build.thiri.ai
Solana · Seeker · Cadenza

THIRI · Chord Intelligence

The deterministic music-intelligence layer for AI.

Generative audio has ears. THIRI is the brain. A computed — not generated harmony engine that parses, spells, voices, and reharmonizes any chord progression — behind a REST API and an MCP server. Correct by construction, on every call.

The control seam

Diffusion models generate texture; they can't reliably control harmony — the changes drift, voice-leading blurs, and "make it Coltrane changes" gets ignored. Harmony is a symbolic problem. THIRI solves it deterministically and hands your model a correct harmonic plan to condition on.

Your model's ears, THIRI's brain. ~40 ms per call. No GPU. Same input → same correct answer, every time.

🎷 Hear it live → future.thiri.ai

Four capabilities, one engine

analyze

chord → quality, intervals, roman numeral, harmonic function

resolve

chord → spelled notes, frequencies, MIDI, scale recommendations

voice

instrument-ready voicings with a voice-leading score

reharmonize

8 techniques — tritone sub, Coltrane changes, backdoor, and more

Built to sit inside

Why THIRI, not a model

Partner with THIRI

Building something that needs to understand harmony? Let's talk about the integration.

THIRI is the intelligence layer of Blues Prince Media's music stack. WoodShed (player tools) and Cadence (rights & settlement) build on the same foundation.