| --- |
| { |
| "pretty_name": "XIM Audio Stems", |
| "license": "other", |
| "license_name": "xim-audio-dataset-licence", |
| "license_link": "https://ximaudio.com/rights", |
| "tags": [ |
| "music", |
| "audio", |
| "ai-training-data", |
| "licensed", |
| "dance-music", |
| "stems" |
| ], |
| "task_categories": [ |
| "audio-to-audio", |
| "audio-classification" |
| ], |
| "size_categories": [ |
| "1K<n<10K" |
| ], |
| "extra_gated_heading": "Request the index and previews", |
| "extra_gated_prompt": "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.", |
| "extra_gated_fields": { |
| "Company or lab": "text", |
| "Country": "country", |
| "What you are building": "text", |
| "I understand the files here are an index and previews, and that the dataset is licensed by XIM Audio on request": "checkbox" |
| }, |
| "extra_gated_button_content": "Request access" |
| } |
| --- |
| |
| # Stems, XIM Audio |
|
|
| Every lane of a finished song, printed by the producer from the session it was made in. One 24-bit WAV per lane, MIDI where the session had it. Pop, house, hip hop and more. |
|
|
| Catalogue page with every number and the licence terms: [https://ximaudio.com/data/stems](https://ximaudio.com/data/stems) |
|
|
| ## 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](https://ximaudio.com/request), or mikas@ximaudio.com. |
|
|
| ## Numbers, measured 2026-10-18 |
|
|
| | | | |
| |---|---:| |
| | Stems packs | 2,817 | |
| | Files | 61,751 | |
| | Audio | 5,351 h* | |
|
|
|
|
| \* Approximated: every lane's length added up, read from the file headers of 2,616 songs and packs on 2026-10-17, plus 201 not yet measured at the average. 167 of these songs are not on the public store yet. |
|
|
| ### DAW |
|
|
| | | stems packs | |
| |---|---:| |
| | Logic | 2,025 | |
| | Ableton | 248 | |
| | Cubase | 14 | |
| | FL Studio | 5 | |
| ### Genre |
|
|
| | | stems packs | |
| |---|---:| |
| | Pop - Rock | 527 | |
| | Deep House - Garage | 401 | |
| | House Music | 345 | |
| | Progressive House | 217 | |
| | Electro House - Big Room | 197 | |
| | Trance | 157 | |
| | Ambient - Chill | 141 | |
| | Breakbeat - Beats | 114 | |
| | Hip Hop - Urban | 83 | |
| | Techno (Peak Time Driving) | 59 | |
| | Funk - R&B - Soul | 48 | |
| | Tech House | 40 | |
|
|
| 23 more on the catalogue page. |
| ### Key |
|
|
| | | stems packs | |
| |---|---:| |
| | E | 212 | |
| | Am | 204 | |
| | C | 178 | |
| | F | 150 | |
| | Em | 147 | |
| | Cm | 146 | |
| | G | 137 | |
| | Gm | 134 | |
| | Ab | 133 | |
| | Fm | 126 | |
| | Dm | 110 | |
| | D | 93 | |
|
|
| 13 more on the catalogue page. |
| ### Tempo |
|
|
| min 63, max 195, median 125 BPM |
|
|
| ### Tempo bands |
|
|
| | | stems packs | |
| |---|---:| |
| | 60-69 | 4 | |
| | 70-79 | 23 | |
| | 80-89 | 78 | |
| | 90-99 | 150 | |
| | 100-109 | 164 | |
| | 110-119 | 225 | |
| | 120-129 | 1,404 | |
| | 130-139 | 359 | |
| | 140-149 | 109 | |
| | 150-159 | 53 | |
| | 160-169 | 30 | |
| | 170-179 | 40 | |
| | 180-189 | 6 | |
| | 190-199 | 2 | |
|
|
| ## What you get per item |
|
|
| One folder per song, one WAV per lane with the producer's lane name, the unmastered mix of the song, a MIDI folder where the session had notes, and the manifest row with one instrument label per lane. |
|
|
| Manifest fields: `song_id`, `creator_id`, `daw`, `bpm`, `key`, `sample_rate`, `stems`, `lane_label`, `duration`, `has_midi`. |
|
|
| No separation model touched these files. Each WAV is one lane of the producer's session, printed from the original project with its plugins loaded. Each lane carries one of 53 instrument labels in the manifest, taken from the producer's own lane name. |
|
|
| Every delivery includes the unmastered mix of each song, summed from its lanes, so a separation model learns from the source the parts came from, not from a guess. Train on these pairs and it pulls cleaner parts out of any finished mix. A mastered reference of the same song on request. |
|
|
| ## Rights |
|
|
| Every creator in this set accepted terms that name AI training. Every stem is printed from the creator's own session, part by part, 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](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](https://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 |
|
|
| WAV 24-bit stereo, AIFF, MIDI |
|
|
| ## Related sets |
|
|
| [daw-sessions](https://huggingface.co/datasets/ximaudio/daw-sessions), [midi](https://huggingface.co/datasets/ximaudio/midi), [acapella-instrumental-pairs](https://huggingface.co/datasets/ximaudio/acapella-instrumental-pairs) |
|
|
| XIM Audio, [https://ximaudio.com](https://ximaudio.com), [LinkedIn](https://www.linkedin.com/company/ximaudio/). |
|
|
