Silence is defined by a threshold
The tool classifies decoded audio chunks by measured RMS level. It does not claim to understand speech, music or intent.
Minimum duration protects natural pauses
Removing every tiny quiet gap can make speech sound unnatural. A sustained-silence rule is safer than a single-sample gate.
Noise changes the result
Air conditioning, room noise or hiss can keep the RMS above threshold even when nobody is speaking.
Review the compacted timeline
Listen near every major transition before publishing important speech or music.
Official marketplace references
Requirements can change. Review the source before publishing an important listing.
Mediabunny browser media toolkit documentation ↗