wassgha/rescript

🎬 Open source, transcript-based video/audio editor that lives in the browser.

Rescript is a transcript-based media editor. You drop in a video or audio file, it gets transcribed on your device with per-word timestamps and speaker labels. Then you edit the text: delete words in the transcript and the corresponding parts of the media are cut. Export the final result. The file never leaves your machine.

What problems does it solve?

Traditional video editing is timeline-based — you drag clips around on tracks. Rescript flips that: you edit the words, and the cuts happen automatically. It also removes the privacy concern of cloud transcription services, since everything runs locally.

Key features (from the README)

  • Private by design — no auth, no uploads; all media processing happens on-device.
  • Word-level editing — select words, press Delete, the cut follows the text.
  • Import your own transcript — skip transcription entirely; edit with an SRT, VTT, or JSON caption file.
  • Export hub — video (MP4/WebM, 720p–4K), audio (M4A/MP3/WAV), transcripts/captions (TXT/MD/DOCX/PDF/SRT/VTT/JSON), or NLE/DAW timeline formats (Resolve/Premiere/FCP/AAF/Reaper/Samplitude EDL).
  • Filler removal — one-click cut of "um", "uh", etc.
  • Silence removal — one-click cut of pauses ≥0.3s.
  • Speaker diarization — transcript grouped by speaker.
  • Timeline — waveform, wordbar with draggable timing handles, Split, cut regions, playhead; scroll to zoom, side-scroll to pan.
  • Split & trim — blade clips at the playhead; drag clip edges to refine cuts beyond word boundaries.
  • Word timing — zoom in and drag a word's edges when speech-to-text alignment is off.
  • Cut edges — drag either edge of a cut to trim independently of ASR timestamps; double-click to reset.
  • Live preview — playback skips your cuts in real time.
  • In-browser / desktop export — frame-accurate re-encode with ffmpeg.wasm.
  • NLE/DAW timeline export — DaVinci Resolve / Premiere XML, Final Cut FCPXML, Pro Tools/Logic AAF, Reaper .rpp, Samplitude .edl.
  • Audio files — edit podcasts, voice notes, interviews the same way as video.
  • Desktop app — macOS, Windows, Linux via Electron (signed + notarized on Mac).

How it works (4 steps)

  1. Extract — ffmpeg.wasm decodes the audio track to mono 16 kHz PCM.
  2. Transcribe — Whisper runs in a Web Worker with per-word timestamps, streaming text as it goes; pyannote assigns a speaker to every word. You can choose Whisper Base, Whisper Small, or import your own transcript.
  3. Edit — deleting words produces "cut ranges" of the original media. The preview player skips them in real time; the timeline shows them in red. Remove fillers / silences in one click.
  4. Export — kept ranges are trimmed, concatenated with an ffmpeg filter graph, and re-encoded (libx264/aac), so cuts are word-accurate.

Technical stack

Piece Tech
App Next.js + React + TypeScript + Tailwind
Desktop Electron + electron-builder (auto-update from GitHub Releases)
Transcription transformers.js running whisper-base_timestamped or whisper-small_timestamped (WebGPU with WASM fallback) in a Web Worker
Speaker labels pyannote-segmentation-3.0 (ONNX)
Media processing ffmpeg.wasm (multi-threaded) for audio extraction and export
State zustand

Where can you run it?

  • Browser: a Chromium-based browser is recommended. Requires SharedArrayBuffer (served with COOP/COEP headers) and uses WebGPU for inference when available, falling back to WASM.
  • Desktop app: macOS, Windows, Linux via Electron — bundles Chromium and sets the same isolation headers.

Privacy & telemetry

  • All media processing happens on-device (no uploads).
  • By default, Rescript reports anonymous usage stats and crash reports (crash reports go to Sentry with stack trace, app version, platform). You can turn this off in Settings → Privacy → “Help improve the app” — that stops it immediately and permanently on that install.

License

Copyright (c) 2026 Wassim Gharbi and Rescript contributors.

  • PolyForm Noncommercial License 1.0.0 — you may use, modify, and share Rescript for noncommercial purposes only, and must retain the copyright notice. Commercial use (including reselling for a fee) is not permitted. Contact the author for commercial licensing.
  • Prior releases published under the MIT License remain available under MIT for those versions only.

Quick links

  • Try in browser: app.getrescript.com
  • Downloads: macOS (Apple Silicon / Intel), Windows, Linux (AppImage / .deb) — auto-update from GitHub Releases.
  • Dev setup: npm install && npm run dev (web) or npm run electron:dev (desktop).
  • Community: Discord · X / Twitter

Related

  • Project
    0xsline/StoryGen-Atelier

    An AI-assisted tool that generates storyboards and stitches them into coherent videos using Gemini and Vertex AI Veo.

  • Project
    anYuJia/better-douyin

    better-douyin is a Rust/Tauri desktop client for Douyin (TikTok) offering browsing, downloading, multi‑account support, and AI‑assisted interactions. The public repo only contains the UI and a mock backend; the full functional binary is distributed via GitHub Releases under a non‑commercial license.

  • Project
    c1overzz/dsh-bili-scroll

    dsh‑bili‑scroll is a DSH plugin that automatically searches recent B‑site videos, reads subtitles (or locally transcribed speech), selects worthwhile items, writes a Markdown report, and can post @‑mentions for you. It runs in a self‑contained Python environment (downloaded via uv) and integrates with the bilibili‑mcp AI model for searching and subtitle processing.

  • Project
    thu-ml/Causal-Forcing

    A framework for high-quality real-time interactive video generation using Causal ODE and Consistency Distillation to optimize autoregressive diffusion distillation.