rescript

wassgha活跃维护

基于转录文本的浏览器端视频/音频剪辑器,改文字就能剪片子,不用拖时间轴。

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

v1.2.5发布于 2026-10-29 17:40 UTC1,058star +13 / 7 天138fork同步于 2026-11-10 11:51 UTC

怎么上手

从 README 代码块原样提取,可能是其中一种安装方式
npm install     # also copies ffmpeg/onnxruntime WASM into public/vendor
npm run dev     # Next.js web app (http://localhost:3000)
npm run electron:dev   # Electron shell + Next.js dev server
硬件要求按 README 关键词推断
支持 Apple Silicon
下载:源码 zip·GitHub Releases(安装包)本站不镜像安装包

项目说明

这个项目的说明还在生成中,先参考下面「最近发版」和 README。

最近发版

5 个版本v1.2.5
  1. v1.2.51.2.5最新正式版2026-10-29 · 2026-10-29 17:40 UTC

    Highlights

    • Blender export — you can now export your edit as a Blender Video Editing sequence project (.blend file).
    • Italian UI — the interface is now available in Italian.
    • Better silence removal — short words that fall between removed silences are no longer incorrectly cut off.
    • Accurate frame rate detection — the source video's frame rate is now read directly from the file when extracting audio, improving export accuracy.

    Changes

    Export

    • Added a new "Blender Video Editing" export option that generates a .blend project file from your edit timeline.
    • Export dialog updated to surface the new Blender export format.

    Editing

    • Fixed a bug where short words sandwiched between two removed silences were being dropped from the transcript/edit.

    Transcription

    • Frame rate is now detected from the source video during audio extraction, ensuring downstream processing uses the correct timing.

    Desktop

    • Italian locale added to the Electron desktop shell.

    Localization

    • Italian added as a supported interface language (joins de, en, es, fr, ja, ko, pt, zh-CN, zh-TW).

    Fixes

    • Fixed Italian locale not being wired into the release build correctly.
  2. v1.2.41.2.42026-10-05 · 2026-10-05 01:15 UTC
    查看更新说明

    Highlights

    • The transcript editor is now accessible to screen readers via a dedicated text layer
    • Auto-updater errors on the desktop app are now handled gracefully instead of crashing

    Changes

    Accessibility

    • Added a screen-reader text layer to the transcript editor, making transcript content and speaker labels navigable with assistive technologies
    • Updated localization strings across all supported languages (de, es, fr, ja, ko, pt, zh-CN, zh-TW) to support accessibility labels

    Desktop

    • Caught unhandled promise rejections from the auto-updater's check and download operations, preventing silent failures or crashes when update checks fail
  3. v1.2.31.2.32026-10-03 · 2026-10-03 18:27 UTC
    查看更新说明

    Highlights

    • Desktop exports are now significantly faster thanks to hardware-accelerated encoding via Apple VideoToolbox
    • Export resolution now defaults to 1080p on desktop

    Changes

    Export

    • Added hardware-accelerated video encoding (VideoToolbox) for faster exports on macOS desktop builds
    • Changed default export resolution to 1080p in the Export dialog
    • Improved FFmpeg argument generation for desktop export paths
    • Extended export options exposed to the desktop app via updated type definitions

    Desktop

    • Updated FFmpeg runner to support hardware codec selection and handle VideoToolbox encoding pipeline
    • Improved native media handling to integrate with hardware acceleration where available
  4. v1.2.21.2.22026-10-03 · 2026-10-03 16:56 UTC
    查看更新说明

    Highlights

    • Desktop exports now use a bundled native ffmpeg — faster, more reliable rendering that no longer depends on the browser-based WebAssembly ffmpeg.
    • 1080p and original-quality exports fixed on desktop — a memory limitation in the previous ffmpeg core prevented high-resolution exports from completing.
    • Improved export stability — half-loaded ffmpeg workers are now properly terminated before falling back to single-threaded mode, preventing stalled or failed exports.

    Changes

    Export

    • Desktop app renders exports using a bundled native ffmpeg binary instead of the WebAssembly ffmpeg used in the browser.
    • Fixed 1080p and original-resolution exports on desktop by switching to a growable-memory ffmpeg core.
    • Fixed a memory leak/hang where partially initialised multi-threaded ffmpeg workers were left running when falling back to single-threaded mode.
    • Export dialog updated to support and surface native export progress and error states.

    Desktop

    • Added ffmpegRunner module to manage native ffmpeg processes from the Electron main process.
    • Exposed native media and ffmpeg capabilities via an expanded preload API.
    • Build scripts now fetch and bundle the appropriate native ffmpeg binary at pack time.

