Skip to main content

​
Upcoming fixes

  • Restored the focused physical Sample Card redemption experience: creators can choose the silver-white, frost-brushed Minimalist theme or dark technical Meiji theme in Studio Experience, and that saved choice now covers every customer state. The selected palette styles the complete document and remains fixed across device color settings, with versioned assets preventing an older cached stylesheet from hiding a newly selected theme. Public pack and redemption pages present uploaded artwork as a physical card with a graceful fallback, subtle pointer and touch-scroll motion, reduced-motion safeguards, preview/delivery/value details, and post-redemption discovery across the creator and live catalog. Public discovery remains read-only and never exposes downloads, serials, storage keys, card batches, or drafts. See Sample Cards And Packs.
  • DJ mode now has its own output buffer at Settings → DJ → Buffer Size, independent of Settings → Audio → Buffer Size. It defaults to 128 samples (2.90 ms) at 44100 Hz while the sampler stays at 256, cutting about 2.9 ms from every jog, scratch, and hot cue: the deck path renders a frame at a time and drains every pending control per frame, so the output buffer is the dominant control-to-speaker delay. The default stops at 128 rather than 64 because the deck’s windowed-sinc reader costs roughly 2.25× per frame and a 64-frame callback leaves too little headroom on slower machines. 64 remains available for the tightest deck, and 256 or higher remains the crackle escape hatch. There is no restart prompt — the size applies the next time DJ audio opens, on re-entering the DJ tab or the next controller hardware attach — and existing configs pick up the new default without editing. See DJ Mode, Settings, and Audio And Playback Problems.
  • Restored the REV7 hardware-output contract after a fallback regression could open the controller as a ten-channel DJ graph. Meiji Sampler now keeps the proven four-channel deck route, tries the controller’s current native sample rate before broad advertised-rate fallbacks, and preserves the configured frame buffer. Runtime diagnostics report the actual CPAL stream width and rate. This protects ordinary playback and stopped-motor jog audio from channel-framing and unnecessarily short callback deadlines.
  • Added contextual PerformFX rows to DJ mode. Every enabled, nonblank bank appears between the deck blocks and dual waveforms, with the same assigned, queued, active, release-pending, releasing, and missing states as the Perform tab. [↑/↓] moves focus through Decks, the visible banks, and Library; [TAB] remains exclusively reserved for switching application tabs. [1]–[9], [0] trigger the focused FX bank; Decks focus keeps [1]–[8] for hot cues, and Library consumes digits. [A] or [B] always returns focus to the selected deck. See DJ Mode and Keyboard Shortcuts.
  • The Pioneer DDJ-REV7 can now be powered on or off while Meiji Sampler is open. A runtime MIDI connection reports Pioneer DDJ-REV7 connected — decks are live; disconnecting reports that keyboard control continues while decks remain on system output. The controller’s MIDI controls can arrive before its audio interface or optional high-resolution HID display, so Meiji Sampler keeps the set running and attaches those delayed services when they become available. Restarting is only an escalation after the controller has completed booting and USB, competing DJ software, and a reconnect have been checked. Automated coverage uses virtual CoreMIDI add/remove devices, so this change does not claim physical REV7 boot-timing certification. See DJ Mode and Audio And Playback Problems.
  • A delayed REV7 HID endpoint now retries independently from hardware audio. When it appears, both loaded decks replay their complete waveform, position, pitch, range, and BPM state instead of receiving transport updates on a blank platter timeline.
  • Corrected the REV7 high-resolution waveform and remaining-time software timeline for the reported 96.6-second track, which is intended to prevent the prior stop near -01:00.4. Post-fix physical display behavior awaits hardware confirmation.
  • Restored clean, responsive controller audio in DJ mode. Every controller report now dispatches with its bounded controller snapshot — jog and pitch-fader gestures no longer wait out a 5 ms aggregation window before reaching the deck, removing up to 5 ms of dead time plus its timer jitter from every scratch and pitch move. The anti-warble smoothing moved into the deck, which estimates platter velocity on a fixed 5 ms grid spanning roughly the last 20 ms of reports — a fixed estimation grid keeps the encoder’s integer tick counts from beating against the estimate and wobbling the pitch, and a wow-and-flutter regression harness now pins that steadiness against a simulated jittery 1 kHz platter stream. The deck follows the velocity curve continuously instead of teleporting its audible playhead to every report, a direction change flushes the window so reversals turn around immediately, and cue capture still uses the event-time controller position. Bend mode measures the platter against its motor over the same coalesced window, so encoder quantization can no longer masquerade as a hand and flutter the pitch. Hot cues and scratch onset use a sample-rate-derived 1 ms click-safe transition; ordinary play, pause, and release keep their 5 ms transport fade. DJ output honors Settings → DJ → Buffer Size, so 64 samples is available for the lowest latency while 128 (the default) or 256 remain effective crackle fallbacks. Deterministic hardware-path gates cover 64-, 128-, and 256-frame callbacks, with hot-cue level reaching 99% in 43 frames (0.975 ms at 44.1 kHz) after command admission. Letting go of the platter is timed the same way: the scratch release, and the pulse nudge snapping back to zero offset, are now checked every 2 ms instead of once per 33 ms interface frame, so the track comes back under the hand without up to a frame of extra delay.
  • DJ deck controls now use a bounded real-time command mailbox without silently moving the UI ahead of audio. If a transport, cue, pitch, keylock, nudge, sync, or scratch action cannot be admitted, the deck keeps its prior visible state and reports not applied — audio queue busy so the action can be retried.
  • Added DJ master recording on the DJ tab. Press [R] to start or stop the safety-limited post-PerformFX software master, with a persistent red REC mm:ss clock and dynamic library-footer command. [CTRL+R] still opens input recording. DJ set files are stereo IEEE Float32 32-bit WAVs at the active output rate in Bounces, named DJ Set YYYY-MM-DD HH-MM-SS.wav, with collision suffixes and numbered Part 2 rollover before classic WAV reaches its size limit. Both output modes record the software master. REV7 hardware mode records the unity-gain sum of processed deck stems before the physical mixer, so physical EQ, faders, crossfader, microphone or auxiliary input, and hardware master processing are excluded. Overflow, tap, disk, and finalization failures retain a .partial recovery artifact that is never announced or scanned as a completed bounce. See DJ Mode, File Formats, and File And Project Problems.
  • DJ pitch ranges cycle per deck as ±8% → ±16% → ±50%. Use [P] for the focused deck or the Pioneer DDJ-REV7 TEMPO RANGE button for its owning deck. A change immediately reinterprets only the owning deck’s physical tempo fader, then saves one global default for the next track load on either deck. Settings adds a DJ → Pitch Range default for later track loads, and persisted 20 values migrate to the REV7-native 16 on disk. On the optional high-resolution jog surface, the platter now receives that owning deck’s live tempo-range slot instead of remaining fixed at ±8, and its BPM uses the same live post-pitch calculation as the TUI. REV7 hardware has confirmed the corrected BPM readout. The controller, TUI, and audio engine now use the same native ±8, ±16, and ±50 ranges end to end. See DJ Mode, Keyboard Shortcuts, Settings, Audio And Playback Problems, and MIDI And Sync Problems.
  • DJ mode now detects beatgrids. Load a bounce that predates beatgrid tagging — or any imported track — and Meiji Sampler analyses its tempo and first beat in the background, writes the grid into the file in Serato’s format, and caches it so every later load is instant. [S] became a real DVS-style SYNC at the same time: it matches tempo and snaps the deck onto the other deck’s beat, click-free, instead of leaving the phase to your nudge. Pressing play on a synced deck now enters on the other deck’s phase too. Detection leans toward the slower tempo reading — an 85 BPM grid on a 170 BPM track still puts beats on beats — and [;] / [\'] halve or double it in one keypress, with [,] / [.] shifting the anchor. A tempo the analyser was unsure about is marked 84.0? BPM so you know to trust your ears. See the DJ Mode guide.
  • Redesigned the DJ tab around a horizontal-waveform layout. Both decks’ waveforms now run full width in their own deck colors and scroll under a fixed center playhead, with a beat ruler carrying bar numbers on every downbeat and hot cues marked the way chop markers are in the trim view — the cue number running as a colored column down the full height of the deck. A phase-match strip between the two waveforms draws deck A’s beats above the line and deck B’s below, turning white where they land together — beat-matching becomes something you see rather than count. The deck blocks gained hot-cue chips tinted in each cue’s own color, a full-track overview strip with the played portion dimmed, and a VINYL badge alongside SYNC and KEYLOCK. Library BPMs are tinted by how reachable they are from the focused deck — green for a straight match inside the pitch range, amber at half or double time, grey when out of range. New [Z] / [X] keys zoom the waveform window between 1.5 s and 24 s, and a taller terminal now spends its extra rows on the waveform instead of the track list. See the DJ Mode guide.
  • Standardized Admin and Creator Studio forms on polished Flowbite controls. Pack Description and Supporting copy fields now provide a rich text toolbar while saving Markdown, and public redemption pages plus every preview that displays pack copy render it safely. Publish readiness is now an actionable five-step flow that opens the exact unfinished field. Existing plain-text pack copy continues to work without conversion. See Sample Cards And Packs.
  • Added a shared Flowbite pack management workspace for creators and administrators. It covers complete pack CRUD, customer experience copy and accent customization, artwork/audio/ZIP persistence, ordering and primary-file selection, card operations, publish readiness, durable audit history, archive versus guarded deletion, retryable storage cleanup, and read-only previews of all seven customer redemption states.
  • Removed the Auto Pitch, Vocoder, and Harmony entries from the PerformFX Library. These SP-404MKII vocal input FX were riding placeholder engines that had nothing to do with the vocal processing their names promise; they will return if and when real pitch-tracking and vocoder engines ship. Existing pad assignments referencing them show as MISSING and can be reassigned.
