Release notes

What's new in Brusmaskin.

3.5 — mastering

August 2026

The last step of making a record, built in. Your export leaves the app sounding like a release — measured, loudness-matched and peak-safe — and the whole album gets the same treatment automatically.

Master your export

• MASTERING — a new row in EXPORT AUDIO. The song is measured the way the pros do it (EBU R128 loudness), raised to the target, and the peaks are caught by a transparent true-peak limiter at −1 dBTP. Nothing pumps, nothing clips — not even between the samples.
• Three targets — STREAM (−14 LUFS, what streaming platforms normalize to), LOUD (−9, a modern electronic release — the default), CLUB (−7, for the DJ hand-off). Or OFF for the raw mix.
• ADJUST — tweak the recipe: the LUFS target itself, TONE (one tilt, darker ↔ brighter), GLUE (gentle mid/side bus compression) and WIDE (a stereo band lift). Neutral by default; picking a target resets the tweaks.
• Album rule — your MASTERING choice is remembered across projects. Export ten tracks, get ten masters with the same loudness and the same character. The export toast shows the measurement: “−13.2 → −9 LUFS”.
• Stems stay raw — per-track normalization would wreck the mix balance, so STEMS exports are never mastered. On purpose.

Master a tape

• MASTER in the cassette view — your jam, mastered to a new tape (“· M”). The original stays untouched, so you can A/B them from the list. Same engine, same targets as the export.

A sweeter engine

• Cleaner oscillators — the VA saw, pulse, sub and all 28 supersaw voices got anti-aliased edges (polyBLEP): the digital hash above the note is down by up to 50 dB. High leads stop fizzing.
• Constant-power pan — panning no longer changes how loud a sound feels; chords, too, keep the same power however many notes you stack.
• Smoother sampler — four-point interpolation on the main path: pitched-down samples lose the gritty stair-steps.
• Same sound at every sample rate — delay tone, reverb damping and sidechain recovery are now clock-true, so a 48 kHz iPhone and the 44.1 kHz web app finally agree. Long 2-beat echoes work all the way down to 60 BPM.

3.4 — tape

August 2026

Press record, go play. This one is about the jam itself — taping it, locking every knob to any step, and steering the song live without ever breaking time.

Jam to tape

• The cassette — press the PRJ tab twice and you're in TPE: a tape deck. NEW TAPE, hit RECORD, and everything that reaches the speakers is written — knob rides, live pads, happy accidents — until you come back and press STOP. The tape keeps rolling while you leave the view and play; a red dot on the tab reminds you.
• A recorder, not a transport slave — the tape doesn't care what the sequencer does. Start it, stop the song, start another one, keep jamming: it's all on the band. Reels spin, the tape migrates, the label tells the time.
• STEMS — flip the toggle and one pass writes ten files: the master, all eight tracks and the FX return — and they sum back to the master exactly. Your performance, ready for a remix in any DAW.
• Out into the world — SHARE opens the sheet with the WAV: AirDrop it, save to Files, send it anywhere. Tapes cap themselves (60 min stereo, 20 min stems) so a forgotten deck never eats the disk.

One sound per pattern

• Pattern default — steps you haven't touched follow the pattern's sound. Change it and they all change with it; lock a step to its own sound only when you mean it. Placing the first step on an empty track opens the picker with an audible starter, so nothing is ever silently blank.
• TRACK FOLLOW — a track can fire from an earlier track's notes instead of its own grid. The whole step row becomes one super-step: its note transposes the incoming one, its sound and envelope are its own. Chains allowed.
• SHAPE — select steps, pick a parameter, and draw ramps, curves and random spreads across the selection in one modal. DONE keeps it, BACK never happened.
• SELECT, properly — its own tab: ALL, SAME SOUND, SAME NOTE, WITH OFFSETS, then refine with EVERY 2ND and INVERT. Reachable from a step and from the track itself.
• COMPACT — squeezes the silence out of a pattern: gaps bake into step LENGTH, the grid shrinks, the sound doesn't change. Roll spans survive, JUMPs are remapped.
• Loop points — the sampler gets LS/LE point-loop markers you can move while the sample plays.

