Summoner

Summoner VST is an open source DAW plugin that takes natural language input, consults an LLM (ChatGPT), to adjusts parameters in an internal synthesizer.

(You can clone it here)

Summoner was first developed as a python script, that would interact with Xfer Records’ VST Serum via Max4Live. Later I adapted it into a VST using the JUCE framework, and updated it as Serum 2 was released. I also played explored the option of adding AWS functionalities to have producers login and be charged for their uses in credits. Despite working towards this for a while, I ultimately abandoned it for the hope of keeping the VST open source.

Wanting more control over the synth’s modulation matrix, and picking up the speed of completing the project by incorporating AI tools like Claude Code (as well as starting a masters program centered around audio software engineering, DSP, and machine learning), I ended up moving on from Serum 2 to adapt an internal synthesizer. It has been a fun ride!

Archived Updates:

April 2025 Update: Serum 2

Serum 2 was shadow dropped. Surprise! (From Xfer records, not from me) I’m working on making modulation matrix manipulation work, the ability to select different LLMs, and implementing user login for API access.

February 2025 Update: Initial VST3 Model

In this video, I showcase my first plugin’s features, run a live demo, break down some of the code, and review it after testing it in a track.

November 2024 Update: Initial Prototype

This shows the prototype model of summoner that attaches to Xfer Records Serum, and adjusts parameters based off of natural language input from users.