  • Added a high-resolution jog surface for the Pioneer DDJ-REV7, on by default behind dj.hi_res_jog in config.json. It gives scratch-grade jog resolution — 10,080.7 encoder ticks per platter revolution, streamed up to the device’s 1,000 messages-per-second cap, instead of the public surface’s coarse direction ticks at 19 to 131 reports a second — and motorized platter drive: playing a deck spins its physical platter and pausing brakes it, with about a second of coast-down. A REV7 therefore scratches at full resolution the first time it is plugged in; the surface is chosen at connect time, so setting dj.hi_res_jog to false to keep the controller’s own idle jog display takes effect on the next replug. Meiji Sampler also attempts to drive that surface’s platter waveform and live post-pitch BPM through the cosmetic HID endpoint; if it cannot open, the display may remain blank without affecting control or audio. Physical display behavior is not yet signed off. See the DJ Mode guide.
  • DJ decks now have a silent lead-in ahead of the track, the way a record has groove before the music: wind back past 0:00 and the deck clock counts negative for 1.8 seconds — one full platter revolution at 33⅓ RPM — with the deck silent. That runway is what lets you cue a drop properly. Hold a baby scratch going in the silence while you wait for the “one”, then let go: the deck counts in and the track begins at exactly 0:00 every time. Previously the playhead stopped dead at the first sample, so any forward motion played immediately and the only way to cue was to hold the record still and release it. Scratching off the end of a track now runs into silence too, instead of the last sample held flat. See the DJ Mode guide.
  • DJ-mode scratching and fast varispeed playback are now band-limited to studio standards — the deck read adapts to platter speed and filters the frames it traverses through a windowed-sinc kernel, pushing the residual aliasing haze about 50 dB further down than a plain average, so hauling the jog wheel or pitching a deck up stays clean instead of turning harsh and digital. Platter velocity is tracked just as smoothly: each jog update now accelerates the record evenly across the interval it measured, so speeding up glides in pitch instead of climbing in 30-per-second steps. Normal-speed playback is unchanged.
  • Added a feature-flagged DJ tab preview that turns your Bounces folder into a performable library: every bounce appears automatically with its BPM, and loads onto two decks with pitch, nudge, hot cues on [1]–[8], tempo SYNC, and keylock. Bounces carry Serato-native BPM, beatgrid, and cue metadata, so the same files open in Serato DJ with everything intact. Pioneer DDJ-REV7 support is plug and play for controls, LED feedback, jog motion, and hardware-mixer audio routing, with deck A on channels 1/2 and deck B on 3/4 so the physical faders, EQ, and crossfader do the mixing. See the DJ Mode guide.
  • Physical sample cards: redeem a card on the web and the pack appears in Your Packs in the app.
  • Added a Turntablism PerformFX reference with an overview, 20 effect pages, all 101 preview permutations, exact TTM diagrams, and synchronized accessible ahh.wav / fresh.wav players.
  • Corrected the Turntablism motion transcriptions so Twiddle, Prism, the Tear family, and the chirp half of Joe Cooley use their authored constant-speed lines instead of sine or acceleration curves. Abrupt linear reversals now produce a short phantom micro-stop. FX Params adds Humanize (0%–100%, technique-specific defaults from 10%–30%) to scale bounded cue, travel, and back-pull variation around the exact base curve. Pattern-dependent rows now follow the selected stroke, old saved Pattern indices remain stable as the catalog grows, and older scratch control sets gain the current Gate, Swing, Rate, and Humanize schema when opened.
  • Added two Turntablism strokes from a turntablist deep-dive — Prism (two constant-speed rub cycles phrased as one gesture: each push split in two by a dividing click, the peak reversals muted chirp-style, the pulls riding fully open — 1-2-3 | 1-2-3) and Bubble (multiple fingers tapping the playing record in quick succession, a dense run of near-instant pauses over a natural-pitch rub — the watery gurgle) — plus the constant-speed Forward Tear (2F-1B) and Clover Tear (2F-2B) as Pattern-choice strokes alongside the classic triplet Tear. Record pauses across the tear/tap family use explicit Holds with pop-free phantom edges, and Pats became the slow deliberate sibling it really is: a transform-like drag patted to a halt on the quarter notes, defaulting to a 1-bar cycle. Every scratch slot’s Pattern choice lists twenty-seven TTM strokes.
  • Added a Swing knob to every Turntablism scratch slot (0–1 in FX Params). Swing warps pairs of stroke cycles into the pocket — the first stroke of each pair stretches while the second compresses, up to full triplet swing — and because a scratch is a physical record motion, the pitch follows: the stretched stroke reads slower and lower, the compressed one faster and higher, exactly like a swung hand. Straight playback is untouched at 0; the Transformer keeps its own Swung gate mode instead of the knob.
  • Redesigned the Transformer as the real closed-fader technique: a very slow half-speed drag (a low growl spanning two divisions) with the fader closed at rest and chopped open on the tempo grid like a hard tremolo. The chops lock to the track’s subdivisions — the Rate control picks 8th or 16th chops — and a new Gate choice in FX Params selects the rhythm: Straight even chops, Swung off-beats delayed to triplet swing, or a Rhythmic funky 16th figure. Preview buttons carry the Swung, Rhythmic, and 16th-gate variants alongside the division sweeps.
  • Redesigned the crab scratches as a head-justified family with selectable finger counts, tuned from turntablist field notes. The clicks spring off the thumb in a fixed roll at the head of a constant-speed push — one 1/8 note of the track’s tempo, clamped to the playable finger-spring range (~200–420 ms) — dividing it into evenly-spaced sound onsets with every click re-opening exactly on the next onset, the final sound riding open for the rest of the stroke, and the pull-back heard fully open like a flare. The old Crab preset is now 4-Finger Crab (four clicks, five sounds — the Q-bert fader bounce), a new 3-Finger Crab preset joins it (three clicks, four sounds), and Twiddle is now the two-finger crab (two clicks, three sounds — the delayed two-click flare) built on the same linear triangle. All three default to a 1/2 division so the roll is followed by an audible ride-out, each crab preset’s preview buttons offer the other two finger counts, and the FX Params Pattern choice lists all twenty-seven TTM strokes.
  • Added nine new Turntablism scratches as first-class library presets, each with its own card in the Turntablism group. Four are faderless techniques shaped by the record hand instead of the crossfader — Uzi (a rapid forearm-tension muscle-spasm vibration, the fastest tremolo-like scratch, with a shallow amplitude buzz), Lazer (an extreme-fast single-finger fling, a bright rising sweep and return, fader open), Hydroplane (a moist finger glide whose stick-slip friction bounces into a bubbly amplitude texture over a slow record), and Pats (one-hand phasing — tapping the record to halt its momentum mid-stroke for “clickless clicks”, no fader). Five are advanced combos of motion and crossfader — Boomerang (a syncopated forward-slice, forward-dice, reverse-slice, reverse-dice run), Autobahn (the busiest combo, a six-note compressed micro-grid run with a couple of clicks), Joe Cooley (a fast electro scratch alternating a constant-speed chirp with an open swirl), Twiddle (a two-finger 2-to-3-click flare), and Aquaman (an off-beat undulating texture from asymmetrical off-center flare/orbit clicks). To power the faderless family the scratch engine gained a vinyl-hand amplitude-modulation track that multiplies alongside the phantom click and the crossfader as a third, independent gain. Every scratch slot’s Pattern choice in FX Params lists all twenty-seven TTM strokes, and each new preset ships division-sweep and kin-stroke preview buttons like the rest of the group.
  • Removed the PerformFX CLAP library tab and its non-functional A/B preview cards. The audition lane, scoring/notes, and feedback export never shipped as working features; effect-quality work now lands directly in the built-in engines instead.
  • Added a Turntablism group to the PerformFX Library: eight real scratch presets — Baby Scratch, Chirp, Flare, Orbit, Crab, Transformer, Stab, and Scribble — built on a new engine that reproduces DJ scratch technique as a record-position curve with a crossfader gate, the exact model of the Turntablist Transcription Methodology (TTM). Each stroke’s pitch follows the hand’s speed, the record naturally micro-silences at every direction change (the TTM “phantom click”), and the crossfader chops are click-free. The patterns carry their inventors’ lineage: the Forward cut (G.W. Theodore, popularised by Grandmaster Flash), the Transformer (DJ Spinbad; Jazzy Jeff and Cash Money), the Flare (DJ Flare, popularised by Q-bert), and the Crab and Orbit (Q-bert). Strokes phase-lock to the grid so cycles land on the beat, and each preset now rubs the exact grid slice engaged — the record is cued to the single division that just played and held fixed there, so one stroke covers one division of material instead of creeping across the following beats. [1]–[5] preview buttons sweep the stroke rate per preset (Chirp adds Forward and Military pattern buttons; Stab adds Drag), and every scratch slot exposes a Pattern choice in FX Params that can switch to any of the 27 TTM strokes.
  • Refined the Turntablism scratch engine for more realistic hand feel. Humanize can scale the ~7% faster back-pull, cue drift, and stroke-length drift down to exact geometry. The crossfader cut-in is direction-dependent (a fast ~1 ms open to catch the transient, a slightly slower ~2 ms close to stay click-free), while every read stays locked to the seeded grid slice. Baby, Flare, Orbit, and Twiddle add a 16th Rub preview button; the Rate choice changes one complete rub per division to two, increasing reversal rate and pitch.
  • Every tape-stop-backed preset now exposes a Brake Curve control in FX Params, so you can dial the shape of the pitch dive. Low values give a hard, punchy snap; the middle is a constant-torque vinyl brake; high values give a long, lazy reel spin-down. Tape Stop keeps its eased tape curve by default, while Motor Kill, Vinyl Brake, the SP-404MKII Stopper, and the RMX-1000 Vinyl Brake keep the faster vinyl curve — so untouched presets sound exactly as before. Curve edits take effect on the next stop rather than jumping an in-flight brake.
  • Overdrive, Distortion, WrmSaturator, and Gt Amp Sim are now a real drive family instead of the vinyl simulator they all shared, so each has its own voice. Overdrive is a soft, warm push; Distortion is harder with a tight low end; WrmSaturator spans up to 48 dB of drive with independent ±24 dB EQ LOW/HIGH shelves; Gt Amp Sim runs a two-stage amp voicing with a BASS/MIDDLE/TREBLE stack and a six-way AMP TYPE (JC, TWIN, BG, MATCH, MS, SLDN) — amp-voiced approximations, not circuit models. All four use antiderivative-antialiased shaping so hard drive stays clean instead of aliasing, and each is loudness-matched so engaging it does not jump the level.
