Automatic Recognition of Musical Phrases
The automatic recognition of musical phrase patterns requires some preliminary stages, such as acquisition of musical fragments by means of a MIDI keyboard, storage of MIDI-encoded musical phrases, MIDI data conversion, parametrization of musical phrases, and discretization of parameter values in the case of rule-based decision systems. These tasks result in the creation of musical phrase databases containing feature vectors. The recognition process includes training and classification phases. All of these phases are to be described on the basis of experiments performed by the author. The results shown here aim at underlining the most general problems which are met while analyzing a musical phrase, form or style.
KeywordsFeature Vector Neutral Point Recognition Score Automatic Recognition Absolute Pitch
Unable to display preview. Download preview PDF.