Music Classification in MIDI Format based on LSTM Mdel

TL;DR

This paper demonstrates that deep learning networks, specifically mLSTM combined with logistic regression, can effectively distinguish between AI-generated and human-composed music in MIDI format with high accuracy.

Contribution

The study introduces a novel approach of converting MIDI music samples into natural language sequences and applying mLSTM for classification, achieving 90% accuracy.

Findings

AI and human music have distinguishable characteristics.

Deep learning can effectively classify music origin.

Achieved 90% accuracy with 10-fold cross validation.

Abstract

Music classification between music made by AI or human composers can be done by deep learning networks. We first transformed music samples in midi format to natural language sequences, then classified these samples by mLSTM (multiplicative Long Short Term Memory) + logistic regression. The accuracy of the result evaluated by 10-fold cross validation can reach 90%. Our work indicates that music generated by AI and human composers do have different characteristics, which can be learned by deep learning networks.