  • Back Spin (SP-404MKII and RMX-1000 Release Back Spin) now wheels the record backward from the current beat instead of playing a forward tape stop. It snapshots the audio that just played and spins it in reverse — accelerating from a standstill up to a fast reverse spin — with SPEED setting how hard the spin is and LENGTH riding the division. The spin-up and terminal stop are click-free.
  • Cassette Sim is a dedicated worn-cassette engine instead of a flanger reroute. Its wow/flutter now warbles the pitch of the whole signal (rather than a comb-filtering delay mixed against the dry), with TONE, HISS, AGE (0–60 years of high-end loss), DRIVE, WOW FLUT, and a CATCH control for occasional tape-stretch pitch sags.
  • Tremolo/Pan is now a smooth cyclic LFO tremolo/panner instead of a hard duty-cycle gate. TYPE picks TRE (equal-power amplitude) or PAN (equal-power left/right that preserves the stereo image), WAVE offers TRI/SQR/SIN/SAW1/SAW2/TRP with click-free edges, DEPTH sets the amount, and RATE runs free or tempo-locked with SYNC.
  • Ring Mod is now a true ring modulator (input × carrier) instead of a standalone siren oscillator, so it makes the input metallic and clangorous. FREQUENCY sets the carrier (30 Hz–4 kHz), SENS lets input level push the carrier frequency, POLARITY flips that push, BALANCE mixes dry/wet, and EQ LOW/HIGH shelve the modulated signal (±15 dB).
  • Slicer now runs a 16-step pattern instead of a plain duty-cycle gate: PATTERN selects one of twelve curated grooves, DEPTH sets how far off-steps are cut, SHUFFLE delays the even steps, and MODE picks LEGATO (adjacent on-steps merge) or SLASH (every step re-attacks). Existing gate presets keep their duty-cycle behavior.
  • Zan-Zou is now a multi-tap echo: MODE picks 2TAP, 3TAP, or 4TAP, tapping the delay line at successive intervals with alternate taps phase-inverted and each successive tap darker — the short answering-echo texture of the hardware. Every other delay preset stays single-tap.
  • Downer and To-Gu-Ro are now cyclic speed effects instead of a one-shot tape stop and a flanger. Downer ramps the playback speed down each RATE period then snaps back; To-Gu-Ro undulates the speed sinusoidally like a coiled spring. Both add a resonant tracking filter (FILTER / FLT MOD, RESONANCE) and, for To-Gu-Ro, an AMP MOD that dips the level with the speed; Downer’s PITCH ON holds the pitch steady while the filter and level still cycle. Read-position lag is crossfaded back to the live head each cycle so it never drifts.
  • Crusher and Lo-fi are now real digital degradation instead of a vinyl simulator. Crusher runs a pre-filter, genuine sample-rate reduction (higher RATE holds each sample longer for a coarser, more lo-fi sound down to roughly 2 kHz), and a dry/wet BALANCE inside the effect. Lo-fi adds a nine-step LOFI TYPE ladder that pairs progressively lower sample rates with lower bit depths, a TONE tilt, and a post-CUTOFF low-pass. Both replace the old Vinyl Sim reroute that only added saturation and flutter.
  • The SP-404MKII Equalizer is now a real three-band EQ (low shelf, mid bell, high shelf, ±15 dB per band) instead of a plain low-pass, so it can boost and independently shape lows, mids, and highs. The Isolator and RMX-1000 Cut/Add ports share the same engine with a deep-cut “kill” range, so a band can be removed entirely.
  • Wah is now a real swept resonant filter instead of a static high-pass — the cutoff wobbles cyclically like a wah-wah pedal. PEAK narrows the band for more quack, MANUAL sets the centre frequency, DEPTH sets the sweep span, RATE sets the cycle (tempo-locked when SYNC is On), and FLT TYPE picks a wide LPF or narrow BPF. Super Filter now cyclically sweeps too, with an extremely sharp two-stage slope and LPF/BPF/HPF types — previously both just sat still. Both sweep in log-frequency space, so the sweep is symmetric in octaves above and below the centre.
  • Chorus and JUNO Chorus are now true dual-line BBD-style choruses instead of flangers. Each channel runs its own modulated delay with no feedback, so there is no stationary comb filtering — the old routing “sounded like a flanger” — and the two channels sweep in anti-phase for a wide stereo image (the way the Roland JUNO-106 inverts one of its two chorus outputs). Chorus exposes DEPTH, RATE (0.33–2.30 s cycle), BALANCE, EQ LOW/HIGH shelves (±15 dB), and a NOISE control for bucket-brigade hiss; JUNO Chorus models the JUNO-106 / JX sections with a five-way MODE (JUNO 1, JUNO 2, JUNO12, JX-1 1, JX-1 2), NOISE, and BALANCE at the hardware’s fixed mix. The internal DP4 Lush, Stereo Chorus, and Ensemble Wash presets moved onto the same real chorus engine, so their lush widening survives; Daft Flanger stays a flanger.
  • Phaser is now a real allpass-cascade phaser instead of a flanger. It sweeps four allpass stages to move notches that are not harmonically spaced (a phaser, not a swept-comb flanger): DEPTH sets the sweep span, MANUAL the centre frequency, RESONANCE the feedback emphasis, RATE the cycle (tempo-locked when SYNC is On), and BALANCE the mix. The plain Flanger is unchanged.
  • Differentiated the SP-404MKII and RMX-1000 hardware-port presets that share an engine. Every delay/echo port now has its own feedback and repeat-brightness voicing (clean Sync Delay, echo-y Ko-Da-Ma, tight dark Zan-Zou, smooth TimeCtrlDly, band-colored BPF/HPF/LPF Echo, rising Spiral Up vs. sinking Spiral Down, washy Cloud Delay), the vinyl/saturation ports (303 vs. 404 VinylSim, WrmSaturator, Cassette Sim, Overdrive, Distortion, Gt Amp Sim) carry distinct character defaults, and SX Reverb, Reverb Up, and Reverb Down get their own sizes and decays — previously all presets sharing an engine sounded identical.
  • Retriggering an Airwindows-backed reverb or echo preset (Room Verb, Hall Throw, Cathedral, Air Raid Echo, and the other space and delay patches) no longer bleeds the previous trigger’s tail under the new attack: the effect is now fully reset on each trigger, so it starts from silence instead of replaying a leftover reverb wash or echo tail.
  • Improved the internal reverb engine’s stereo and tail behavior: each channel now diffuses through its own allpass pair for a wider, more decorrelated stereo image instead of sharing one mono diffusion, the ring-out tail fades smoothly to silence at maximum decay instead of hard-cutting mid-decay, and the comb feedback path is protected against denormal CPU spikes as tails fade out. Retriggering the reverb no longer allocates on the audio thread.
  • Smoothed the Rhythmic Gate edges: gate open/close transitions now use raised-cosine ramps instead of linear ones, so gating bass-heavy material no longer “ticks” on each edge. Very short gate-open windows fade in and out cleanly instead of clicking, and both stereo channels of a frame now receive the same gate gain so odd (triplet) divisions no longer flip the stereo image.
  • Engaging Vinyl Sim (and the SP-404MKII vinyl, cassette, and lo-fi character ports built on it) no longer pumps loudness or dips pitch: the saturation stage used to add up to ≈ +7 dB the moment the effect came on even with Drive at zero — it now passes quiet material at unity, with Drive 0 a true clean bypass — and the wow/flutter wobble no longer deepens with the trigger envelope, so the attack ramp no longer bends the program flat on every press. Vinyl hiss is also shaped darker (rolled off ~4 kHz) and mixed in before the high-frequency rolloff, so it sits inside the program like record surface noise instead of floating brightly above it.
  • Vinyl crackle sounds natural instead of mechanical: every retrigger now produces a fresh click pattern rather than replaying the identical one, clicks keep the same brightness and duration on any audio device sample rate (previously their energy sat at the device Nyquist frequency — nearly inaudible at 96 kHz), and each click’s resonant tail rings out instead of freezing mid-ring. This also covers Dusty Vinyl’s Crackle control.
  • Fixed a buzz in SP1200 45RPM mode: the pitch chain restarted its interpolation phase on every audio block, adding a faint block-rate discontinuity buzz on top of the program — the resampler now runs continuously across blocks. SP1200 style changes and retriggers no longer allocate on the audio thread, and pooled SP1200 clones keep the configured device sample rate instead of resetting to 44.1 kHz.
  • Filter Sweep presets (S950 LPF/HPF gestures) now glide the cutoff with the same 5 ms log-domain smoothing as the mixer filters, removing zipper transients from fast resonant sweeps.
  • Changing the ASR10 Boost Level while the effect is running no longer clicks: the level switch keeps the saturation state instead of resetting it, matching the mixer-side behavior.
  • Fixed Echo/Delay timing at low tempos: bar-length echoes (2-bar and 4-bar delays below 120/240 BPM) were silently clamped to a shorter time and landed off-grid — the delay buffer now holds the full 8 seconds so those echoes stay locked to the beat. Repeats also darken as they regenerate, rolling off the highs a little more on each pass like an analog tape delay instead of staying bright and digital, and a DC blocker keeps the feedback loop from drifting. Triplet (odd) delay divisions no longer swap the stereo channels on each repeat, retriggering an echo with the same time now carries the ringing tail through the new trigger instead of cutting it to silence, and changing the delay time fades the old repeats out cleanly.
  • The Flanger no longer colors the dry signal: engaging it used to gently compress the whole mix even at zero depth, causing a small level step when it turned on — the output limiter now applies only in proportion to the wet amount, so the dry program passes through untouched until you dial the effect in. Its modulated delay tap also uses smoother 4-point interpolation, reducing high-frequency flutter on the sweep.
  • AutoPan now preserves the stereo image: the panner used to collapse to mono internally, so anti-phase stereo material went silent at full wet — it now applies the equal-power pan gains to the left and right channels directly, keeping mono-and-centered material bit-transparent while widened content survives the sweep.
  • Motor Kill is now a real deck brake instead of a quick volume duck: triggering it pitch-dives the master to silence over about 200 ms with a constant-torque vinyl curve — like cutting the power on a turntable — and releasing brings the program back through the normal envelope. Its FX Params Target control is gone with the reroute, and its library preview buttons now offer timing variants (Brake, 1/4 Brake, 1/2 Brake, Bar Brake). Digital Blackout keeps the instant hard-mute behavior and its Target control.
  • Rebuilt Halftime playback quality: the slowed signal now uses the same 4-point interpolation as Tape Stop, so half-speed playback is smooth instead of stair-stepped, loud material no longer overshoots at higher Speed settings, and a speed-tracking low-pass rolls off digital imaging so the slowdown sounds analog. Resync jumps and the phrase wrap now use click-free raised-cosine crossfades, and Beat Repeat’s Feedback LPF now genuinely darkens each repeat cycle — every wrap gets progressively warmer, like a tape echo regenerating — instead of filtering the loop only once. Triplet divisions no longer swap the stereo channels on loop wraps, and the Hype Button’s duck timing is now identical at 44.1 kHz and 48 kHz with no level pop on rapid re-presses.
