The first professional
in-browser sample editor.
No install, no plugin, no desktop app. Pick the recordings you want to analyse and get a playable instrument in minutes.
Two ways your samples arrive.
One way they leave.
Whether your source is a single long take with every note packed inside, or a folder where each note is already its own file, the editor reads both — and lands you in the same place: a playable instrument mapped across the keyboard.
One file. Every note inside.
A single long WAV — recorded in one pass with notes played back-to-back, or dumped from a hardware sampler — drops onto the editor as-is. No splitting beforehand, no DAW round-trip.
Energy-aware silence detection scans the file and finds the boundary between every distinct note. Each segment is isolated into its own clip in memory, the AI pitch ensemble runs over each one in parallel, and the detected MIDI notes map themselves across the keyboard. You see the split points on the waveform before anything commits — nudge a boundary, merge two segments that landed too close, or drop a marker the detector missed.
- Silence-aware splitter with adjustable threshold and minimum-gap controls
- Up to 61 notes per session, processed in parallel by the pitch worker pool
- Per-segment override before commit — wrong split? Drag the boundary
- Source WAV stays untouched on your disk; everything happens in browser memory
A folder. One file per note.
Already split? Open the file picker and select the notes you want — pick all of them, or just the ones you need for this instrument. The editor accepts mixed-case filenames, accidentals as # or s, octave numbers, and arbitrary prefixes — the parser uses filenames as hints, not as truth.
Every file gets pitch-detected by the same AI ensemble, in parallel. If a filename says C2 but the audio is actually C#2, the audio wins and the note moves where it belongs — silently, with a flag you can review. Stereo files (detected by header) claim a stereo pair; mono files stay mono. Both live in the same instrument without forcing a sample rate or bit-depth conversion.
- Multi-select the notes you want from the file picker — pick all, or just the subset for this instrument
- Hard cap of 61 samples per session (one per MIDI note across the playable range)
- Filenames are hints; audio content is authoritative — mislabelled files self-correct
- Stereo and mono coexist in the same instrument, per-side handling preserved
Either workflow ends at the same place: 61 notes on a keyboard, every one auditionable, every one editable, every one ready to export — SFZ, SF2, KMP, WAV+SMPL, or the native HLI project.
Pick the recordings you want. Get a keyboard.
Advanced AI pitch detection reads each sample and assigns it to the right note. The result is accurate enough that most content needs zero manual correction. Anything ambiguous is flagged for you to review; the rest maps itself.
- Auto-fill missing semitones with intelligent interpolation
- Built-in keyboard for instant audition (mouse, MIDI, or QWERTY)
- Per-note edit: pitch, loop, fade-in, normalize (ADSR is preview-only — informative, so you can hear how the note would sound on a keyboard)
- Stereo and mono in the same instrument — handled per-side
One loop in. A kit on the keyboard out.
Drop a one-shot drum loop. Onset detection slices it into hits, each hit gets mapped to a pad, and the whole thing is playable in seconds. Quantize keeps slices on the grid, anti-overlap fade prevents clicks, live trigger lets you audition before you export.
Seamless loops. On stereo. On hardware.
Five next-generation loop finders for multisamples (and three for drum slices) propose loop candidates — you pick which one to trust per note. Each algorithm is a unique combination of parameters we built ourselves, not a stock library you'll find anywhere else. The crossfade editor snaps to zero crossings. Stereo samples keep independent loop points per channel — so when your file lands on arranger hardware that expects two mono halves, the engine gets exactly what it expects. No forced L = R, no destroyed left channel past the 10-second mark.
Export to the format
your sampler speaks.
Open standards plus the proprietary formats used by professional arranger keyboards. No conversion tools in the middle.
Four steps. Five minutes.
WAV files (one per pitch for a multisample, or a single loop for the drum slicer). Folder drag-and-drop included.
AI pitch detection analyses each sample and assigns it to the right note. Anything ambiguous is flagged for you to review; the rest is mapped automatically.
Pick from five next-generation loop finders for multisamples and three for drums — each one a unique combination of parameters we engineered ourselves, not a library import. Audition with a real keyboard, tweak with a crossfade editor that snaps to zero crossings, per-channel on stereo.
Export to the format your hardware speaks. Per-channel loops preserved, sample names in the right naming convention, ZIP delivered to your downloads.
Three things we won't compromise on.
We don't keep your audio. Both Drum and Multisample modes work the same way — we don't store your samples, we don't index them, we don't keep them on our servers a moment longer than the job needs.
Per-channel stereo loop points. SF2 stereo pairs that don't break on PA-series hardware. KMP/KSF with byte-accurate chunk layout. Tested against the spec, not against a heuristic.
Drag, detect, audition, export. No round-trips to a desktop app, no plugin install, no project file format you have to maintain.
Ship the instrument
you imagined.
No install. No round-trip. No project file to back up. Just a tab, your samples, and the format your hardware speaks.