this work we describe a set of experiments in automatic musical genre classification. An important contribution of this work is the comparison of the automatic results with human genre classification on the same dataset. The results show that, although there is significant room for improvement, genre classification is inherently subjective and therefore perfect results can not be expected from either automatic algorithms or human annotation. The experiments also show that the use of features derived from an auditory model have similar performance with features based on Mel-Frequency Cepstral Coefficients (MFCC). 1.

The ever-growing amount of available music induces an increasing demand for Music Information Retrieval (MIR) applications such as music recommendation applications or automatic classification algorithms. When audio-based, a crucial part of such systems are the audio feature extraction routines. In this paper, we evaluate how well a variety of combinations of feature extraction and machine learning algorithms are suited to classify music into perceptual categories. The examined categorizations are perceived tempo, mood (happy / neutral /sad), emotion (soft / neutral / aggressive), complexity, and vocal content. The aim is to contribute to the investigation which aspects of music are not captured by the common audio descriptors; from our experiments we can conclude that most of the examined categorizations are not captured well. This indicates that more research is needed on alternative (possibly extra-musical) sources of information for useful music classification. 1.

Abstract. With the popularity of multimedia applications, a large amount of music data has been accumulated on the Internet. Automatic classification of music data becomes a critical technique for providing an efficient and effective retrieval of music data. In this paper, we propose a new approach for classifying music data based on their contents. In this approach, we focus on monophonic music features represented as rhythmic and melodic sequences. Moreover, we use repeating patterns of music data to do music classification. For each pattern discovered from a group of music data, we employ a series of measurements to estimate its usefulness for classifying this group of music data. According to the patterns contained in a music piece, we determine which class it should be assigned to. We perform a series of experiments and the results show that our approach performs on average better than the approach based on the probability distribution of contextual information in music.

Abstract. In this paper, we present a technique to automatically create music maps labeled with semantic descriptors, the so called Music Description Maps (MDM). Based on a Self-organizing Map (SOM) trained on audio features, we create term profiles that characterize the type of music in the various clusters. To this end, we efficiently retrieve musicrelated term descriptors for music artists from the Web. These descriptors are used in conjuction with a SOM-labeling strategy to identify words and phrases commonly used in the context of the associated music. Additionally, regions of similar clusters are uncovered. Music maps labeled in such a manner can aid the user in retrieving desired music from a very large repository, either by providing landmarks on the map or by allowing the formulation of queries consisting of terms describing the musical content. 1