How the service sorts a playlist

A playlist in random order feels like radio with jarring jumps: a heavy track after a calm one, electronic after acoustic. The service reorders the list so neighboring tracks are as similar as possible, and the playlist plays like one continuous mix from start to finish.

For every pair of tracks the service fetches a similarity score from Last.fm and builds the full similarity matrix of the playlist; an optimization algorithm then searches for the order with the smoothest transitions. Tracks from the same album stay together and in the right sequence; for tracks Last.fm does not know, artist and tag similarity is used instead.

The bigger the playlist, the better the sorting works. If the list holds a handful of completely different tracks, no magic will make the transitions smooth — the service can only place the least clashing ones side by side. But a playlist of hundreds or a thousand tracks truly opens up: the transitions become gradual, style to style and mood to mood.

Apple Music exports (.txt) and .m3u/.m3u8 playlists are supported — the sorted playlist downloads in the same format, ready to import straight back into your player.