This repository holds the public numbers of the set and, where present, anonymised previews. The dataset itself is licensed by XIM Audio per set, non-exclusive, on request, with a manifest and a datasheet. Tell us who you are and what you are building and we approve by hand.

Paired production sets, XIM Audio

One song, everything that made it: the producer's session, every lane printed from it, the MIDI it played, and the flat and the mastered master.

Catalogue page with every number and the licence terms: https://ximaudio.com/data/paired-sets

What this repository is

The public numbers of the set as data/catalogue.json, this card, and the preview lanes. The dataset is not downloadable here. It is licensed by XIM Audio per set, non-exclusive, on request: sample and quote at https://ximaudio.com/request, or mikas@ximaudio.com.

Numbers, measured 2026-10-18


Songs	2,172
Files	66,074
Audio	3,588 h*

* Approximated: every lane of 2,009 of the 2,172 songs read from the file headers, the rest at that average, plus 2,335 master files at the average listed length of these songs. Projects and MIDI are not audio and add no hours.

DAW

	songs
Logic	1,950
Ableton	222

Genre

	songs
Pop - Rock	447
Deep House - Garage	320
House Music	249
Progressive House	180
Electro House - Big Room	140
Ambient - Chill	107
Breakbeat - Beats	92
Trance	89
Hip Hop - Urban	61
Funk - R&B - Soul	33
Techno (Peak Time Driving)	33
Tech House	32

23 more on the catalogue page.

Key

	songs
E	186
Am	165
C	133
Em	126
Cm	111
Gm	107
Ab	103
Dm	93
Fm	92
G	89
F	87
D	69

13 more on the catalogue page.

Tempo

min 63, max 195, median 124 BPM

Tempo bands

	songs
60-69	4
70-79	14
80-89	64
90-99	114
100-109	144
110-119	179
120-129	1,070
130-139	232
140-149	83
150-159	44
160-169	23
170-179	32
180-189	4
190-199	1

What you get per item

One folder per song: the project zipped, one WAV per lane, the MIDI folder where the session had notes, the flat master and the mastered master where it exists, with the manifest row. Prepared per request.

Manifest fields: song_id, creator_id, daw, bpm, key, lanes, midi_files, has_mastered, duration.

What you get per song: the project as the producer saved it, one WAV per lane printed from that project with its plugins loaded, the MIDI the session played, the unmastered mix, and the mastered master where one was made. Four views of one piece of music, from one file.

Because every part comes from the same project, they line up sample for sample: the lanes sum to the mix, the MIDI matches the lanes it drove, and the arrangement in the project is the arrangement you hear. No separation model touched any of it, so the lanes have no bleed.

That is the pairing an automatic mixing or mastering model needs, and the symbolic to audio pairing a transcription or synthesis model needs, with the end state of the session as ground truth.

Rights

Every creator in this set accepted terms that name AI training. Every part comes from the creator's own session, and the manifest says which session it came from. Every creator sells on WeMakeDanceMusic, a music production marketplace online since 2012, under terms that name machine learning, keeps ownership and is paid on every licence. Nothing is scraped. Deliveries are anonymised: pseudonymous IDs, no titles, no artist or producer names.

The rights of every set, stated one by one: https://ximaudio.com/rights.

Preview

preview/ holds the 26 lanes of one house song from our own records, the same lanes the player on ximaudio.com streams: AAC 160k, bars 65 to 80, one file per lane, with preview/lanes.json naming the lane and its group. Deliveries are 24-bit WAV, full length.

Formats

.logicx, .als, .wav, .mid

Related sets

stems, daw-sessions, midi

XIM Audio, https://ximaudio.com, LinkedIn.

Downloads last month: 11

Total file size:

7.19 MB