  • Rebuilt Tape Stop and Vinyl Brake so the brake reaches a true terminal stop — the playback rate now falls all the way to zero and dives faster near the end instead of crawling at a floor. Both presets continue seamlessly from the moment you trigger them rather than replaying older audio, roll off highs naturally as they slow so the dive sounds like analog tape or vinyl instead of digital aliasing, and fade the final stop to silence with no click. Tape Stop keeps its eased reel spin-down and longer release; Vinyl Brake uses a constant-torque brake curve that dives more sharply into the stop.
  • Fixed the first [SPACE] press after launch so cued loops start playing as soon as synced audio finishes warming. Previously the pending transport start was dropped a moment after the press, so nothing played until [SPACE] was pressed a second time.
  • Made channels with stem separation ready faster after launch: cached stems now load without re-running audio analysis, and startup logs distinguish loading cached stems from running Demucs, so a cache hit is no longer reported as a new separation.
  • Replaced the silent FX Library quick-assign with a PerformFX Assign dialog. [A] on a focused preset now opens a modal with three [←/→] selectors — Effect, Preset, and FX Slot — plus an informational pad list. Effect cycles assignable FX Library effects and opens on the active pad’s assigned effect when present; Preset cycles that effect’s labeled variants from the [1]–[6] preview buttons; FX Slot picks Bank A or Bank B. Move focus between Effect, Preset, FX Slot, and Pad with [↑/↓], and change the focused selector with [←/→]; [←/→] is a no-op on Pad. Pad keys [1]–[0] assign the chosen variant to that pad of the selected bank right away — the first pad appears on the Pad: row, an empty pad shows [1] - Empty, and assigned pads keep their dimmed key visible with the assignment label ([1] SP1200: Chan1&2, or SP1200: Custom after edited settings). The dialog stays open so you can place several variants in one visit; pressing an occupied pad’s key does nothing. [ENTER] and [ESC] close the dialog. Each assignment shows a status line such as Assigned "King Tubby" to Bank A pad 3.
  • Rebuilt the PerformFX dub siren tone generator around classic siren hardware architecture: band-limited Square and Saw oscillator waveforms join Sine and Triangle, the pitch LFO now offers six shapes (Sine, Square, Ramp Up, Ramp Down, Stepped, Sine+Oct) with analog-style glide between steps, a one-shot pitch Sweep (± semitones over an adjustable Sweep Time up to 10.0 s) retriggers on every fire, and the output passes a soft-clip tone stage for hardware-style warmth. Classic Dub Siren, Air Raid, and Robot Beep were retuned around the new engine — Air Raid now winds up like a real siren and Robot Beep chatters through a stepped staircase — and four new sirens joined the SP Series group: Bomb Drop, Laser Zap, Two-Tone Horn, and Steppers Siren.
  • Restored the terminal immediately on quit and shortened telemetry flush waits so closing the app no longer pauses in the alternate screen.
  • Fixed Windows installer updates so startup checks and Settings → About → Updates use the native WinSparkle dialog with the correct Windows feed.
  • Redesigned the Perform tab around the shared audio waveform overview, stacked Bank A and Bank B ten-slot rows mapped to [1] through [9] and [0], a group-colored FX Library with per-preset preview buttons, [A] assignment via the Assign dialog without leaving the library, a focused inspector, and FX Params on [ENTER]. The first Vintage previews use [1] through [4] as A/B/C/D variants for SP1200 channel pairs, S950 LPF/HPF cutoff points, and ASR10 Boost levels; DJFX Looper uses [1] through [6] for 1/2 through 1/64 grid captures.
  • Split the PerformFX Library into grouped boxes: Vintage, DJ FX, SP Series, Time, Space, RMX Series, Utility, and Custom when present.
  • Added a Vintage PerformFX Library group for SP1200, S950 LPF, S950 HPF, and ASR10 Boost, including SP1200 filter/45RPM metadata, S950 frequency ranges, and ASR10 Lo/Mid/Hi boost levels in the inspector debug notes.
  • Added shared PerformFX One Shot, Length, PreGain, and Wet controls plus effect-specific FX Params so every slot can auto-release after a dedicated total engagement duration, show the remaining one-shot time while active, trim or boost the signal before processing, set a dry/wet ceiling or use the preset default, reopen with the saved values, reset with [R], and apply those saved parameters when the slot triggers. Time-based one-shot Length now reaches 20,000 ms. Gain-heavy presets default to -6 dB, color/saturation presets default to -3 dB, and Lofi Repeat, S950 LPF/HPF, SP1200, ASR10 Boost, Vinyl Sim, Halftime, dub echo, reverb, gate, flanger, AutoPan, dub siren, duck, and utility slots can save their musical controls.
  • Fixed one-shot echo and delay slots so completing Length cuts the tail and returns the slot to READY instead of lingering in a phantom RELEASE state.
  • Added Engage: Always On for PerformFX slots that should stay active automatically; assigned Always On slots activate without waiting for a trigger, ignore trigger/release presses and One Shot auto-release, and stay active until the slot is cleared or changed.
  • Added [i] on PerformFX to show or hide the inspector, keeping the FX Library full-width until detailed focused-slot or library metadata is needed.
  • Set every built-in PerformFX Library preset’s default engagement to Smart Hold. Fixed slots and custom presets still keep any explicitly saved Hold or Toggle setting.
  • Renamed the FX Params Timing column to Grid. Grid divisions run from 1/64 through 4 Bars, including 1/16T, 1/8T, 1/4T, and 1/2T.
  • Added Grid as an Attack and Release value when Sync is Tempo (beats), so envelopes can follow the slot’s current Grid value while [-] and [+] adjust that slot from the Perform tab.
  • Filter Sweep Up and Filter Sweep Down now default to Tempo (beats) with Attack set to Grid and Release set to 1/4. Cap Sweep keeps a 1 Bar Attack/Release sweep and now uses the internal resonant LPF sweep path instead of a fixed Airwindows filter patch.
  • DP4 Lush, Daft Flanger, Stereo Chorus, and Ensemble Wash now use internal modulation engines instead of the generic Airwindows insert mapping — DP4 Lush, Stereo Chorus, and Ensemble Wash on the real dual-line chorus engine, Daft Flanger on the flanger. AutoPan now uses an internal equal-power pan engine with editable Rate and Depth, so it moves clearly without a dry-level drop.
  • Room Verb is now a subtler short-room preset, and Room Verb plus Hall Throw now ring out naturally after transport stop by stopping new input into the reverb while the existing tail drains.
  • Bright Room and Infinite Space now use the controlled internal reverb engine instead of the older multi-stage Airwindows chains. Bright Room is a more audible short ambience, and Infinite Space uses a lower-gain large wash to avoid clipping.
  • Gate Filter and Digital Gate now use the internal Current Grid gate engine so they produce reliable rhythmic chops instead of depending on input-threshold Airwindows gate settings. Beat Surge is stronger and no longer defaults to -6 dB PreGain, while Block Limiter uses a lower Pound setting with a parallel wet default for a more musical limiter gesture.
  • Digital Blackout now uses the internal full-kill path so it is an obvious hard mute, Tape Stop now uses a longer release so playback eases back in after the slowdown, and Vinyl Brake defaults to 1/4 with a silent stop floor instead of aliased near-stop playback.
  • Halftime now captures and wraps the active timing division with a smoothed phrase boundary, and transport stop clears active PerformFX so captured effects do not continue after playback stops.
  • Added SP-404MKII and RMX-1000 hardware-port presets to PerformFX with the original effect names and editable hardware-style custom parameters.
  • Rendered live PerformFX gestures into master WAV bounces and visualizer video audio while keeping individual channel stem exports pre-master-FX for clean mix handoff.
  • Renamed PerformFX Stutter to DJFX Looper, marked it with ★ as the recommended loop-repeat path, kept its default engagement on Smart Hold, and kept its post-trigger capture loop behavior. Lofi Repeat keeps the existing lookback-repeat behavior.
  • Added PerformFX Current Grid quantize. [q] now cycles the active bank’s visible Quantize Bank row from Default → Current Grid → Next Grid → Next Bar → Next 2 Bars. DJFX Looper defaults to Current Grid so it can finish the active grid cell live before looping the completed cell from its grid start, and Build Gate and Half Gate default to Current Grid so their gate phase locks to the active cell instead of restarting at the press point.
  • Renamed per-slot PerformFX quantize to Quantize Start and added Quantize End for manual releases. Quantize End uses the same Off, Current Grid, Next Grid, Next Bar, and Next 2 Bars values, keeps the slot active while the release is waiting, and is hidden when One Shot is on because Length controls one-shot end timing.
  • Added a Gate Offset FX Param for shifting Build Gate and Half Gate phase while staying locked to the active grid.
  • Fixed PerformFX grid timing so Current Grid and queued Next Grid triggers use the active loop transport start as the grid origin.
  • Added an inverted tile flash when a queued PerformFX trigger becomes active, retriggers, or advances into a new loop cycle, plus QUEUED: 1000ms countdowns for slots waiting to activate and ACTIVE: 1000ms countdowns for active slots waiting for a quantized release.
  • Fixed MIDI-mapped PerformFX Hold slots so Note button releases and CC falling edges release the effect instead of retriggering it on the next tap, and made PerformFX MIDI Learn slots bank-specific with all Bank A slots listed before Bank B slots.
  • Added named PerformFX FX Library targets to MIDI Mapping, including custom presets. These trigger the selected library item directly with its current settings instead of requiring a fixed slot assignment.
  • Fixed PerformFX wet mix routing so delay presets such as Tape Echo use their preset default or slot wet override when triggered, instead of starting as full-wet and briefly dropping the dry signal.
  • Fixed Airwindows delay dry-through so Tape Echo, Space Echo, Pitch Echo, Four Tap Echo, and Double Delay keep the live dry signal at unity while layering the delayed return.
  • Added a PerformFX Tail: Cut / Ring Out setting for tail-capable presets. Echo, delay, throw, and reverb presets default to Ring Out, so stopping transport lets their tails decay instead of hard-cutting them; echo tails now retire once a full delay cycle falls below the inaudible threshold, avoiding unnecessary silent time at the end of bounces. Gates, repeats, brakes, ducking, filters, and utility gestures hide Tail and keep the Cut behavior.
  • Added an echo-only PerformFX Throw setting. Set it from 1/64 through 4 Bars to feed only that tempo window into Dub Echo, Tape Echo, Space Echo, and related delay presets, then let wet-only repeats ring without interrupting the playhead. Manually releasing a thrown echo restores the dry signal and cuts the thrown tail.
  • Added Active FX Debug Notes for active or queued PerformFX presets, and fixed FX Params preset browsing so deep preset selections stay visible while scrolling.
  • Fixed the [M] shortcut so it opens the metronome controls dialog across terminals that report shifted letters differently.