Everything locks

• Every parameter, every step — all ~130 sound parameters can now be locked per step: envelopes, curves, every modulation route, LFO everything. Hold a step, turn any knob. (Engine type and sample stay on the sound — that's what the pool is for.)
• LFO phase is alive — phase is a waveform shift that works in every mode now, and it answers live: turn it and the wave rotates under your finger, mid-note. The DEPTH and PHASE tiles never play dead again.
• Locked enums read as results — a filter-mode lock shows NOTCH, not +2.

Steering the song live

• Jumps in time — hold a phrase or part and press play while the song runs: the switch queues to the next part boundary, sample-exact, and the mode follows what you held. The new CUE toggle next to BPM gives plain taps the same patience.
• FREE chords stay in key — your exact voicing at home, and every chord note follows part transpose in scale degrees. No more accidental chromatic clashes three parts in.
• Live recording, finished — count-in on Shift+metronome, quantize that pushes early notes forward instead of dropping them, velocity, and knob moves recorded as step offsets you hear immediately.
• Save mid-performance — no crackle. The project preview renders at the next stop instead of fighting the audio for the bus.

Ableton Push

• A full control surface — plug in a Push 1, 2 or 3 over USB and play the whole instrument: pads in key, step editing with locks, browse and audition, the sampler, scale mode, the song view with queued jumps, resample gestures, metronome, tap tempo, undo. Encoders tuned to Ableton's own feel, two-handed knob storms included.

Fixed

• The SELECT tab in the track panel showed empty — its content was unreachable behind ACTIONS.
• RANDOM and EUCLID could place silent steps — they were locked to an empty sound slot instead of following the pattern's sound.
• Playing a hair early during live recording could double-trigger the step.
• Notes fiddled on the pads no longer overwrite pattern steps unless Record is armed.

3.3 — grains

August 2026

A seventh engine — granular synthesis over any sample — and a calmer, more consistent library built around one question: what do you want to use this sound as?

GRAIN

• A cloud, not a cursor — GRAIN plays the sample as a swarm of small grains. POS picks the spot (and scrubs — route an envelope or a slow LFO there), SPRAY scatters around it, SIZE sets the grain length, DENS the overlap, JITTER detunes every grain and loosens its timing, WINDOW goes from soft washes to buzzy edges.
• Built like the big ones — density is overlap, so the weave holds up whatever the note or grain size; low notes lengthen their grains instead of falling apart into clicks; grain playback is interpolated so high notes don't hiss; and every knob answers mid-note — each grain latches its settings at birth, so the cloud glides, never zippers.
• It knows your chops — on a sliced sample the cloud quantizes to slice starts, and with keymap on, the keyboard picks the slice. Same slices, trim and settings as the sampler — it is the same material.
• 32 presets — from Still Cloud and Fine Mist to Broken Radio and Everything Cloud. And SYNTHESIZE, chords, arps, the whole FX tail — it is a full citizen.

The library

• One grammar everywhere — every browse page (samples, six preset libraries, your own presets) now has the same two-tab panel: the selected row's verbs up front, the page's tools behind. Nothing shows until it means something.
• USE, not MAKE — a sample is taken into play with USE AS SMPL or USE AS GRAIN (zero audio copies — one file, two instruments), and ON TRACK asks how it should play before where. The sample editor is a real EDIT button now.
• Dangerous things moved out — CLEAR BANK lives with the other mass deletes under BRUS, not next to the everyday verbs.

Stereo samples

• The bank is stereo — imported files keep both channels, and the sampler and the grain cloud both read them. PAN becomes a balance on a stereo sound; fold it with the new MONO flag when you want to place it at a point in the field instead. The flag belongs to the sound, so a clone can stay wide.
• Mono is untouched — a mono file is still stored as one channel. Nothing costs more until you actually record or import something wide.

