Catálogo de publicaciones - libros

Over the last decade, a series of publications has brought and established new research areas related to music, and intensified the research verging on several disciplinary boundaries, typically dealt with separately. The explosion of collaboration and competition was triggered by the Internet revolution. Research achievements published in the Internet, along with audio and video available through the Internet have made research more efficient. This creates enormous possibilities and synergy. Also standards are more easily defined and implemented. On the other hand, content search of the Internet resources must in response bring new solutions to the problem – most possibly in the form of new standards and technology. Among new emerging areas are: Music Information Retrieval (MIR), Semantic Audio Analysis (SAA), music ontology, and many others. Music Information Retrieval refers to data extraction and retrieval from musical databases found on the Internet. The MIR strategic plans were defined and re-defined many times. Strong collaboration, and at the same time strong competition, afforded solutions to many problems defined within the scope of MIR, and overcame some of the largest obstacles found in this field. In addition, these problems have been addressed by technology, thus no research plans have been immune to the demands of an increasingly competitive technology environment.

The human auditory system pertains to the entire peripheral auditory mechanism. Classically, the peripheral auditory system has been divided into three basic parts - the outer ear, the middle ear, and the inner ear. Each part of the ear serves a specific purpose in the task of detecting and interpreting sound. The outer and middle parts form the conducting apparatus (Durrant and Lovrinic 1997; Gelfand 1998).

This chapter is devoted to intelligent classification of the sound of musical instruments. Although it is possible, and in some applications sufficient to process musical data based on statistical methods, clearly such an approach does not provide either computational or cognitive insight. The principal constituents of intelligent computation techniques are data mining, machine learning, knowledge discovery algorithms, decision-systems, learning algorithms, soft computing techniques, artificial intelligence – some of these notions have become independent areas, and some of them are nearly synonymous. Data mining, also referred to as Knowledge Discovery in Databases – KDD, has been defined by Frawley et al. as “the nontrivial extraction of implicit, previously unknown, and potentially useful information from data”. Soft computing aims at using machine learning to discover and to present knowledge in a form, which is easily comprehensible to humans. Physical systems described by multiple variables and parameter models having non-linear coupling, frequently occur in the fields of physics, engineering, technical applications, economy, etc. The conventional approaches for understanding and predicting the behavior of such systems based on analytical techniques can prove to be very difficult, even at initial stages of establishing an appropriate mathematical model. The computational environment used in such an analytical approach is perhaps too categorical and inflexible to cope with the complexity of physical systems of the real world. It turns out that when dealing with such systems, one has to face a high degree of uncertainty and to tolerate imprecision. Trying to increase precision can be very costly.

Digital signal processing is one of the most rapidly developing areas of science. With the explosive expansion of the Internet, the number of very demanding computer network users increases. Content analysis and searching for specific content are relatively new areas, and therefore new concepts and algorithms of processing them appear quite often. Currently there are no faultless solutions. Sound data analysis is connected with difficulties in analytical description as well as with significant redundancy characterized by high entropy included in the very type of information. Such characteristics also prevail for sound separation problems, hence the number of algorithms for sound separation from musical material. In addition, one should notice that there are some limitations regarding percussion sounds and other non-harmonic sources. Easy extraction of such sounds by means of existing algorithms is not possible. Therefore one could state that such instruments are a source of noise for an algorithm, which makes its operation more difficult, and decreases the accuracy and credibility of the result. Concurrently, one should select the musical material for analysis based on the instrumentation, avoiding non-harmonic sounds. In addition, the articulation such as glissando or tremolo causes the problem of detecting fundamental frequency in the spectrum. Another important factor should also be considered: the music of Western culture is based on simple relations of frequency to fundamental frequency. Therefore it is an obvious consequence that harmonic tones overlap in the spectrum, which makes the operation of most algorithms more difficult. The most basic notion in musical acoustics i.e. sound timbre, as mentioned in Chapter 2, remains unresolved, even if many important research works have been done in the field (e.g. McAdams et al 1995; Cosi et al 1994; Grey 1997; Krimphoff et al 1994; De Poli and Prandoni 1997; Toiviainen et al 1998, Wessel 1979).

The idea of vagueness (contrary to bi-valent logic) appeared at the end of the 19th century, and was formally applied to the field of logic in 1923 by Russell. A Polish logician ukasiewicz first formulated multi-valued logic in 1930. These research studies were carried out long before the assumptions of fuzzy logic which Lofti A. Zadeh originally defined in 1965 (Zadeh 1965), but thanks to his work multi-valued logic was once more discovered. Later, numerous scientists such as Sugeno (1985), Kosko (1997), Kacprzyk and Feddrizzi (1992), Yager (1992), Yamakawa (1989) and others (Bosc and Kacprzyk 1995; Bose 1994; Dubois and Prade 1999; Dubois et al 2002; Larsen 1980; Mamdani 1977; Mendel 1995; Takagi and Sugeno 1998; Zadeh 1999a; Zadeh and Kacprzyk 1992; Zemankowa and Kacprzyk 1993) worked on the idea and further developed it. Also lately, many research works appeared on the use of fuzzy sets, fuzzy logic, and possibility theory for dealing with imprecise information in database management systems (Fuller 1999; Kuncheva and Kacprzyk 2000; Szczepaniak et al 2003; Yu and Kacprzyk 2003). Both theoretical aspects and implemented systems are discussed within the scope of these studies. Since fuzzy logic theory and its applications are covered extensively in literature, only the main features of this theory will be pointed out here.

Interaction between two perceptual modalities, seeing and hearing, their interaction and mutual reinforcement in a complex relationship was a subject of many research studies. The term synesthesia, under which the phenomenon is known, is related to involuntary joining in which the real information of one sense is accompanied by a perception in another sense. Attempts to assess the degree of this interaction have for many reasons been of special interest.

The choice of problems presented in this study is intended to emphasize that in some cases even the classical problems of acoustics can be addressed and solved by means of new methods, especially those arising from the soft computing domain. Before soft computing methods were introduced, all applications dealing with uncertainty were based on the probabilistic approach. Meanwhile, in the case of some of the studied applications, such as automatic recognition of musical phrases, it is impossible to base the research on such an approach only, because each musical phrase has its unique character that cannot be sufficiently described by any statistics. Similarly, the statistical processing of subjective testing results is not fully reliable in most practical applications in which relatively small data sets are available. Moreover, the hitherto used statistical analyses do not allow for directly formulating rules showing the relations between assessed parameters. Such rules are needed to analyze the acoustical phenomena underlying the preference of subjective quality of sound. In the above mentioned applications a rule-based decision systems are necessary to ensure a more accurate data analysis and a better understanding of the phenomena under scrutiny on the basis of obtained results.