Probabilistic Models for the Transcription of Single-Voice Melodies
The thesis proposed a method for the automatic transcription of single-voice melodies from an acoustic waveform into a symbolic musical notation. The system consisted of a signal processing front-end which calculated a continuous pitch track and of a probabilistic model which converted the pitch track into a discrete musical notation. The proposed probabilistic model consisted of three parts operating in parallel: a pitch trajectory model, a musicological model, and a duration model. The first handled imperfections in the performed/estimated pitch values using a hidden Markov model, the second estimated the musical key signature to improve the transcription accuracy ant the last models the duration of the notes. The thesis covered a literature review on human voice production and on the theory of pitch estimation. In addition, an inspection of an acoustic database recorded for the training and the testing of the proposed model was introduced. The most important part of the thesis is the chapter with the three probabilistic models.