AI Music Generation with LSTM

Project Overview

This system trains on classical MIDI files to learn musical patterns, then generates new compositions using a stacked LSTM neural network architecture.

Model Architecture

3 LSTM layers (512 units each)
Dropout (0.3) for regularization
Dense layers with softmax output
Trained for 20 epochs with Adam optimizer

AI Music Generation with LSTM

Project Overview

Model Architecture

Generated Music Sample (MIDI)