Capture a track

• CAPTURE — one button on a track's ACTIONS tab. Nothing drops out: the track keeps playing while the capture runs through its own private effects. It starts where that track's loop starts, runs exactly one loop, keeps rolling until the sound has died away — then you audition it and choose what it becomes: SAMPLE, GRAIN or KIT.
• It sounds like what you heard — filter, drive and the track's own reverb and delay tail are in the file, in stereo. The result lands straight in the pool, armed for the next step you place.
• Grain arrives chopped, kits arrive playable — GRAIN slices the capture on its transients so the cloud jumps between the hits; KIT does the same and switches KEYMAP on, so each note plays one hit. SAMPLE stays whole, because a slice table there would play one fragment instead of the loop. The receipt's OPEN jumps straight into the new sound.

Fixed

• Samples played a semitone and a half sharp on iPhone. The bank is 44.1 kHz but the engine runs at the device's rate — 48 kHz on iPhone — and nothing bridged the two. Every imported sample has been playing 8.8 % fast since sampling was added. It is correct now, which means projects with samples will sound lower than you are used to on iPhone. That is the right pitch, not a new bug.
• On PLUCK, DAMP and the filter's cutoff were the same modulation target — routing an LFO to DAMP swept both. They are separate now: DAMP moves the string, CUT moves the filter.
• Modulation targets that an engine does not use are hidden instead of listed. They were selectable and did nothing.
• Pad notes now follow the track's pattern speed — a pad and a step with the same note finally sound identical on speeded tracks.
• A freshly added sound no longer looks assigned to every empty track.
• Empty tracks show a dash instead of a phantom sound id.

3.2 — strings & routes

August 2026

A sixth engine — a plucked string — and a rebuilt modulation system where every envelope and LFO points anywhere: three routes each, on every sound.

PLUCK

• A real string — Karplus–Strong physical modelling: a noise burst rings through a tuned feedback loop. DECAY is measured in time, so the same setting rings equally long across the whole keyboard — and 255 is an endless drone that lives until the track's next note takes over.
• The pick is per note — COLOR (soft thumb to hard plectrum), EXCITE (click to scrape) and POS (the hollow near-the-bridge comb) are sampled at every note-on. Route a random LFO there and every pluck lands differently, like a hand that moves.
• DAMP, STIFF and MODE — the tone knob (bright steel to felted nylon, pitch stays true while you turn it), allpass dispersion toward piano wire and bell, and the original 1983 noise-drum mode hiding in FB POL.
• It behaves like an instrument — chords ring as real multi-string stacks, slides glide the ringing string without re-plucking, and the whole fold / BRUTE / filter / drive / crush tail is there. 32 presets, from Nylon Gut to KS Snare.

The modulation system

• Three envelopes, three LFOs, three routes each — ENV A/B/C and LFO A/B/C are tabs now (tap again to page through), each with the same shape: envelope controls on the left, TARGET/DEPTH routes on the right. ENV A's first route shows its fixed AMP wiring — the hardware is visible. ENV C is free: silent until you point it somewhere.
• Every target works on every engine — the dead slots are gone. Samples get real pitch vibrato and per-note START/SLICE/LENGTH (a random LFO rolls a new chop every note), drums get a moving noise filter, real RATIO and FEEDBACK modulation, CLICK and sweep control — and the target list finally speaks each engine's language.
• Sends are targets — point an LFO or envelope at REV/DEL/CHO SEND for dub throws and tremolo delay.
• New envelope shapes — HOLD on the mod envelope, CURVE (log to exponential) on all three, START/END floors everywhere — and END on the amp envelope turns any sound into a drone that sustains until the next note. At your own risk.
• Your old sounds are untouched — FLT AMT and PIT AMT became routes automatically, verified bit-identical. Everything you saved sounds exactly the same.

Small things

• Presets know their octave — drums and basses audition and land at C2 instead of C4, and your own presets remember the note they were saved on.
• FM's shapers woke up — FOLD, BRUTE and DRIVE now work on the FM engine without engaging the filter, matching every other engine.
• Panels tidied — drum controls in playing order, CLICK/CLAP with the sound design knobs, sends in the mixer's order, and the picker grew a PLK tab.

