Music Transcription by Deep Learning with Data and "Artificial Semantic" Augmentation

Abstract

In this progress paper the previous results of the single note recognition by deep learning are presented. The several ways for data augmentation and "artificial semantic" augmentation are proposed to enhance efficiency of deep learning approaches for monophonic and polyphonic note recognition by increase of dimensions of training data, their lossless and lossy transformations.

Tasks

Reproductions