​
Meiji Sampler v1.2.0

v1.2.0 is a workflow release focused on recording tempo, metronome/count-in control, project browsing, audio setup, and more reliable loop sync.

​
At a glance

  • 1. Recording tempo overhaul. Fixed BPM recording, metronome, count-in, First Loop Setup, and Project Tempo management now form one clearer first-loop workflow.
  • 2. Project browsing and audio setup. The project browser is searchable, supports scrolling through larger libraries, and Audio settings add selectable stereo channels, buffer latency labels, and restart prompts.
  • 3. Reverb, PerformFX, and loop sync. Reverb is massively improved, Airwindows-backed PerformFX presets are available, and loop sync is more stable across loading and loop length edits.
  • 4. Recording, loading, and polish. Trigger recordings start at the threshold crossing, analysis work is cached, startup waits for the splash before login, and mixer warmup messaging is clearer.

​
1. Recording tempo overhaul

Meiji Sampler’s first-loop workflow has been rebuilt around fixed BPM recording, metronome, count-in, and stronger Project Tempo management. First Loop Setup gives the first take an explicit tempo path instead of making every recording depend on an inferred loop length. The metronome and count-in controls now support more intentional recording starts, including fixed BPM takes and clearer count-in behavior. Project Tempo management has also been tightened so tempo decisions made before the first recording flow into the loop workflow consistently.

​
2. Project browsing and audio setup

The project browser is now searchable, no longer limited to a fixed project count, and can scroll through larger project libraries. Audio setup is clearer for multi-output and latency-sensitive setups. You can select stereo channels, see calculated buffer latency beside buffer size, and use an in-app audio engine restart prompt after changing settings that require the engine to restart.

​
3. Reverb, PerformFX, and loop sync

The Reverb algorithm has been massively improved, and PerformFX adds Airwindows-backed presets for more immediate performance effects. Loop sync is more reliable across common tempo workflows: first loops can snap to a looped anchor tempo, sample sync stays stable across loop length edits, Loop: ON samples can auto-stretch on load, and Sync warmups debounce stale work.

​
4. Recording, loading, and polish

Record In now defaults to Threshold mode, and Trigger recordings start at the threshold crossing so the saved take begins at the intended transient. Transient analysis results are cached, samples avoid unnecessary reprocessing on app load, the login modal waits until the startup splash finishes, and mixer warmup status now communicates the waiting state more clearly.

