← All work

Product design · Music tooling

Murrsick

Learn a song by its numbers, so you can play lead over it in any key.

Category
Practice tool
Role
Concept, music-theory system, UX, front-end
Year
2026
Status
Concept — deck & engine spec

Overview

Murrsick is a practice tool for lead guitar. It never stores chord letters — a song's progressions live as Nashville scale degrees, relative to the key, so the same song map is instantly correct in every key.

A guitarist on stage doesn't need another chord sheet. They need to know what to play over the part that's happening right now. Existing tools chart songs in fixed chord letters, and the moment the band starts a song a whole step down, that chart is the wrong document.

I built it for the situations where that failure actually bites: improvising softly behind a sermon close in whatever key the backing track is holding, moving a congregation from one key to the next song to song, or remembering a song as the chorus is 1 5 6m 4 and needing to solo over it.

  • 4Modules on one canvas
  • 12Scales in the library
  • 0Chord letters stored

01 — The core idea

Change the key, not the song.

Because the degrees are relative, key-independence is structural rather than a transpose button bolted on afterwards. One field changes, and every chord name, suggested scale, degree strip and fretboard diagram re-derives at once — because none of them were ever written down.

Stored on disk

  • 1 5 6m 4 — the chorus, and it never changes
  • Sections are ordered lists of number tokens
  • Nothing in the file knows what key you're in

Derived on screen

  • In C: C · G · Am · F
  • In E♭: E♭ · B♭ · Cm · A♭
  • Scales, chord tones and neck diagrams follow automatically

02 — Information architecture

One song, studied from the centre out.

The Song Canvas holds a utility bar, an always-visible degree strip of the seven in-key chords, and four modules. Selecting a section makes it active — and the active section drives everything else. Every other module is a lens on it.

  1. Structure — the spine

    The song as an ordered list of sections, each typed as space-separated number tokens. Add, rename, reorder, delete.

  2. Scales — what fits

    For the active section, the scales that fit the whole progression are ranked first, then the per-chord choices for each chord in it.

  3. Arpeggios — what to land on

    Strip the scale away and show only chord tones, labelled by interval. It teaches you to spell a chord on the neck instead of running a shape.

  4. Chart, time and drone

    Personal notes rather than a chord sheet, a metronome at the song's tempo, and a key-pad drone that follows the song's key automatically.

What the engine computes

Resolve & transpose. Any number token to a real chord in the key — 6m becomes Am in C — degree-preserving across every key.

Diatonic sets. The seven chords and numerals of any major or minor key, plus the chord tones for each.

Scale library. Twelve scales — the modes, harmonic and melodic minor, both pentatonics and blues — filtered for relevance to the key.

Capo ↔ shape family. Which open-chord family a capo position gives you, and the capo that gets you back to a family you like.

The fretboard diagrams in the deck are generated, not drawn: a small renderer maps pitch classes onto six strings and twelve frets, so every scale and arpeggio shape is computed from the same note data the app would use.

03 — Presentation

The product deck

Eleven slides on the number system, the Song Canvas, the four modules and the theory engine underneath — with live fretboard diagrams rendered from real note data.

Open the deck full screen ↗

Outcome

What was produced. An eleven-slide product deck specifying the whole system — the number-token format, the Song Canvas layout, all four modules and the four engine capabilities — plus a working fretboard renderer that draws scale and arpeggio shapes from computed pitch classes.

What I learned. Choosing the storage format was the product decision. Once songs are degrees rather than letters, transposition stops being a feature and becomes a property, and half the interface I'd sketched turned out to be unnecessary.

What could be improved. Entering a song is still typing number tokens by hand. Import from an existing chord chart — resolving letters back to degrees against a detected key — would remove the only tedious part of using it.

Where it goes next. The theory engine is the piece to build first, since every module is a view over it. After that, the honest test is taking it into a live set and seeing whether it survives a key change called from the stage.