Template music transcription for different types of musical instruments

Abstract

Automatically recovering the musical notation for a given signal allows reproducing and modifying the original performance. Music transcription can be considered as extracting the sequence of the notes that best models our audio signal. So it is understood as the process of analyzing a music signal in order to write down the parameters of the sounds occurring in it. Another principal application is structured audio coding: a MIDI-like representation which is highly compact but holds on to the identification and characteristics of a musical note to a significant extent. The applied notation can be the traditional musical notation which gives sufficient information for performing the piece using the available musical instruments. Automatic Music Transcription (extracting musical notes from an audio stream) is a complicated task due to the complexity of musical sounds that requires an effective solution. The scope of this article is in the automatic transcription of the harmonic and melodic parts of the real-world music signals. Algorithms are proposed that address two distinct sub-problems of music transcription. The main part of the article emphasizes the estimation of the several concurrent musical notes. The other sub-problem discusses musical meter estimation. This has to do with rhythmic aspects of music and refers to the estimation of the regular pattern of strong and weak beats in a piece of music. Signal processing methods for the automatic transcription of music are developed in this paper.