How music
moved through
the decades.

Every release notched against its year. The mountain rises steadily from the post-war decades, surges through the digital boom, and tells you in one glance just how much more music exists today than at any prior moment in recorded history.

A stacked century: the top-ranking AI-classified genres, plotted by their share of each year's indexed releases. Watch rock swell through the seventies, electronic sneak in during the eighties, and hip-hop's long climb after.

Mood labels are derived from the same audio model — a quick read on whether the decade was feeling triumphant, melancholic, dreamy, dark. Lines are rolling shares within each year's indexed pool.

Average beats-per-minute across each year sits beside an average energy score — a 0–1 axis the model assigns each track. They don't always move together. Sometimes the world wants fast and soft.

A mean BPM line can hold steady while the spread underneath it moves. Here is the distribution itself — what share of each year fell into each tempo band, and the overall shape across the whole catalogue.

Average track length by year sits beside the overall spread of durations — from short stingers and edits to full-length cuts.

What share of each year's catalog is purely instrumental? Each point counts only tracks whose audio we have measured, so the line reflects what the music sounds like rather than whether lyrics were published for it.

Cumulative rankings across every track indexed from streaming platforms — not biased by year, just raw volume.

Two simple distributions — energy level and vocal type — to close out the atlas. Both flatten time entirely; they're the catalog's overall fingerprint.

Every Apple cover carries a dominant background colour. Bucketed into families and tracked across time, the palette of released music becomes its own trend line — the warm-to-cool swing of an era, made visible.

Beyond how much gets made — how popular is it? Each genre and mood ranked by its average streaming popularity (a 0–100 index from Deezer's rank) across the tracks where we have it. Not volume — the sounds the world actually plays.

Not how much, not how it sounds — who commands the audience. Artist fandom measured by Deezer fan counts, matched to the commercial catalog. A handful of names carry tens of millions; the rest form a very long tail.

Music never moves alone. Pandemics, wars, new machines and cultural ruptures all leave their fingerprints on what gets made — and when. These are the years worth holding against the curves above.

How much of the music arriving since 2023 is AI-generated. Computed by MatchTune AI Detector, trained on nine known generators, so every figure here is a floor: music from generators it has never heard stays invisible. Shares are computed over the tracks scanned so far — coverage is shown alongside.

The artist's home country — not the label's, not the listener's. Known for — artists behind — tracks: the recognisable part of the catalogue, with the long tail of unknown uploaders left out. Shares are of that slice.

Gender, groups versus solo acts, and how old an artist is when a track comes out — drawn from the public record of — artists. This is the notable end of the catalogue; the anonymous majority has no such record.

— tracks matched to — films, series and games whose composer is on public record. Song-only soundtracks and unlisted scores are not counted, so these are indexed scores, not all screen music.