    Fixes

    • Various i18n message additions across all supported languages (de, en, es, fr, ja, ko, pt, zh-CN, zh-TW) to cover new export-related UI strings.
  5. v1.2.11.2.12026-09-14 · 2026-09-14 22:46 UTC
    查看更新说明

    Highlights

    • Fixed macOS app signing so that notarized builds are correctly distributed to Mac users

    Changes

    Desktop

    • Fixed macOS release signing on GitHub Actions runners (Darwin 25.6 compatibility)

用户点评

登录用户的真实使用记录,可按版本和硬件筛选
共 0 条

    还没有人点评。第一个说出真实体验的人最有价值。

    README 原文

    未翻译

    rescript.

    Join the Discord Follow @wassgha on X

    ✨ Try it in the browser app.getrescript.com or download the Desktop App

    Rescript Demo

    Rescript is an open-source, transcript-based media editor. Drop in a video or audio file and it is transcribed locally with per-word timestamps and speaker labels. Delete words in the transcript and the corresponding clip is cut from the media. Export the final cut — without your file ever leaving your device.

    Download

           

    See the Releases page. Desktop builds auto-update from GitHub Releases. Prefer the browser? Use the web app — same editor, no install.

    • 🔒 Private by design — no auth, no uploads; all media processing happens on-device
    • 📝 Word-level editing — select words, press ⌫, the cut follows the text
    • 📥 Import your own transcript — skip Whisper and edit with an SRT, VTT, or JSON caption file
    • 📤 Export hub — video (MP4/WebM, 720p–4K), audio (M4A/MP3/WAV), transcript / captions (TXT/MD/DOCX/PDF/SRT/VTT/JSON, optional timestamps on documents), or NLE/DAW timeline (Resolve/Premiere/FCP/AAF/Reaper/Samplitude EDL/Blender)
    • 🧹 Filler removal — one-click cut of "um", "uh", and similar fillers
    • 🔇 Silence removal — one-click cut of pauses and dead air (≥0.3s)
    • 🗣️ Speaker diarization — the transcript is 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 ASR alignment is off
    • 🔴 Cut edges — drag either edge of a cut to trim independently of Whisper 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, Blender .blend (Video Editing project linked to the original file; save it next to the source)
    • 🎧 Audio files — edit podcasts, voice notes, and interviews the same way as video
    • 🖥️ Desktop app — macOS, Windows, and Linux via Electron (signed + notarized on Mac)

    Stack

    PieceTech
    AppNext.js + React + TypeScript + Tailwind
    DesktopElectron + electron-builder (auto-update from GitHub Releases)
    Transcriptiontransformers.js running whisper-base_timestamped or whisper-small_timestamped (WebGPU with WASM fallback) in a Web Worker
    Speaker labelspyannote-segmentation-3.0 (ONNX)
    Media processingffmpeg.wasm (multi-threaded) for audio extraction and export
    Statezustand

    Development

    npm install     # also copies ffmpeg/onnxruntime WASM into public/vendor
    npm run dev     # Next.js web app (http://localhost:3000)
    npm run electron:dev   # Electron shell + Next.js dev server
    npm run build   # production web build
    npm run dist    # unsigned desktop installers into dist/
    npm run lint    # eslint
    

    For desktop packaging, signing, and cutting releases, see RELEASING.md.

    How it works

    Rescript Promo

    1. Extract — ffmpeg.wasm decodes the audio track to mono 16 kHz PCM.
    2. Transcribe — Whisper runs in a Web Worker with return_timestamps: "word", streaming text as it goes; pyannote assigns a speaker to every word. Choose Whisper Base, Whisper Small, or Import transcript (SRT / VTT / JSON) on the homepage.
    3. Edit — deleting words produces "cut ranges" of the original media. The preview player skips them in real time and the timeline shows them in red. Remove fillers cuts every detected "um" / "uh" / etc. in one click. Remove silences cuts pauses and dead air of 0.3s or longer.
    4. Export — the kept ranges are trimmed and concatenated with an ffmpeg filter graph and re-encoded (libx264/aac), so cuts are word-accurate.

    Browser support

    A Chromium-based browser is recommended for the web app. It requires SharedArrayBuffer (served with COOP/COEP headers) and uses WebGPU for inference when available, falling back to WASM otherwise. The desktop app bundles Chromium via Electron and sets the same isolation headers on its app:// protocol.

    Telemetry

    By default, Rescript reports anonymous usage stats and crash reports so we can tell how many people actually use it, which features are worth the maintenance, and what's breaking. You can turn both off in Settings → Privacy → Help improve the app, which stops them immediately and permanently on that install. Crash reports go to Sentry and carry a stack trace, the app version, and the platform.

    License

    Copyright (c) 2026 Wassim Gharbi and Rescript contributors.

    Licensed under the PolyForm Noncommercial License 1.0.0. You may use, modify, and share Rescript for noncommercial purposes only, and you must retain the required copyright notice. Commercial use (including reselling or redistributing the software for a fee) is not permitted under this license. Contact the author for commercial licensing.

    Prior releases published under the MIT License remain available under MIT for those versions only.


    License: PolyForm Noncommercial Platforms Electron Stars Latest release

    Built by @wassgha — follow along on X for updates, or come say hi in the Discord.