Drop in a video — small clips to 10 GB+ files. Get it back without filler words, without long dead pauses, and with a transcript + subtitles. Everything is processed on your device — nothing is uploaded to any server.
First run downloads a speech model (~80 MB) and the video engine (~32 MB) — both are
cached by your browser afterwards. Big files (10 GB+) are streamed straight from your
disk in slices, so they never load into memory at once; processing time depends on your
device and the video length. Very large results save as .ts (plays in VLC,
QuickTime and most players) — use “Save to disk” for those.
🎙️ Filler words — the video is transcribed locally with Whisper
(word-level timestamps). Every uh, um, … is cut out.
⏱️ Dead space — quiet pauses longer than 0.5 s are shortened to a quick 0.2 s breath.
🎬 The cut — ffmpeg (WebAssembly) trims those moments and stitches the rest back together, so audio and video stay in sync.
📝 Transcript — the surviving words are written out as a clean transcript and timed subtitles matched to the final video.
🔒 Privacy — no accounts, no uploads, no watermark. Works offline after the first load.