Miyaesi: java based implementation for automatic music transcription

Abstract

Music Notes play a major role in the music world. They are extremely important for musicians and composers. Sometimes people like to know music notes of an already composed music which raises the need of music transcription. In the past decade up to now researches and engineers have come up with vivid techniques to do music transcription automatically using probability and signal processing. With the advent of computer science which facilitated encoding and recording music in digital format it became an important topic. Nevertheless, implementation specific details are still rare to find addressing Automatic Music Transcription (AMT). In this paper we discuss concepts behind automatic music transcription and how they are applied in the system Miyaesi, an automatic music transcription system implemented using Java programming language. Further, we discuss how time domain signal analysis and spectrum analysis leads to automatic instrument identification.