​
Fixes

  • Fixed neutral pad playback tone.
  • Fixed loop auto-detection by ignoring files that are too short to produce useful loop candidates.
  • Fixed MIDI keypress mappings so shifted letters stay distinct from lowercase shortcuts, and [ESC] can be captured as a MIDI keypress target. Existing lowercase-stored letter mappings now display their true lowercase label, such as [m], while new shifted mappings display as [M].
  • Fixed FX Params option rows so [ENTER] opens value editing, arrow keys change the selected value, [R] resets the current target slot editor to defaults, and [ESC] returns to row selection.
  • Fixed FX Params closing so [ESC] saves edited slot values, including Quantize changes, instead of discarding them.
  • Fixed FX Params from the FX Library so [ENTER] targets the last focused fixed slot instead of falling back to slot [1], and preserved legacy letter-key PerformFX assignments when migrating to fixed slots.

​
Meiji Sampler v1.1.0

v1.1.0 is a fast-follow hot off the heels of our huge v1.0.0 update. MIDI in & mapping gets a lot of love. Other important updates involve playback reliability, tempo handling, mute and solo behavior, and trim mode.

​
At a glance

  • 1. MIDI improvements. MIDI knob & fader support, device automap, new-device prompts, keypress mapping, MIDI transport controls, per-mapping CC behaviour, and a Mapping column in MIDI Monitor reduce setup friction.
  • 2. Playback reliability. Synced stretch warmup, overdub playback fixes, loop-mode pad playback, multi-output devices, and empty loop-slot navigation all received targeted fixes.
  • 3. Tempo, metronome, settings, and loop workflow. The Tempo Settings page includes first-loop range normalization, metronome count-in/click controls, fixed first-loop lengths when BPM is set, Audio settings show calculated buffer latency, and track solo in the sequence editor now toggles consistently.
  • 4. Trim, radio capture, and input recording. Internet Radio captures begin transient analysis as Trim opens, Trigger recordings start at the threshold crossing, and trim playback is guarded against invalid bounds.

​
1. MIDI improvements

​
MIDI Knob & Fader Support

A Behavior step appears when you map a CC fader or knob. Pick Gate to wire the control to on/off state, or Trigger to fire once when the value crosses a threshold. Adjust threshold with [←/→], use [SHIFT+←/→] for larger steps, and press [I] to invert direction. The Listening list shows each mapping’s behavior as a chip in the Trigger column, such as [GATE@64] or [TRIG@64]. To edit an existing mapping’s behavior, highlight it and press [B] to jump straight to the Behavior step. Existing mappings keep their pre-1.1.0 behavior so projects load unchanged. New CC mappings get defaults based on the target: loops, channel mute, transport, and scene slots default to Gate at threshold 64; pads, cue actions, and keypress targets default to Trigger.

​
Device automap and setup prompts

Meiji Sampler can now prompt you when a new MIDI device appears and guide you into setup instead of making you hunt through Settings. The new device automap flow builds a batch of mappings for common controller layouts from Settings → MIDI → Devices, which is useful when you want pads, transport, loops, or scene triggers configured in one pass. See MIDI And External Gear and MIDI Implementation Reference.

​
Conflict handling, monitor details, and keypress mappings

When a MIDI Learn control or target is already mapped, Meiji Sampler returns to the Listening list, selects the existing mapping row, and shows already mapped. This protects established mappings instead of replacing them from the conflict path. The MIDI Monitor now includes a Mapping column so you can see which incoming events are already assigned. MIDI Learn also keeps properties inline while you assign controls, preserves hotkey styling, and auto-advances through trigger targets for faster pad, loop, and scene setup. Keypress mappings are part of the MIDI action surface in v1.1.0. Map a controller control to a keyboard key and Meiji Sampler routes it through the normal keyboard handler, so it remains context-sensitive like the real key.

​
2. Playback reliability

​
Synced stretch warmup

When a cued loop or scene references sync-stretched audio that is still decoding, loading from cache, or rendering, Meiji Sampler keeps transport cued at warming synced audio… and starts once the stretched buffer is ready. This prevents transport from beginning on time while the stretched channel enters late. After you change Loop values in Channel Detail, closing the modal starts warming the channel’s stretched audio when a session clock is available. Sync multiplier changes now wait one second before warming, show Stretching beside the Sync control once warmup starts, and supersede older warmups when you keep flipping values. Transport still waits if the buffer is cold, but adjusted loops can now finish warming before you press play. Channel Detail now groups Loop, Scene Ends, and Sync under Loop & Stretch, so loop behavior and stretch timing live in one place. Sync values also show the effective post-sync BPM when the loaded loop has detected BPM and a session clock exists. Loaded samples with Loop: ON now receive the best-fit Sync value automatically when a session clock exists. This applies to loops detected on load and to samples that already had Loop: ON from file defaults or copied pad settings. Disable it from Settings → Audio → Stretch Loops on Load, which defaults to ON; turning it back on also applies Sync to loaded loop pads that are ready.

​
Loop and overdub reliability

v1.1.0 fixes several loop-path edge cases: overdub playback beds start correctly, loop-mode pad playback works in loop slots, looped pad playback behaves on multi-output devices, empty loop-slot navigation no longer panics, and inverted trim bounds no longer panic playback progress. Multichannel interfaces can now route the stereo master with the Stereo Channels setting, such as Ch 9 & 10. Scene playback now restarts the metronome clock on the scene downbeat, so the click no longer inherits a stale phase from earlier transport. When exactly one recorded loop is stopped, pressing [SPACE] now starts it immediately instead of showing the Playback needs 1+ cued loop guidance first. During OVR, pressing [SPACE] now commits the overdub and stops transport instead of showing first-loop guidance. With multiple stopped loops, Meiji Sampler still asks you to cue or choose the loop before transport starts. Channel mute transitions are also smoothed, reducing clicks when muting or unmuting during playback. Audio output setting changes now offer an in-app audio engine restart when you exit Settings. You can apply host, output device, buffer size, and sample rate changes without relaunching, or defer them until the next launch.

​
3. Tempo and loop workflow

​
Tempo range normalization

Tempo Settings now include a first-loop Range selector: 65-135, 87-175, 133-300, or None. Out-of-range first-loop detections are doubled or halved into the selected range, which helps keep quantize, synced stretch, and loop timing aligned with your intended BPM.

​
Metronome

Meiji Sampler now includes a recording-focused metronome tied to the effective project BPM. Use [M] to open the metronome controls dialog for count-in mode, click mode, count-in beats, subdivision rate, volume, accent, and fixed first-loop length. In the dialog, use [←/→] to change values, [SPACE] to control transport without closing the dialog, and [ENTER] to accept. A right-side pulse expands on accented beats so the dialog acts as a visual metronome while the click is active. Pressing [R] before the first loop now opens First Loop Setup. The default Fixed mode shows Mode: FIXED TEMPO, Start with a blank loop of fixed tempo & duration, and [SPACE] to tap tempo, [ENTER] to record, records into the selected Length, starts at 90.00 BPM, initializes Quantize from Settings with OFF as the default, enables a 4 BEATS record count-in, and uses a REC + PLAY click. Highlight Tempo to type a BPM directly, or press [SPACE] anywhere in the dialog to tap tempo with an audible click, return to Fixed mode, and focus Tempo. During a Fixed take, the loop details view shows the full selected Length with a moving playhead, and pressing [ENTER] early ends recording while preserving that Length and queuing overdub as OVR+ for the next fixed interval. Empty selected loop slots now show Press [R] to record a Loop as a persistent toast instead of drawing that prompt inside the loop details panel; active recording shows Press [ENTER] to save, [ESC] to retry, and retry returns to the armed pad or count-in prompt. Select Auto with the Mode field to see Record freely, set tempo by ending cleanly, then press [ENTER] to start recording and [ENTER] again to end the loop; Auto keeps the flexible elapsed-time details view while still applying the dialog’s Quantize, Count-in, and Click choices, then immediately enters overdub. Press [ESC] to cancel the dialog without arming. Pad and chop keys that start a record count-in now play immediately. The armed prompt also advertises [SPACE] when count-in is available, so you can start the countdown without playing a pad first. While the count-in runs, a centered numeric toast counts down once per beat and stays on screen for that beat’s length. Pad or chop hits during the final count-in beat are captured at 0:00; earlier hits are consumed as unrecorded start-signal updates. When BPM is set before the first recording, the first loop can be constrained to a clean bar length instead of relying only on the initial performance duration. The click is routed outside bounce, resample, and internal capture paths, so it cannot bleed into exported or internally sampled audio. See Settings Reference and Recording And Loops Troubleshooting.

​
Audio buffer latency labels

Settings → Audio now shows the calculated buffer duration beside Buffer Size, such as 256 samples (5.80 ms) at 44100 Hz. The value updates with Sample Rate so latency tradeoffs are visible while tuning audio settings.

​
Track solo is consistent

Track solo in the sequence editor is now a true toggle. Press [S] to solo the selected track, then press [S] again to un-solo and unmute every track in the loop. This matches channel solo and loop solo. See Muting And Soloing.

​
4. Trim, radio capture, and input recording

​
Trigger recordings start at the threshold crossing

Audio input recording in Trigger mode now starts the saved take at the first frame that crosses the threshold. Armed input is still monitored so the trigger can be detected accurately, but audio before the crossing is not written into the recording. See Sampling From Audio Input.

