The idea
Instead of generating raw audio, this project explored symbolic composition: learning musical event patterns and producing MIDI sequences that can be played back with virtual instruments.
The experiment
01
Represent music as sequences
Used a notebook-based workflow to examine musical patterns and prepare model inputs for sequential generation.
02
Generate with and without conditioning
Produced two MIDI artifacts: an unconditioned sample and a conditioned sample.
03
Review examples
Created a separate project demonstration video showing the work in context.
What the artifacts show
The uploaded materials include two generated MIDI files. These demonstrate model outputs, but audio quality and model-comparison claims are subjective unless supported by a controlled evaluation.
Unconditioned MIDI ↗Conditioned MIDI ↗See the demonstration
Watch the project video ↗Explore the work
Read the original project materials behind this case study.
Restaurant Recommender