Summary
I designed and shipped a live browser instrument that turns LibreView glucose exports into playable wavetables — solo ownership from data parsing through Web Audio synthesis, MIDI, and a public demo.
Overview
Glukoscillator is a personal creative tool aimed at people who want to hear patterns in continuous glucose data rather than only chart them. The challenge was to turn sparse clinical CSV exports into something musical without requiring a DAW for the first version.
Goals
- Make LibreView CSV data immediately playable in the browser
- Keep the first release usable without install friction
- Leave a clear path to a DAW-native plugin later
Approach
I constrained the MVP to client-side TypeScript and Web Audio so anyone can open a URL and try it. Product decisions focused on mapping curve data to wavetable layers, then adding an effect chain and MIDI only after the core mapping felt musical. Privacy stayed simple: files stay in the browser.
Solution
The shipped app builds a multi-layer wavetable synth from glucose CSV exports, with an effect chain, virtual piano, and MIDI support. A companion VST3 port with JUCE is underway so the same idea can live inside a DAW without rewriting the creative concept.
Impact
A public live demo proves the full loop works — import, hear, play. The project also demonstrates product judgment under constraints: local data handling, progressive feature depth, and a path from web prototype to native plugin.
Looking ahead
Finish the VST3 companion, harden edge cases in CSV formats, and tighten the mapping presets based on what sounds useful in real sessions rather than adding more effects for their own sake.