Music Information Retrieval in P2P Networks

In this chapter we present the most significant trends in recent research in the field of content-based music information retrieval in peer-to-peer networks. Despite the diminished attention the area has received in general terms, the relatively close area of metadata MIR in P2P is by far new. As metadata prove to be inefficient for the purposes of MIR as well as the peculiarities of music in comparison to text and image data, developing dedicated solutions for CBMIR in P2P networks becomes a necessity while the challenges faced therein, unique. Depending on the type of P2P network, a number of prominent research works are presented and compared in this chapter.

Identifying Saxophonists from Their Playing Styles

We describe a novel approach to the task of identifying performers from their playing styles. We investigate how professional Jazz saxophonists express and communicate their view of the musical and emotional content of musical pieces and how to use this information in order to automatically identify performers. We study deviations of parameters such as pitch, timing, amplitude and timbre both at an inter-note level and at an intra-note level. Our approach to performer identification consists of establishing a performer dependent mapping of inter-note features to a repertoire of inflections characterized by intra-note features. We present and discuss some results of our approach and comment on future trends in this exciting research area.

An Expert System Devoted to Automated Music Identification and Recognition

This chapter describes a logic and computer algebra based Expert System that automates identification and recognition of the cult music styles of the period XVII century - beginnings of the XX century. It uses a table that contains a list of characteristics (identifiers) of the music styles of the epoch, developed after interacting with a panel of experts. The user, while or after analysing a score, introduces the identifiers and the expert system returns the score’s style. These tentative identifications and recognitions could be interactively compared with those of experts. Therefore it may be a useful tool for teaching and learning history of music. The objectives are to organize the musical knowledge in an way admissible by the inference engine, to adapt and implement the inference engine (based on a powerful tool for effective polynomial computations named Gröbner bases) and to implement a GUI.

MARSYAS-0.2

Marsyas, is an open source audio processing framework with specific emphasis on building Music Information Retrieval systems. It has been been under development since 1998 and has been used for a variety of projects in both academia and industry. In this chapter, the software architecture of Marsyas will be described. The goal is to highlight design challenges and solutions that are relevant to any MIR software. Keywords: Information Processing, Music, Information Retrieval, System Design, Evaluation, Fast Fourier Transfer (FFT), Feature Extraction, MFCC

Pen-Based Interaction for Intuitive Music Composition and Editing

This chapter deals with pen interaction and its use for musical notation composition and editing. The authors first present existing pen-based musical notation editors and argue that they are dedicated to classical musical notations and are often constraining for the user. They then focus on their generic method that is in particular based on a formalism that models how to interpret the strokes drawn in on-line structured documents. To analyze an element, it models a coupling of a global vision (its position relatively to the other elements of the document) with a local vision (its shape and that of its components). The authors also present the hand-drawn stroke analyzer associated to this formalism. Finally they demonstrate how the presented method can be used to design pen-based systems not only for classical musical score notations, but also for plainchant scores, drum tablatures and stringed-instrument tablatures.

A P2P Based Secure Digital Music Distribution Channel

This chapter presents a model enabling content providers to successfully sell digital music. We show that content providers must overcome three main hurdles to successfully sell digital music. The first is to establish an efficient and economically viable distribution channel. Second, they need to develop a secure and interoperable framework for protecting copyrighted digital music from piracy by integrating Digital Rights Management Systems into the distribution channel. The third hurdle is to set-up a robust payment mechanism that meets the content providers’ needs for revenue capturing and the consumers’ needs for hassle-free and legal content acquisition and usage. This chapter finally presents a DRM supported peer-to-peer network which could address and overcome the three hurdles. We conclude that a DRM supported P2P network gives content providers as well as consumers the secure, legal and most cost-efficient and user friendly digital distribution channel they have been searching for.

DART

This chapter introduces the DART (Distributed Audio Retrieval using Triana) project as a framework for facilitating the distributed processing and analysis of audio and Music Information Retrieval. The chapter begins by discussing the background and history of Grid and P2P technologies, the Triana framework, the current tools already developed for audio-rate signal processing, and also gives a description of how Triana is employing a decentralized P2P framework to support MIR applications. A music recommendation system is proposed to demonstrate the DART framework, and the chapter also documents the DART team’s progress towards the creation of a working system. The authors hope that introducing the DART system to the MIR community will not only inform them of a research tool that will benefit the entire field of MIR, but also establish DART as an important tool for the rest of the audio and research communities.

Motivic Pattern Extraction in Symbolic Domain

This chapter offers an overview of computational research in motivic pattern extraction. The central questions underlying the topic, concerning the formalization of the motivic structures, the matching strategies and the filtering of the results, have been addressed in various ways. A detailed analysis of these problems leads to the proposal of a new methodology, which will be developed throughout the study. One main conclusion of this review is that the problems cannot be tackled using purely mathematic or geometric heuristics or classical engineering tools, but require also a detailed understanding of the multiple constraints derived by the underlying cognitive context.

Content-Based Indexing of Symbolic Music Documents

Indexing is the core component of most information retrieval systems, because it allows for a compact representation of the content of a collection of documents, aimed at efficient and scalable access and retrieval. Indexing techniques can be extended also to music, providing that significant descriptors are computed from music documents. These descriptors can be defined as the “lexical units” of music, and depend on the dimensions that are taken into account – melody, harmony, rhythm, timbre – and are related to the way listeners perceive music. This paper describes some relevant aspects of indexing of symbolic music documents, giving a review of its basic concepts and going in more detail about some key aspects, such as the consistency at which candidate index terms are perceived by listeners, the effectiveness of alternative approaches to compute indexes, and how individual indexing schemes can be combined together by applying data fusion approaches.

Music Representation of Score, Sound, MIDI, Structure and Metadata All Integrated in a Single Multilayer Environment Based on XML

In this chapter, we will analyze the heterogeneous contents involved in a comprehensive description of music, organizing them according to a multilayer structure. Each layer we can identify corresponds to a different degree of abstraction in music information. In particular, our approach arranges music contents in six layers: General, Music Logic, Structural, Notational, Performance, and Audio. In order to reflect such organization, we will introduce a new XML-based format, called MX, which is currently undergoing the IEEE standardization process (IEEE SA PAR1599). In an MX file, music symbols, printed scores, audio tracks, computer-driven performances, catalogue metadata, and graphic contents related to a single music piece can be linked and mutually synchronized within the same encoding. The aforementioned multilayer structure allows us to gather and organize heterogeneous contents, leaving them encoded in well-known and commonly used formats aimed at music description.