​
Internet Radio captures analyze transients as Trim opens

Captured Internet Radio buffers start transient analysis as soon as [ENTER] opens Trim. Marker-based snapping and chopping are ready without needing to nudge the playhead first.

​
Trim playback safeguards

Playback progress is now guarded against inverted trim bounds, so malformed or rapidly edited trim regions do not crash the UI while playback state updates.

​
Meiji Sampler v1.0.0

The biggest release of Meiji Sampler ever. 119 changes since v0.16.0, headlined by native MIDI, Splice support, three vintage sampler emulations, an improved timestretching engine, intelligent chopping, three new in-app radio stations, and our first official documentation site. This is also the version that earns the 1.0 badge — production-ready, fully documented, and stable across macOS, Linux, and Windows (including Windows ARM64).

​
At a glance

  • MIDI is here. Full MIDI Learn covers pads, individual chop slices, loops, scenes, mixer mutes, transport, and one-key Keypress mappings. Mappings keep working inside Channel Detail and Chop views. Built-in MIDI Monitor for identifying devices and events.
  • Splice support. Sign in with your Splice account and browse Splice’s growing library of over 2 million royalty-free loops & samples. License and download with a single keypress.
  • Three vintage emulations. SP-1200 Mode (4 styles, refined against a real SP-1200), ASR-10 Boost (3 levels), and rebuilt S950 Butterworth HPF/LPF.
  • New timestretching engine + smarter chopping. Formant-preserving spectral stretch replaces the old engine, plus intelligent spectral transient snap (Shift+[ / Shift+]) color-coded by frequency band.
  • Six new audio formats. MP3, AAC, FLAC, OGG, ALAC, and CAF in addition to WAV/AIFF.
  • Meiji Sampler Radio. Three live 24/7 stations — Drums, Originals, and Originals II — usable as sample sources in-app and online at radio.meiji.industries.
  • Official docs site at docs.sampler.meiji.industries.
  • Vim-style navigation (h j k l) everywhere.
  • Tap tempo, manual project BPM override, and first-loop snapping from looped anchor pads.
  • Linked pad triggers, choke groups, playback overrides, and Sample Offset for layering and humanization.
  • Tabbed Settings with a dedicated Project management category.
  • Optional login — skip sign-in, skip telemetry.

​
Tentpoles

​
1. MIDI, Splice, and Documentation

The integrations release. Three things you’ve been asking for, all shipped together.

​
Native MIDI mapping

The new MIDI tab inside Settings provides a complete mapping workflow over the entire action surface:
  • MIDI Learn — move a control on your device, then navigate a hierarchical menu to assign it to any action
  • Built-in MIDI Monitor — identify devices and inspect incoming events in real time
  • Loop actions — slot triggers, mute, solo, undo, or clear individual slots
  • Scene actions — trigger, cue, or stop scenes
  • Mixer actions — mute individual channels
  • Pad chop mapping — map controls directly to individual chop slices on any pad
  • Transport — play/stop and record/overdub independently
  • Keypress mappings: map a controller control to a normal keyboard key such as 1, b, r, p, Space, or a function key
Direct MIDI action mappings keep working when Channel Detail or Chop overlays are open, so you can trigger pads and control transport without leaving these views. Stateful surfaces (Settings, Bounce, Trim, Search, MIDI Learn) still block direct app-action mappings to prevent mapped actions from dismissing work in progress. Keypress mappings follow the normal keyboard handler after the MIDI control is resolved, so they work in focused contexts such as the Trim/chop editor. See MIDI and External Gear and MIDI Implementation Reference.

​
Splice support

Sign in with a Splice account directly from the Splice browser and browse Splice’s growing library of over 2 million royalty-free loops & samples — without leaving Meiji Sampler.
  • Navigate with filters, natural-language search, and preview
  • Automatically license + download sounds with one keypress — Enter licenses and loads the selected result at full quality
  • Licensed files live on disk by Splice sample id with a license sidecar, next to your recordings for unified file management
  • Long samples open directly in Trim for fast chopping
See Splice Integration.

​
Official documentation site

Meiji Sampler now has a real documentation site at docs.sampler.meiji.industries — guides, recipes, troubleshooting, full keyboard shortcut reference, and a release notes archive. Built on Mintlify and updated alongside every release.

​
2. Vintage sampler emulation

Meiji Sampler now models three pieces of legendary sampling hardware. Stack them per channel for the exact tonal signature you want.

​
SP-1200 Mode — four output styles

Incredibly accurate modeling of the E-mu SP-1200’s four hardware output pairs, refined repeatedly against a real SP-1200 for optimal tone matching:
  • Unfiltered (7 & 8) — raw SP-1200 crunch
  • Filtered (1 & 2) — SSM2044 dynamic ladder filter with an 8-bit-stepped cutoff envelope
  • Filtered (3 & 4) — aggressive active low-pass around 9 kHz
  • Filtered (5 & 6) — gentler high-treble low-pass for pads and melodic content
All settings run the full SP-1200 DSP chain: 11.07 kHz input anti-alias, tape saturation, 12-bit quantization, and a 26.04 kHz multiplexed DAC + sample-and-hold stage. SP-1200 sits after pitch in the signal chain, so aliasing and quantization respond to pitched content just like the real hardware. See SP1200 Filter.

​
ASR-10 Boost — three intensity levels

Accurate emulation of the famous ASR-10 “Boost” mode — a per-channel signal chain that combines tape-style hysteresis saturation → soft-knee limiting → 18 kHz analog output filter.
  • Lo (+4 dB) — subtle warmth
  • Mid (+8 dB) — classic character
  • Hi (+12 dB) — aggressive
Cycle with Left/Right in Channel Detail (Tone section). See ASR-10 Boost.

​
S950 Butterworth HPF/LPF

The HPF and LPF have been rebuilt from the ground up using three cascaded biquad sections with the exact Butterworth Q factors from the Akai S950’s MF6-50 switched-capacitor chip. The result: a maximally flat passband, a precise -3 dB point at the cutoff, and a steep 36 dB/octave cliff beyond it. Cutoff changes are now smoothed in the log-frequency domain with a 5 ms time constant — no more zipper noise on rapid sweeps. PerformFX uses the same biquad core and now supports proper resonance via Q-factor scaling. PerformFX Talkover Duck and Motor Kill now reduce the master to about 40% volume instead of muting it, making them usable as live transition and talkover moves.

​
Airwindows CreamCoat Reverb

The per-channel Reverb control now uses Airwindows CreamCoat for general-purpose ambience, room tone, and longer space. The control stays the same in Channel Detail; existing projects keep their saved reverb amount, mapped directly to CreamCoat Wetness.

​
Airwindows PerformFX Expansion

PerformFX now includes a broad set of selected Airwindows-powered presets for live performance: plate and infinite reverb throws, tape and pitch echoes, vinyl and tape color, DJ isolator color, glitch pitch effects, and Airwindows-colored dub siren throws. Dusty Vinyl now adds event-based vinyl crackle with an editable Crackle control, and Pocket Lofi uses a more audible Pockey2 setting for stronger pocket-sampler character. Existing bindings continue to resolve by preset ID. SP1200, S950-style HPF/LPF behavior, ASR-10 Boost, ducking, pan, width, Motor Kill, and Tape Stop keep their existing engines and controls. Meiji Sampler still does not load arbitrary CLAP, VST, LV2, or AU plugins; these Airwindows effects are compiled into the app as MIT-licensed source.

​
3. Loop sync, stretch, and smarter chopping

​
New timestretching engine

The time-stretch engine has been completely replaced with a formant-preserving spectral stretcher.
  • Better quality for melodic and vocal material — formants stay put when you stretch
  • Tighter loop sync — synced chop stretch reuse eliminates redundant computation and removes audible artifacts at loop edges
  • Stem playback now stretches correctly alongside the rest of the channel
No workflow or shortcut changes. It just sounds better and syncs tighter.

​
Intelligent Transient Snap

The trim view can now snap the playhead to useful transient points before you set a trim point or place a chop marker. Inspired by Cool Edit Pro’s bracket-key snapping.
  • Shift+[ / Shift+] — previous / next snap point (legacy { } still work)
  • Subtle ghost markers show possible snap points on the waveform
  • ReCycle-style sensitivity — raise for softer hits, lower for stronger candidates; press a to autochop the first ten accepted markers to 1-9,0
  • Color-coded by frequency band — low hits, midrange attacks, and high-frequency details are easier to scan
  • b cycles between All / Lows / Mids / Highs when a sample is crowded
  • < / > make snapping stricter or more sensitive
  • Snap targets are refined to nearby zero crossings to reduce click risk
  • Chops and trim points snap to safe playback boundaries automatically
See Chopping And Slicing.

​
Six new audio formats

Meiji Sampler now reads MP3, AAC, FLAC, OGG, ALAC, and CAF alongside the existing WAV and AIFF support. Drop any of them into the browser, Splice library, or recordings folder.

​
Bounce in your bit depth of choice

The bounce modal now exports WAV at 16-bit, 24-bit, or 32-bit float. Default is 24-bit. The setting persists. See Exporting And Bouncing.

