A probabilistic approach to melodic similarity (original) (raw)

A probabilistic model of melodic similarity

Proceedings of the International …, 2002

Melodic similarity is an important concept for music databases, musicological studies, and interactive music systems. Dynamic programming is commonly used to compare melodies, often with a distance function based on pitch differences measured in semitones. This approach computes an "edit distance" as a measure of melodic dissimilarity. The problem can also be viewed in probabilistic terms: What is the probability that a melody is a "mutation" of another melody, given a table of mutation probabilities? We explain this approach and demonstrate how it can be used to search a database of melodies. Our experiments show that the probabilistic model performs better than a typical "edit distance" comparison.

A measure of melodic similarity based on a graph representation of the music structure‖

2009

Content-based music retrieval requires to define a similarity measure between music documents. In this paper, we propose a novel similarity measure between melodic content, as represented in symbolic notation, that takes into account musicological aspects on the structural function of the melodic elements. The approach is based on the representation of a collection of music scores with a graph structure, where terminal nodes directly describe the music content, internal nodes represent its incremental generalization, and arcs denote the relationships among them. The similarity between two melodies can be computed by analyzing the graph structure and finding the shortest path between the corresponding nodes inside the graph. Preliminary results in terms of music similarity are presented using a small test collection.

Tree-structured representation of melodies for comparison and retrieval

2002

The success of the Internet has filled the net with lots of symbolic representations of music works. Two kinds of problems arise to the user: the search for information from content and the identification of similar works. Both belong to the pattern recognition domain. In contrast to most of the existing approaches, we pose a non-linear representation of a melody, based on trees that express the metric and rhythm of music in a natural way. This representation provide a number of advantages: more musical significance, more compact representation and others. Here we have worked on the comparison of melodies and patterns, leading to motive extraction and its use for the identification of complete melodies.

An evaluation of melodic similarity models

2008

The advance of music information retrieval (MIR) has brought about a strong interest in melodic similarity models. In fact, the majority of these models derive their rational from the implementation within a MIR context. This is, so authors argue in general, a model is suitable if it retrieves the desired melody or a close variation of it. This post hoc method does not permit the evaluation and comparison of these models (although such a “comparison” competition has been proposed for ISMIR 2005). The author of this paper approaches the issue form a different angle; rather than testing models in an unsystematic fashion, the author will discuss the underlying cognitive principles of the similarity models. Here, it will be shown that four principle strategies exist: The contrast models, the distance models, dynamic programming and transition matrices. The author will then demonstrate that a variety of distance measures based upon a specific representation of melodies appears most promi...

Classifying Melodies Using Tree Grammars

Lecture Notes in Computer Science, 2011

Similarity computation is a difficult issue in music information retrieval, because it tries to emulate the special ability that humans show for pattern recognition in general, and particularly in the presence of noisy data. A number of works have addressed the problem of what is the best representation for symbolic music in this context. The tree representation, using rhythm for defining the tree structure and pitch information for leaf and node labeling has proven to be effective in melodic similarity computation. In this paper we propose a solution when we have melodies represented by trees for the training but the duration information is not available for the input data. For that, we infer a probabilistic context-free grammar using the information in the trees (duration and pitch) and classify new melodies represented by strings using only the pitch. The case study in this paper is to identify a snippet query among a set of songs stored in symbolic format. For it, the utilized method must be able to deal with inexact queries and efficient for scalability issues.

Tree-Structured Representation of Musical Information

2003

The success of the Internet has filled the net with lots of symbolic representations of music works. Two kinds of problems arise to the user: content-based search of music and the identification of similar works. Both belong to the pattern recognition domain. In contrast to most of the existing approaches, we pose a non-linear representation of a melody, based on trees that express the metric and rhythm of music in a natural way. This representation provide a number of advantages: more musical significance, more compact representation and others. Here we have worked on the comparison of melodies for identification.

A Melodic Similarity Measure Based on Human Similarity Judgments

Music software applications often require similarity-finding methods. One instance involves performing content-based searches, where music similar to what is heard by the listener is retrieved from a database using audio or symbolic input. Another instance involves music generation tools where compositional suggestions are provided by the application based on user-provided musical choices (e.g. genre, rhythm and so on) or samples. The application would then generate new samples of music with varying degrees of musical similarity. Although several similarity algorithms such as edit distance methods and hidden Markov models already exist, they are not fully informed by human judgments. Furthermore, only a few studies have compared human similarity judgments with algorithmic judgments. In this study, we describe an empirically derived measure, from participant judgments based on multiple linear regression, for determining similarity between two melodies with a one-note change. Eight standard melodies of equal duration (eight notes) were systematically varied with respect to pitch distance, pitch direction, tonal stability, rhythmic salience, and melodic contour. Twelve comparison melodies with one-note changes were created for each standard. These comparison melodies were presented to participants in transposed and non-transposed conditions. For the non-transposed condition, predictors of similarity were pitch distance, direction and melodic contour. For the transposed condition, predictors were tonal stability and melodic contour. In a follow-up experiment, we show that our empirically derived measure of melodic similarity yielded superior performance to the Mongeau and Sankoff similarity algorithm. We intend to extend this measure to comparison melodies with multiple note changes.

Generalized N-gram Measures for Melodic Similarity

In this paper we propose three generalizations of well-known N-gram approaches for measuring similarity of single-line melodies. In a former paper we compared around 50 similarity measures for melodies with empirical data from music psychological experiments. Similarity measures based on edit distances and N-grams always showed the best results for different contexts. This paper aims at a generalization of N-gram measures that can combine N-gram and other similarity measures in a fairly general way.

Evaluation of approaches to measuring melodic similarity

This paper describes an empirical approach to evaluating similarity measures for the comparision of two note sequences or melodies. In the first sections the experimental approach and the empirical results of previous studies on melodic similarity are reported. In the discussion section several questions are raised that concern the nature of similarity or distance measures for melodies and musical material in general. The approach taken here is based on an empirical comparision of a variety of similarity measures with experimentally gathered rating data from human music experts. An optimal measure is constructed on the basis of a linear model.

Trees and Combined Methods for Monophonic Music Similarity Evaluation

This abstract describes the four methods presented by us with the objective of obtaining a good trade-off between accuracy and processing time [3]. Three of them are based on a summarization of the input musical data: the tree rep-resentation approach [5, 6] (UA T-RI2, and UA T3-RI3), and the quantized point-pattern representation [1] (UA PR -RI4). The fourth method is an ensemble of methods [4] (UA C-RI1). The summarization methods are expected to be faster than approaches dealing with raw representations of data. The ensemble combines different approaches try-ing to be more robust and are expected to give equal or bet-ter accuracy than the summarization methods. Thousands of different parametrizations of those methods are possi-ble. The parameters of the presented methods are chosen based on previous experiments.