Exact normalized-key agreement between Serato and Rekordbox across active repertoire.
Music technologyRepertoire archaeology2026
Music Atlas
A 51,302-track archive is useless if the only way back in is remembering what to search for. Music Atlas starts with one record and surfaces nearby tracks from the real corpus by tempo and musical key.
Loading the sanitized corpus…
01 / THE IDEA
The library remembers files. I wanted it to remember relationships.
DJ software is good at storing records and weak at helping a person recover the connections between them. I wanted to choose one track I already knew, then ask what else in the archive could sit near it by pace or harmonic position. The result is a retrieval tool for forgotten repertoire.
Search by artist or title, choose one analysis source, set the tempo tolerance and key rule, then follow any result as a new seed. Every match says why it appeared. There is no hidden taste model.
02 / THE FRICTION
The software disagrees about pitch.
Direct or half/double-time BPM compatibility across valid active comparisons.
Tempo can be normalized conservatively. Key cannot. The interface keeps Serato, Rekordbox, and Traktor separate and makes the visitor choose which analysis lane controls the match.
03 / PUBLIC BOUNDARY
Real corpus. Deliberately narrow export.
The private archive contains 51,302 audio tracks. The public instrument ships 5,451 filtered title-and-artist records drawn from 5,873 tracks in the resolved active repertoire, plus each tool's BPM and key value.
No audio, file paths, filenames, play counts, activity dates, crates, comments, labels, URLs, albums, or library structure are public. The browser receives a static reviewed snapshot. It never connects to the source catalog.