This is a project that was created as part of creating my master's thesis. It is a program that creates new songs using the capabilities of the LSTM network.
- Please make sure you are using Python version 3.6. To run this project it is also required to configure the local environment so that Tensorflow can use the graphics card. Here you will find an article that will guide you step by step through the process of configuring your environment: link.
- Create a virtual development environment using the
python3 -m venv .venvcommand. - Activate the virtual environment using the
source .venv/bin/activatecommand. - Install dependencies that are in the requirements.txt file using the
pip install -r requirements.txtcommand - To run the neural network learning process, run the
python3 -m src.train.pycommand. This step is required before performing the next step. - The next step is to start the song generation based on the parameters determined in the previous step. To do this, run the
python -m src.generate.pycommand. A file with the .mid extension will be generated, which should then be run in the appropriate program (e.g. LMMS on Ubuntu) for verification.