72.67 Rekordbox 7measured by us, same 566 tracks
held-out, shipping appDJ software for your desktopWindows firstyour own files
It reads the music.
It builds the set.
It DJs the night.
A DJ engine that measures its own work before you hear it.
Point it at a folder. It listens to every track, plans a set you can read, edit in the workbench, and export with cue points to Rekordbox or Serato - or press play and let it hold one tempo all night. Every transition is built as real audio and graded before it reaches the room, so nothing untested ever gets out.
One analysis passdiagram
- Grid
- every beat, every downbeat, the 4-bar phrase
- Key
- Camelot, with a confidence
- Tempo
- BPM, and whether it holds
- Shape
- intro / build / drop / breakdown / outro
- Stems
- drums, bass, vocals, melody, separated
A drawn diagram of one analysed track, not a screenshot. The sections are the labels the analyzer uses; the fields are what it stores on your disk.
01One transition
This is the whole product, in one join.
Not a fade out and a fade in. Two tracks locked to one grid, the low end handed over on a phrase boundary, and a gate at the end that can still refuse the whole thing. Switch the technique and the figure redraws at that technique's own length.
Seam study, 32 beatsgrid lock
24 to 40 beats The classic. The lows change hands on a phrase boundary.
Positionbeat 00 / 32
Tempolocked, 0.00%
Low bandoutgoing
Gatemeasuring
A diagram, drawn from the engine's own thresholds and its own beat ranges. It is not a recording and not a screenshot. The real thing is built as audio, measured, and only then allowed into a set.
A diagram is not a recording. Full sets played by the engine, and single joins on their own, land on this site with the first release: what is coming.
02The receipts
Accuracy you can recompute.
Five public benchmarks, scored on music the analyzer was never tuned on, each with a per-track results file behind it. Mixterio is the only DJ product that publishes reproducible, held-out analysis accuracy.
89.1 Beat This!Foscarin, ISMIR 2024
held-out, never trained on78.3 Beat This!Foscarin, ISMIR 2024
held-out, never trained onno published comparisonnobody publishes a figure for this metric, so we draw none
raw grid, general material0.472 Foote (2000)measured by us, same corpus
held-out, coarse HR3FGold is ours, grey is the strongest figure anyone else has on the same row. Bars are drawn to each benchmark's own scale, so no gap is stretched, and where nobody publishes a comparable number we say so instead of implying a win. Measured 2026-07-16. Read the whole board.
03The product
Three tools that happen to live in one app.
Most people will use one of them more than the others. A library you can trust, a set you can plan, a night you do not have to babysit.
01
It reads the music first
If you have thousands of tracks and no idea which ones belong together.
Every track in your folder gets listened to before a note of it plays. What you get back is a library that knows which of your records actually go together.
- Beats and downbeats, tempo, Camelot key, loudness, energy over time
- Section shape: intro, build, drop, breakdown, outro
- Search by how a track sounds, not only by tempo and key
- Pairings proven by building the blend, not guessed from tags
02
Compose without playing
If you plan your sets at home and want the boring half done for you.
A whole set as a document you can read: the sequence, the style of every join, the length of every blend. The Set Workbench lets you swap or remove tracks with ranked measured replacements. Export directly to Rekordbox XML with cue points, or take the tracklist.
- Compose first, listen later, or never
- The Set Workbench: swap or remove tracks with ranked measured replacements
- Audition one transition at a time instead of the whole set
- Export to Rekordbox XML with cue points on every transition, m3u8 and mix sheet
03
Hand it the decks
If a room needs music for six hours and you need to be somewhere else.
One locked tempo, every transition built and graded before you hear it, and a night that survives a spilled drink.
- One tempo for the whole set, held to within 0.01%
- If the laptop dies at hour four, one click picks the night back up
- No track plays twice in a night, and it says so if your folder is too small
- Endless mode that keeps going, and steering in plain language
04The difference
Everyone else predicts. We render it and listen with an instrument.
Compatible key, close tempo, similar energy: that is a guess, and the room is where you find out. We do the work first.
The usual way
- Compare key, tempo, energy
- Suggest a cue point
- Apply a curve live
- Hear it for the first time in the room
Our way
- Build the blend as real audio
- Grade it: timing, tempo, low end, arrival
- Below the floor, it is refused
- The room only ever hears a pass
What the other tools do better today: they mix a brand-new track with no preparation, and they reach streaming catalogues. We listen first and work from your own files. That trade is what buys you a transition that was graded before you heard it.
05Progressive mastery
Your DJ learns your library.
It plays well the hour you install it, on the same quality gates that carried a real gig. When your machine is idle and plugged into power, a background supervisor measures candidate transitions across your collection at the lowest priority. The moment you touch the app, load the processor, or unplug from power, it stops instantly.
The count of proven pairings is on screen in the app. It starts at zero on your machine and grows every day it sits plugged in, prioritising the tracks you actually play. The standard never moves as it grows. Only how much of your library has cleared it.
Measured on our library0k+transitions built and graded so far
On yours, at install0and climbing from the first night
Pair matrix, schematic22k+
- strong
- passed the floor
- tried, rejected
- not tried yet
Rows are outgoing tracks, columns incoming. The diagonal band is dense because neighbouring tempo and key is where good pairings live. A drawn diagram, not real data.
06Honest edges
The four things a sceptic should know first.
Published here rather than discovered later. Each one is a real limit, and each one is a deliberate trade.
01
We analyse before we mix
You point Mixterio at a folder and it studies the music once, before it can play any of it. There is no instant mix of a track it has never heard. That pass is the price of a transition that was graded before it played.
02
Local files, no streaming catalogues
Mixterio works on audio you already own, on your own disk. It cannot mix a streaming library, and being offline is a feature rather than a limitation on a venue's network.
03
Structure is our newest number
Section detection scored at half the field until August 2026, and we published that. Measured at the operating point and decode rate the app runs, it is level with our own Foote reproduction at the strict tolerance and 0.070 behind it at the lenient one - but the benchmark's own human annotators disagree with each other near the top. Fresh numbers earn less trust than old ones; the per-track results file exists so you do not have to take ours.
04
Four genres are calibrated
Techno, hip-hop, jazz and ambient are the material the engine was tuned and listened to on. Anything else will play, and we are not going to pretend it was measured.
07Price
Free is a real tier, not a teaser.
One folder, the whole engine, no countdown clock. Pro lifts the library and session limits and adds recording, exports, saved sets and personalization. There is a lifetime licence too, for people who would rather own a tool than rent one.
Free$0
Pro$12/mo
Founder Lifetime$199
Be there at launch
Hear it before anyone else.
The installer is not out yet. Leave your email and you get one message the day it is, with the demo sets landing on this site at the same time.