This monograph presents the author's studies in musicrecognition aimed at developing a computer system forautomatic notation of performed music. The performance ofsuch a system is supposed to be similar to that of speechrecognition systems: acoustical data at the input and musicscoreprinting at the output.The approach to pattern recognition employed is thatofartificial perception, based on self-organizing input datain order to segregate patterns before their identificationby artificial intelligencemethods. The special merit of theapproach is that it finds optimal representations of datainstead of directly recognizing patterns.