Inside Rescript: on-device Whisper transcription, voice-clone retakes, zero uploads

JeremyNguyenPhD · x · 2026-09-20

This thread follow-up links Rescript's full feature set: Whisper transcribes on-device with word-level timestamps and speaker diarization (or import SRT/VTT/JSON); deleting words cuts the matching footage, with one-click removal of filler words and silences over 0.3s; you can rewrite a fumbled line and hear it re-spoken in the original speaker's cloned voice, all rendered locally with ffmpeg — nothing is ever uploaded. Available on macOS, Windows, Linux, and web.

Related event: Rescript: Open-Source Video Editing by Editing Text(2 posts)→

Original post →

More from Apps

Apps channel →