Artificial Music Generator

Overview

This project implements an Artificial Music Generator using LSTM (Long Short-Term Memory) networks, a type of recurrent neural network (RNN). The system generates music character by character based on a given input dataset. The core idea is to train an LSTM model on a corpus of music data and then sample from the trained model to generate new music compositions.

Implementation

File Structure

• train.py: Python script for training the CharRNN model.
• model.py: Contains the definition of the LSTM model used for music generation.
• sample.py: Python script for sampling text (music) from the trained model.
• data/: Directory containing input data and related files.
  • char_to_idx.json: JSON file mapping characters to their corresponding indices.
  • input.txt: Text file containing input music data.
• logs/: Directory for storing training logs.
  • training_log.csv: CSV file containing training logs such as loss and accuracy.
• model/: Directory for storing model weights.
  • weights.X.h5: HDF5 files storing model weights for different epochs.