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

TL;DR

This paper explores deep learning techniques for music transcription, introducing data and "artificial semantic" augmentation methods to improve note recognition accuracy for monophonic and polyphonic music.

Contribution

It proposes novel data augmentation strategies, including "artificial semantic" augmentation, to enhance deep learning performance in music transcription tasks.

Findings

Data augmentation improves recognition accuracy.

Artificial semantic augmentation increases training data diversity.

Enhanced methods outperform previous approaches.

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.