3.0 — the plugin update

August 2026

Brusmaskin becomes an AUv3 plugin — the whole instrument inside AUM, Logic and GarageBand — plus live offset recording, offsets that sweep the whole way, and presets that sync and merge.

The plugin

• Brus is a plugin now — Brusmaskin loads as an AUv3 instrument in AUM, Logic, GarageBand and friends. Sequencer, sounds, sampler and projects inside your session, on the host's clock: sample-accurate start, fractional tempo, seeks and loops followed.
• Twelve outputs — MAIN carries the mix, and any track can leave it for its own stereo bus (plus an FX return). Tap OUT on a mixer strip and that track goes dry to its own host channel — mix and effect it there instead. Saved with the project.
• Sessions remember — saving a host session saves your working state inside it. Reopen it and Brus comes back exactly as it was — and it sounds even before you open the panel.
• One library — the plugin sees the samples, presets and projects you collected in the app: everything lives in shared on-device storage.

The instrument

• A real sample editor — overview + pinch-zoomable detail down to single samples, one marker at a time selected by ear, zero-crossing snap on trim, slices living inside the trim region, and swipe controls that follow your zoom.
• Color themes — Paper, STHLM, Norrland, Kosmische, Motown and Tokyo restyle the track colors and the project palette together, each with its own night variant (BRUS > SYSTEM). Saved project colors keep their slot in the new dress.
• Projects wear their sound — every row in the project list shows its own waveform, rendered from the audio preview, plus BPM and age.
• LIVE REC — arm REC and hold a parameter tile while the sequencer runs: whatever you set is written to the step playing right now, and you hear it on the ringing note as you move. Hold still to paint the value across every passing trig, use two fingers for two parameters, sweep to 0 and hold to erase.
• Offsets go all the way — a step offset now spans the whole parameter range: take the filter from wherever it sits down to zero or up to full, no matter where it starts.
• Presets that travel — presets are files now: they sync over iCloud, two devices merge their collections instead of overwriting each other, and SAVE PRESET bakes the step's offsets in — the preset is what you hear on that step.
• FOLLOW knows polymeter — with tracks of different lengths, every track pages with its own playhead.
• Duplicate what you hear — DUPLICATE on the active project saves your working copy first, then forks it, samples included.
• What syncs, visible — BRUS > PROJECT shows where projects, presets, samples and settings live, and when they last synced.

2.0 — the project update

August 2026

Projects become real files you can share, preview and sync — plus dark mode, sidechained effects and a faster arrangement workflow. Everything from 1.3 is in here too.

Projects & sharing

• Projects are files — every project is one self-contained .brus file: patterns, sounds and samples. Stored on the device first, with an optional iCloud mirror (BRUS > PROJECT) that follows your projects between iPhone, iPad and Mac.
• Share a project — SHARE PROJECT packs the whole thing; the receiver opens it in Brusmaskin and it just plays. Send it back and forth — replace or keep both on collision.
• One-bar previews — SAVE renders the busiest bar into the file; tap a row in the list to hear a project before loading it.
• Project names & colors — generated names (with a dice to reroll), and a color palette drawn from the eight track colors.
• Backups — EXPORT ALL zips every project file; IMPORT ALL restores them.

Instrument

• Dark mode — the paper flips to night: same instrument, fewer colors (BRUS > PROJECT > DARK).
• Sidechained effects — each effect return (chorus/delay/reverb) can join a sidechain group: the kick ducks the delay tail, repeats pump in time (MST > BUS).
• Pattern grid — tap PAT on a song cell to see every pattern as a tile; audition in context, tap again to bind.
• DOUBLE LEN & CLONE — double a pattern with its content copied, or clone the track's pattern to a free slot and keep editing independently.
• Free-run arp fix — tempo changes no longer push a free-running arp out of phase.
• Mac window — freely resizable; the phone layout scales with the window height.

