Automatic extraction of multimedia information from files is recently of great interest. Usually multimedia data available for end users are labeled with some information (title, time, author, etc.), but in most cases it is insufficient for content-based searching. For instance, the user cannot find automatically all segments with his favorite tune played by the flute in the audio CD. To address the task of automatic content-based searching, descriptors need to be assigned at various levels to segments of multimedia files. Moving Picture Experts Group has recently elaborated MPEG-7 standard, named ”Multimedia Content Description Interface” [8], that defines a universal mechanism for exchanging the descriptors. However, neither feature (descriptor) extraction nor searching algorithms are encompassed in MPEG-7. Therefore, automatic extraction of multimedia content, including musical information, should be a subject of study.