Samples

Overview

Cuts a recording into a corpus of individual sounds. Record from your microphone or open any audio file and Samples draws the waveform, then finds each sound for you — a sample begins at the onset of a new audio event and runs until relative silence or the start of the next event. Preview any sample on the waveform, drag its start and end to adjust, and untick the ones you don’t want. Save them at their natural length or all trimmed to one duration with a tunable fade-out, optionally normalized, as mono WAV files in a folder you choose.

How it works

Samples segments audio with an energy-envelope onset detector.

1. The recording is mixed to mono and reduced to a short-term power envelope in decibels.
2. A silence threshold is set relative to the loudest part of the file (−40 dB by default), and never below the recording's own noise floor.
3. A sample starts when the level rises above that threshold, or rises sharply (8 dB by default) out of a dip while another sample is still sounding — that is the next event. Starts are then snapped to the actual transient.
4. A sample ends when the level stays below the silence threshold for a hold time (80 ms), or when the next event begins, whichever comes first. Anything shorter than the minimum length (60 ms) is dropped.

Every threshold is adjustable, and every detected sample can be auditioned, trimmed by dragging its edges on the waveform, or left out before saving. On export, samples are kept at their natural length or trimmed to one duration with a tunable fade-out, optionally normalised.