Rescript

Rescript is a transcript-based video and audio editor for speech-driven media. It transcribes a recording with word-level timestamps, then maps text edits to the underlying audio and video: deleting words removes the corresponding footage, while playback previews the cut as you work. Existing SRT, VTT, or JSON transcripts can also be imported instead of running transcription.

The workflow is designed for spoken recordings that need a quick rough cut or correction. Rescript can remove filler words and pauses, identify speakers locally, and provide a conventional timeline with waveforms, cut regions, and draggable timing handles for manual adjustments. Its Regenerate feature lets you rewrite a selected line and synthesize it in that speaker’s voice from clean audio already in the recording, with the generated speech fitted to the replaced span.

Transcription, editing, voice synthesis, and export run on the user’s device. The browser editor and desktop builds work without an account, and the Whisper model is downloaded once and cached. Exports include video, audio, transcripts, and subtitle files. Rescript is intended for transcript-led editing rather than full production work: the source specifically says it does not provide recording, multitrack editing, layers, transitions, titles, color tools, effects, noise removal, or burned-in caption styling.