SURVEY ON AUDIO SOURCE SEPARATION AND NOTE PROCESSING ALONG WITH EDUCATION

Abstract: Users focus upon real-time music played to them, and at the same time, also enjoys making music. We are conducting a group survey attempting to obtain information that can help eradicate wrong assumptions in designing systems involving music-based learning systems. Our main purpose is to present an overall midi system product. In this paper, we exhibit our initial findings and analyses based on the music requests by users we have received to date. This paper also deals with the separation of music into individual instrument tracks which is known to be a challenging problem. We describe two different deep neural network architectures for this task, a feed-forward and a recurrent one, and show that each of them yields state-of-the-art results on the SISEC DSD100 dataset. The accuracy is estimated for each note played by the user.

Keywords: Video Deep Learning, RNN, LSTM, Note Music, Chord Estimation