Cut a loop into its hits. The slicer measures how sharply the spectrum changes from one frame to the next, picks the peaks as onsets, and cuts the file from each onset to the next. Drums, plucked instruments and spoken syllables slice cleanly; sustained material needs the sensitivity raised, and the result is still an estimate.
Sensitivity and shortest slice
Sensitivity sets how far above its surroundings a peak has to stand to count. Low keeps only the strongest hits, high picks up ghost notes, hat ticks and the start of held notes. Shortest slice merges onsets that fall closer together than its value, so a flam reads as one hit, and drops any slice shorter than it.
Lead-in and the slice count
Lead-in starts each slice a few milliseconds before its onset so the attack is not clipped. Most slices caps how many are written: a ZIP of a hundred one-shots is fine, a thousand is not. The selected range confines slicing to part of the file.
Auditioning
Every slice is a chip under the waveform. Click one and playback is confined to it from its start, which is the quickest way to check a cut point before exporting. The slices are also drawn as numbered bands on the waveform.
Export
Slices are written in the source’s format with no fade, one file per slice, numbered. The onset list goes into the ZIP as CSV or JSON with start and end in seconds and the start sample, and can be downloaded on its own for a sampler or a script.