Every chord. Every function. Every cadence. Every surprise.
MusicMind detects chord progressions using De Clercq & Temperley template correlation, maps each chord to its harmonic function, scores harmonic sophistication, identifies secondary dominants and cadence types, and measures surprise from IDyOM information content.
Analyze a track — free →What you get
Chord timeline
Every chord with start/end timestamps, synced to the waveform
Roman numeral analysis
Harmonic function (I, IV, V, vi...) relative to detected key
Harmonic sophistication score
Tymoczko-inspired metric measuring voice-leading efficiency, chromatic density, and non-diatonic chord usage across the track
Secondary dominants
Identifies V/V, V/vi, and other secondary dominants that tonicize non-tonic chords — the "borrowed" tension that makes progressions interesting
Cadence detection
Authentic (V-I), plagal (IV-I), deceptive (V-vi), and half cadences identified at phrase boundaries with timestamps
Surprise score (IDyOM)
Information-theoretic surprise from the IDyOM model — how unexpected each chord is given the preceding context. High surprise = the moment that catches your ear.
Tension scoring
Harmonic tension mapped across the track from 0 (resolved) to 1 (maximum tension)
Progression patterns
Common patterns identified (I-V-vi-IV, ii-V-I, etc.) with frequency counts
Chord transition probabilities
From De Clercq & Temperley (2011) corpus analysis. Rows = current chord, columns = next chord. Brighter = more likely.
| I | ii | iii | IV | V | vi | bVII | other | |
|---|---|---|---|---|---|---|---|---|
| I | 10 | 8 | 5 | 25 | 20 | 18 | 8 | 6 |
| ii | 5 | 5 | 3 | 15 | 45 | 10 | 7 | 10 |
| iii | 8 | 5 | 5 | 30 | 10 | 25 | 7 | 10 |
| IV | 25 | 8 | 5 | 10 | 30 | 10 | 5 | 7 |
| V | 40 | 5 | 5 | 15 | 10 | 12 | 5 | 8 |
| vi | 10 | 15 | 8 | 30 | 20 | 8 | 4 | 5 |
| bVII | 30 | 5 | 5 | 25 | 15 | 8 | 5 | 7 |
Harmonic sophistication
Inspired by Dmitri Tymoczko's "A Geometry of Music" — sophistication isn't complexity for its own sake.
Tymoczko showed that great harmony balances familiarity with surprise. MusicMind computes a harmonic sophistication score by measuring three dimensions: voice-leading efficiency (how smoothly chords connect), chromatic density (how much the harmony moves outside the home key), and non-diatonic chord ratio (borrowed chords, secondary dominants, modal interchange).
A four-chord pop song scores low. A Radiohead track with constant modal shifts scores high. Neither is better — the score tells you where a track sits on the spectrum and how its harmonic language compares to genre norms.
How it works
MusicMind extracts chroma features from the audio — a 12-dimensional representation of pitch energy over time. These features are compared against chord templates using correlation, identifying the most likely chord at each time step.
Each detected chord is then mapped to a Roman numeral function using the detected key as reference. Tension is scored by measuring harmonic distance from tonic, accounting for chord complexity and voice leading.
Ivory Cascade
Zeltor Kemp • Piano
86 chords total • Tension bars shown above
Demo data, not research results. Try your own track.
Analyze free →A report also shows (demo data)
Key detected
C major
Tempo
72 BPM
Narrative arc
crescendo
Your analysis includes all 14 cognitive layers.
See all 14 layers →Full MusicMind Report
This tool shows 1 of 8 sections.
The full MusicMind Report delivers all 8 in one document — shareable, exportable, and ready for your team.
Not ready to upload?
Get notified about chord progression analysis updates.