How Traklyst identifies DJ set tracklists

The short answer: Traklyst first looks for a tracklist someone already published, in the video description, the chapters, the comments or on MixesDB. It then listens to short fragments across the whole set and matches them with audio recognition services such as ACRCloud. Every track is marked as confirmed by the audio or taken from a published list.

Figures from the live library, last updated 15 October 2026.

1. Finding what was already published

Many sets already have a tracklist somewhere, just not in one place. Traklyst reads the places people actually put them:

• The description of the YouTube or SoundCloud upload.
• Chapters, the timestamps some uploaders add to a video.
• Comments and replies, where fans post full lists and answer each other's “ID?” questions.
• MixesDB, a community wiki of DJ mix tracklists.
• Other uploads of the same set, found by the DJ, the event and the year, when the upload you pasted has no list of its own.
• YouTube's own music list, the songs YouTube's audio matching found in the video. It names records but not where they play.

These lists are a strong lead, but they can be incomplete, in the wrong order or simply wrong. So nothing is taken on trust.

2. Listening to the set

Traklyst takes short fragments of audio from the start of the set to the end and sends them to audio recognition services such as ACRCloud, which match them against catalogues of released music. Where the recognised track changes between two fragments, it listens again in between to pin the transition down to within about 15 seconds.

The recording is downloaded only for as long as the scan runs, to cut those fragments, and is deleted afterwards. Traklyst does not host or share the mix itself; every tracklist links back to the original upload.

3. Checking one against the other

One match on its own proves little: a fragment can land on the wrong recording. A track only counts as confirmed by the audio when the recording was heard at least twice with the same result, or heard once at the point where a published tracklist names it.

• A published entry that the audio hears where the list puts it becomes verified, at the time it was heard.
• A track no list mentions is added as recognised once the audio has heard it at least twice, with the same result each time.
• A published entry the audio never matched stays unconfirmed, with an estimated timestamp marked with a ~.

What the labels mean

• ✓ verified: named by a published tracklist and heard in the recording at that point. The most reliable entry.
• No label: heard in the recording at least twice with the same result, without a published list naming it.
• uncertain: recognised, but the match was weak. Worth a listen before you buy.
• ✓ YouTube: named by a published tracklist and by YouTube's own music list. The record is almost certainly in the set; its timestamp comes from the list.
• unconfirmed: from a published list only. Often right, not proven.
• ID: a stretch where nothing was identified. If you know the track, the page lets you say so.

Why some tracks stay unidentified

Recognition can only find music that the recognition services have in their databases. Sets built on unreleased promos, dubplates, the DJ's own edits or bootlegs can come back with large gaps, however carefully they are scanned. In our own tests, a hard techno set full of unreleased edits matched none of its fragments, while commercial house mixes matched far more. Heavy layering, pitch changes and crowd noise in live recordings make matching harder too.

When a published list covers those gaps, its entries stay visible as unconfirmed, so you still get the lead without it pretending to be proof.

Corrections and removal

Tracklists are public, so anyone can find them. If a track is wrong, or you are the artist, label or uploader and want a tracklist taken down, send a message with the link to the page.