Features
Everything earns its place.
Real audio analysis
Threshold, minimum gap, pre/post padding over 20ms RMS windows with hysteresis confirmation.
Three detection modes
Speech-focused, Balanced, Aggressive presets for clean VO through noisy rooms.
Cut-by-cut review
Timestamped list with duration and confidence. Jump-to, keep, remove, undo/redo.
Readable timeline
Waveform with speech / will-remove / kept states, playhead, keyboard seeking.
Edited preview
Original vs. edited modes with silence-skipping playback before you commit.
Honest export
Original / 1080p / 720p, balanced / high / small, MP4-when-supported capture with human progress.
Projects
Recent videos with duration, cuts, and status stored locally — open, rename, delete.
Private by architecture
No server, no upload queue. Decoding and detection run in your browser.