iOS · Feature-complete · Coming with iOS 26.5
DrumFillGenerator writes strictly linear fills — one voice per step, never two — then solves the sticking with a search that refuses to cross your hands, lunge across the kit, or park you on a single drum. Plug in a MIDI e-kit and every hit you play is measured against the grid to the millisecond. A bank of ~200 hand-transcribed fills rides along for the days you want someone else's ideas.
What it does
One writes fills your hands can play. The other tells you, to the millisecond, whether you played them. Both are free — along with all ~200 curated fills, humanized playback, a rock-solid audio engine, and two ways to read every pattern.
Every generated fill goes through a dynamic-programming search over your kit's geometry. It costs out hand travel, refuses to cross your right hand under your left, penalizes direction reversals, and won't ask for a leap the note spacing can't carry.
Connect a USB-C MIDI e-kit and every Note On is timestamped against the grid as you play. Inside 25 ms is a hit, out to 80 ms reads early or late, past that is a miss — green, amber, red, flashing on the step you actually struck. Timeline markers and loop cycle history included.
Robot, Tight, Natural, or Loose — inject micro-timing slop and velocity swing into audio playback only. Visuals and metronome stay on the mathematical grid. Responds live mid-playback, no restart needed.
Read every fill as a chunky color-coded step grid (free) or a traditional 5-line drum staff (premium) — both render from the same source. Tap a step to cue, drag to scrub or seek live.
Bookmark a fill you like with a single tap — generated or curated, both save the same way. Rename them, delete the ones that stopped teaching you anything, and reload any of them into the workspace from the Favorites tab.
AVAudioEngine with per-instrument voice pooling, three velocity tiers, ergonomic hand orchestration via DP-based spatial search, metronome, and a prefill groove that wraps your fill in full-song context.
One voice per step, never two — the defining constraint of the linear vocabulary, enforced while the pattern is built rather than filtered out after. No limb three times in a row, the kick stays on your foot, and the final two steps are never both kicks, so looping never turns your fill into a stutter.
Alongside the generator sits a professionally-authored bank of linear fills across genres — 4/4 and 3/4, quarter through triplet, including fills that switch subdivision mid-bar. Generate presses alternate between the two sources, so at least half of what you get is freshly written. All free, and updated remotely without an app release.
A synced click with accent on the downbeat, prefill/afterfill groove for full-song context, per-instrument gain control, rudiment detection, and tap-to-cue transport. Built for serious practice.
How it works
Pick a time signature (4/4 or 3/4), subdivisions from quarters to triplets (multi-select supported), 1–4 bars, and toggle which of the six voices play.
The generator builds a linear fill inside exactly those constraints and solves the sticking for your kit via a DP-based spatial search. Don't like it? Press again. Every other press pulls from the ~200 curated bank instead, so ideas arrive from both directions.
Loop at any tempo with the click on, toggle the prefill groove for full-song context, and dial in your Playback Feel — from robotic to loose. Flip between grid and notation, tap to cue, drag to scrub.
Connect a MIDI e-kit, enable tracking, and play along. Every hit lands on the timeline against where it should have been, scored green/amber/red to the millisecond, with a per-cycle history waiting when the loop comes round. Drop the tempo until the bar is all green, then climb.
Under the hood
Be first on the kit
One email when DrumFillGenerator hits the App Store. No spam, no drip campaign — just the launch.