Sampling

• Your own slice points — AUTO finds the transients (spectral-flux detection tuned on the actual Amen break), markers drag on the waveform with hold-to-zoom precision, and the slice workshop gets its own tab with destructive TRIM/REVERSE/DUPLICATE actions.
• Videos become samples — drop a screen recording (or any video) into ADD SAMPLES and the audio track is pulled out, in the app and on the web.
• Your own presets — save any sound from a step's ACTIONS tab, browse them under USR, and share them as tiny .brus files (SMPL presets carry their sample).
• Samples work like presets — picking a sample copies its slices/trim into the sound, living with the project; EDIT SAMPLE on a step edits that sound's own copy, and cloning a sound forks the slicing without duplicating audio.
• Leaner engine — only the samples your project actually uses are loaded into the audio engine; big libraries no longer cost memory just by existing.

1.3 — the pool update

August 2026

Your own chords, a mutable step, five new LFO destinations, a faster path from browsing sounds to playing them — and the iPad finally uses the whole screen.

Sequencing

• FREE chords — a second chord mode beside the scale chord: NOTE 2–6 set exact semitone intervals on top of the root (±36, chromatic), so your exact voicing survives transposition. All six voices feed the arpeggiator.
• Step MUTE — silence a step without removing it. Behaves exactly like a trig condition that fails (visit counters keep counting, so mute/unmute never re-phases the pattern), and muted steps survive KEEP.

Sounds

• Preset browsing from the top bar — tap SMP again to cycle the FM / MONO / SUPER / DRUM preset libraries: tap to listen, ADD TO POOL to take it, and the steps you place play it until you pick something else.
• The sound pool everywhere — ADD TO POOL from any sample, a POOL tab in the instrument picker, and the SOUND parameter opens the pool browser with the current sound marked.
• Five new LFO destinations — noise, saw (on SUPER the detune spread breathes), pulse, FM feedback and drive. LFO on RATIO sweeps the slice on samples — wavetable scanning through a chopped break.

Synthesize

• Retrained neural seeding that understands envelope speed, proposals across the new LFO destinations, EVOLVE now mutates the amp and LFO genes, auditions match the song tempo and the saved sound's level — and drum patches keep their pitch when saved.

iPad

• The app scales to fill the whole screen — one big instrument, portrait and landscape.

1.2 — the synthesize update

August 2026

Point Brusmaskin at any sample and it grows a synth patch that plays it. Plus a new tuning stage, three new filter modes, and a round of fixes for live tweaking while the sequencer runs.

Synthesize

• SYNTHESIZE — open any sample in the sampler and let the four synth engines race to recreate it. An on-device neural network proposes starting points, a genetic search refines them against the sample's actual spectrum, and every new best candidate auditions automatically. Tap to compare, EVOLVE to breed from a favourite, SAVE to drop the patch into your sound pool — perfectly in tune, ready to sequence.
• Runs entirely on your device. No cloud, no upload — your samples never leave the phone.

Synth engines

• TUNE — constant transpose (±24 semitones) on every melodic engine, offsettable per step.
• PIT AMT — the MOD envelope can now sweep pitch: zaps, drops and 808-style bends.
• NOTCH, PEAK and VOWEL filter modes — VOWEL morphs U-O-A-E-I formants on the cutoff knob.

Fixes & polish

• Tweaking a sound while the sequencer plays now updates ringing notes correctly — step offsets survive, and edits never leak into the wrong voice.
• New steps inherit the track's actual sound after loading a project.
• Older projects load with correct tuning for every sound.
• Sharper progress reporting and candidate selection in the sampler tools.

1.1 — the jam update

July 2026

• SNAP & KEEP — freeze a moment, jam freely, keep the takes you like as new phrases.
• Six demo songs under the BRUS menu — techno, drum & bass, house, deep house, a one-track FM study and an ambient drift.
• Export audio — WAV or M4A, full master or per-track stems, with or without effects, with live render progress.
• EQ per sound — high-pass and low-pass base/width stage on every engine.
• Live phrase/song switching without pattern restarts, a cleaner navigation strip, and init sounds on new tracks.

1.0

June 2026

First release: an 8-track groovebox — four synth engines and a slicing sampler, deep step sequencing with trig conditions and per-step offsets, patterns, parts, phrases and songs, per-track mixing into chorus, delay and plate reverb, MIDI learn, iCloud projects — free, offline, no accounts.