LSTM generative music model
For this project, I developed an LSTM-based generative music model capable of learning musical structure from MIDI data and producing original compositions. It details the full pipeline—from preprocessing MIDI files to hyperparameter tuning, performance optimisation using GPUs, and multi-branch LSTM architecture design. Quantitative testing (loss/accuracy metrics, optimiser comparison, gridsearch for dropout rates) shows that the model can generate musically coherent pieces.