​
Recording normalization

Press N during recording review to toggle peak normalization to -0.1 dB. Compare normalized vs original before keeping the take.

​
4. Meiji Sampler Radio

Three new Meiji Sampler internet radio stations that double as sample sources inside the app — perfect for inspiration and quick ideation:
  • Drums — live 24/7
  • Originals — live 24/7
  • Originals II — live 24/7
Tune in directly from Meiji Sampler as a sound source, or listen anywhere on the web at radio.meiji.industries. Hear something you like? Sample, chop, and build with it inside the same workflow.

​
Everything else, stack-ranked

The full picture beyond the marquee work, ordered by user impact within each section.

​
Workflow & navigation

  1. Vim-style navigation (h j k l) everywhere in the app — browser, pads, loops, scenes, mixer, editors, modals, settings. See migration notes for key remaps (e.g. h-on-loops moves to Shift+H).
  2. Linked & copied sample playback — assign the same local sample to another channel and choose Link (shared loop/trim/chop/pattern-end via file defaults) or Copy (snapshot then diverge).
  3. Per-channel playback overrides — loop mode, trim points, and chop edits now write per-channel overrides instead of mutating shared file defaults. Significant workflow change; see Per-Channel Playback.
  4. Linked pad triggers — link a secondary pad to any channel so both fire on every hit. Layer kicks with subs, double snares with claps. Only the source pad is recorded into loops.
  5. Per-channel choke groups — assign channels to one of 10 choke groups for hi-hat exclusivity or any one-at-a-time variation.
  6. Sample Offset — delay pad and chop playback from +0ms to +100ms in 1ms steps. Humanize quantized parts without moving the recorded events.
  7. Tap tempo + manual project BPM override — Settings → Tempo. Tap, type, nudge with arrows, or choose a first-loop Range (65-135, 87-175, 133-300, None) so double-time and half-time detections normalize automatically. When the first recorded loop starts from a LOOP: ON pad, close timing within 10% snaps to an exact number of that pad’s loop cycles.
  8. Tabbed Settings with Project management — save, save as, open, new, browse recents, delete projects, all inline. Categories: Project / Audio / Addons / General.
  9. Settings as a top-level tab — open with , or via Tab cycling (which also reveals Perform when PerformFX is enabled).
  10. SPACE transport guidance — pressing Space on Pads/Loops/Scenes is now row-aware, with toast guidance instead of silent no-ops. Loop slot keys also follow the queued-stop feel of scenes during playback.
  11. Track solo toggles consistently: press s on a selected sequence-editor track to solo it, then press s again to un-solo. When a running loop is muted, m or s on the selected track un-mutes the loop, solos that track, and restarts it immediately. See Muting And Soloing.
  12. Hide loops & scenes rows until needed — cleaner default layout that surfaces UI as you actually use it.
  13. Channel Detail sample actions — manage assigned samples directly from Channel Detail without leaving context.
  14. Bounce WAV bit depth selection — 16 / 24 / 32-bit float; persisted.
  15. Recording normalization toggle (N) — peak-normalize takes during review.
  16. Confirm before replacing autochops — protects existing chop work from being clobbered.
  17. Trim view improvements — snap playhead to sample edges with up/down, e restarts from start point, o toggles loop mode (replacing l), better playhead visibility, wrapped hotkey hints, smaller Shift+arrow step size.
  18. Auto-navigate into single-pack downloads — one-pack factory downloads jump directly into the new pack.
  19. Random colors for loaded audio sources — visual variety without fiddling.
  20. Settings source availability page — see which sample sources are configured at a glance.
  21. Wider Settings pane layout — more breathing room for dense forms.
  22. Compact hotkey labels — input device modal and other dense surfaces.
  23. Tab for top-level navigation, , for Settings — predictable and one-keypress.
  24. Optional login — skip sign-in entirely; telemetry is disabled when you do.
  25. Caps Lock detection — warning callout when caps is on, since the app is heavily case-sensitive.
  26. Grooves renamed to Originals — clearer naming for your own loop content.
  27. Replaced Shift+digit combos for better international keyboard layout compatibility.

​
New sample content

  • DK Presents — Detroit Breaks sample pack (factory)
  • DK Presents — Memphis Breaks sample pack (factory)

​
Audio engine & DSP

  • New timestretching engine — formant-preserving spectral stretcher, better quality for melodic and vocal material
  • S950 Butterworth biquad HPF/LPF with proper resonance support and log-domain cutoff smoothing
  • SP-1200 Mode with full DSP chain (anti-alias, tape sat, 12-bit quantization, 26 kHz S&H stage) and four output styles, refined against a real SP-1200
  • ASR-10 Boost with hysteresis saturation, soft-knee limiting, 18 kHz analog output filter, three intensity levels
  • Synced chop stretch reuse — eliminates redundant stretch computation
  • Stretch ratio computation deferred until after decode — prevents stale ratios
  • Stretched loop edge rendering corrected
  • Stretching for stem playback fixed
  • Chop playback applies time-stretching correctly when source is stretched
  • MP3 / AAC / FLAC / OGG / ALAC / CAF decode support added

​
Platform & distribution

  • macOS deployment target lowered to 11.0 via SwiftTerm upstream fix — broader Mac compatibility
  • Stems Addon download validation now checks the published platform bundle and checksum before install, so missing Demucs artifacts fail with a clear platform/version message instead of a broken archive download.
  • Windows ARM64 Demucs bundle published — native ARM64 stem separation
  • Windows demucs CLI builds added (x64)
  • Windows installer fixes — VM dev build path and font-bundled installers unblocked
  • Windows ARM64 ssstretch build corrected
  • Linux + Windows dev environment improvements for contributors
  • Switched remaining CI jobs to Blacksmith runners for faster builds

​
Reliability & fixes

  • Trim modal CI SIGSEGV resolved
  • Late transient marker placement corrected
  • First transient detection in autochop fixed
  • Loop start delay on playback removed
  • Loop playback issue resolved
  • Delayed loop-triggered pad offsets corrected
  • Inverted sync speed multipliers fixed
  • Quantize key hint labels corrected
  • MIDI pad + chop trigger conflict with Channel Detail dialog resolved
  • DEL no longer exits trim unintentionally; confirm-discard added for unassigned samples
  • Enter on stems install prompt behavior fixed
  • Playback event loss between loop tick and UI tick eliminated
  • Pending stretch and stem playback are now cancelled on stop
  • Downloaded sample files preserved when URL cache evicts entries
  • Radio streaming log spam from blank-source proliferation eliminated
  • Radio metadata corrected for stream identification
  • Radio deploy healthcheck, metadata, and silence-gap fixes
  • Idle VU placeholders now scale by channel gain
  • Caps Lock warning prevents silent input failures
  • Skip no-op transient sensitivity steps
  • Windows nonce TTL test stabilized
  • Tutorial copy corrections

​
Internal & contributor improvements

  • Mintlify documentation site — full content authoring infrastructure with local dev tasks
  • Documentation authoring skill + CLAUDE.md acceptance criteria for AI-assisted doc updates
  • Cargo target lock contention across worktrees avoided
  • Local test workflows sped up significantly
  • Removed unused feature flags
  • Expanded unit and integration test coverage across the codebase
  • DRY’d channel detail layout rendering
  • Hardcoded analytics event counts removed to prevent merge conflicts
  • Inlined claude-skills, removed submodule
  • Default to git worktrees for issue branches
  • /merge-and-resolve-conflicts and /wipe-pr developer skills added

​
Upgrade notes

A few breaking-ish changes worth knowing about:
  • Vim navigation key remaps — h on the loops row is now Shift+H (shorten loop by half), j on pads is Shift+J (jump mode), k in Perform is Shift+K (kill all FX), l in browser is Shift+L (toggle loop mode), and l in trim is now o. See Upgrade And Migration Notes.
  • Per-channel playback overrides — channel detail and trim view edits now write per-channel overrides instead of mutating shared file defaults. If you relied on file-level edits propagating everywhere, you’ll see different behavior. Browser and details preview still edit file defaults.
  • F1–F4 navigation removed — replaced by Tab cycling and , for Settings.
  • Shift+digit shortcuts were replaced for international keyboard compatibility — see the keyboard reference for new bindings.
  • Stretch cache schema bumped to v4 — legacy caches are purged automatically on first launch. No manual action.
  • Splice login is required for downloads (Sounds+, Creator, or Creator+).
Full details in Upgrade And Migration Notes.

​
What 1.0 means

This release earns the version number. Meiji Sampler now ships:
  • Stable across three platforms — macOS, Linux, Windows (x64 + ARM64)
  • Documented — full guide, recipe, reference, troubleshooting, and migration content at docs.sampler.meiji.industries
  • Integrated — first-class MIDI, first-class Splice
  • Sonically opinionated — three vintage modeling effects covering the most influential samplers in hip-hop and electronic history
  • Privacy-respecting — login is optional, telemetry is disabled when you skip it
  • Self-published — no GitHub access required, downloads at sampler.meiji.industries
Pure keyboard. Zero mouse. All terminal.