Open-Source Licenses
Last updated: September 28, 2026
Most GrabCast tools run entirely in your browser. They are built on the open-source projects below. We thank their authors. Each project is used under its own license; the license texts are available at the linked project pages.
FFmpeg notice (GPL)
The video and audio tools use FFmpeg compiled to WebAssembly by the ffmpeg.wasm project (@ffmpeg/core 0.12.6). We host an unmodified copy of that build at /vendor/ffmpeg/. It is built with --enable-gpl and includes GPL components such as libx264 and libx265, so the compiled FFmpeg core is distributed under the GNU General Public License, version 2 or later. It also contains libraries under LGPL, BSD, MIT and similar licenses (for example LAME, libvpx, Opus, Vorbis, Theora, libwebp, zlib, zimg, FreeType, FriBidi and libass).
- Build scripts and source of this build: github.com/ffmpegwasm/ffmpeg.wasm (package @ffmpeg/core 0.12.6).
- FFmpeg source code: ffmpeg.org/download.html. FFmpeg is a trademark of Fabrice Bellard; GrabCast is not affiliated with the FFmpeg project.
- Written offer: for at least three years from the date we last distribute this build, anyone may request the complete corresponding source code by emailing support@grabcast.click. We will provide it for no more than the cost of physically performing the distribution.
FFmpeg runs as a separate program inside your browser. The rest of GrabCast is not a derivative of it and is not covered by the GPL.
eSpeak NG notice (GPL)
Text to Speech, AI Video Dubbing, the Template Studio AI voice and the Marketing Studio video voice-over (stock voices Heart, Bella, Michael and Fenrir) load the unmodified kokoro-js 1.2.1 web bundle from the jsDelivr CDN. That bundle contains an English-only build of eSpeak NG compiled to WebAssembly (through the phonemizer package), which is licensed under the GNU General Public License, version 3 or later. It only converts English text into phonemes; the voice itself comes from the Kokoro-82M model (Apache-2.0).
- eSpeak NG source code: github.com/espeak-ng/espeak-ng.
- Source of the bundle: kokoro-js and phonemizer (package kokoro-js 1.2.1).
- Written offer: for at least three years from the date we last use this build, anyone may request the complete corresponding source code by emailing support@grabcast.click. We will provide it for no more than the cost of physically performing the distribution.
The phonemizer runs as a separate program inside your browser. The rest of GrabCast is not a derivative of it and is not covered by the GPL.
JavaScript libraries
| Project | Version | License | Used for |
|---|---|---|---|
| FFmpeg (compiled to WebAssembly by ffmpeg.wasm) | @ffmpeg/core 0.12.6 | GPL-2.0-or-later (see notice below) | video and audio tools |
| ffmpeg.wasm JavaScript wrapper | @ffmpeg/ffmpeg 0.12.10, @ffmpeg/util 0.12.1 | MIT | loading FFmpeg in the browser |
| mp4box.js | 0.5.3 | BSD-3-Clause | reading MP4 sample tables for the Studio video editor |
| pdf.js | 3.11.174 | Apache-2.0 | reading and rendering PDFs |
| pdf-lib | 1.17.1 | MIT | creating and editing PDFs |
| @pdf-lib/fontkit | 1.1.1 | MIT | embedding fonts in PDFs |
| qpdf-wasm (qpdf) | 0.3.0 | ISC (qpdf: Apache-2.0) | PDF encryption, unlocking and repair |
| pdfmake | 0.2.10 | MIT | Word and Excel to PDF |
| html-to-pdfmake | 2.5.13 | MIT | Word to PDF |
| mammoth.js | 1.8.0 | BSD-2-Clause | reading .docx files |
| docx | 8.5.0 | MIT | PDF to Word |
| PptxGenJS | 3.12.0 | MIT | PDF to PowerPoint |
| SheetJS Community Edition | 0.18.5 and 0.20.3 | Apache-2.0 | spreadsheets (PDF to Excel, Excel to PDF) |
| Tesseract.js | 5.1.1 | Apache-2.0 | OCR (image to text, OCR PDF) |
| Transformers.js | 3.x | Apache-2.0 | in-browser AI models |
| ONNX Runtime Web | 1.22 and 1.30 | MIT | AI image upscaler, Magic Eraser, background remover, vocal remover |
| demucs-web (pre/post-processing, re-implemented) | 1.0.2 | MIT | vocal remover & stem splitter |
| kokoro-js | 1.2.1 | Apache-2.0 | text to speech, AI video dubbing, Template Studio AI voice |
| eSpeak NG (English phonemizer compiled to WebAssembly, bundled inside kokoro-js via phonemizer) | as bundled in kokoro-js 1.2.1 | GPL-3.0-or-later (see notice below) | turning English text into phonemes for the voice model |
| MediaPipe Tasks Vision | 0.10.14 | Apache-2.0 | face detection (blur, passport photo, Studio auto-reframe) and person segmentation (Studio background effects) |
| heic2any (includes libheif) | 0.0.4 | MIT (libheif: LGPL-3.0) | HEIC to JPG |
| UTIF.js | 3.1.0 | MIT | TIFF input |
| UPNG.js | 2.1.0 | MIT | PNG compression |
| gifenc | 1.0.3 | MIT | GIF maker |
| ImageTracer.js | 1.2.6 | Unlicense (public domain) | image to SVG |
| mp4-muxer | 5.2.2 | MIT | MP4 export (waveform video, screen recorder) |
| lamejs | 1.2.1 | LGPL-3.0 (loaded unmodified as a separate library) | MP3 encoding |
| JSZip | 3.10.1 | MIT (dual MIT / GPL-3.0, used under MIT) | ZIP downloads |
| pako | 1.0.11 | MIT and Zlib | compression |
| SortableJS | 1.15.2 | MIT | drag-and-drop ordering |
| highlight.js | 11.9.0 | BSD-3-Clause | code to image |
| html2canvas | 1.4.1 | MIT | screenshots of rendered elements |
| marked | 12.0.2 | MIT | Markdown to HTML |
| DOMPurify | 3.1.7 | MPL-2.0 or Apache-2.0 (used under Apache-2.0) | HTML sanitizing |
| qrcode-generator | 1.4.4 | MIT | QR code generator |
| jsQR | 1.4.0 | Apache-2.0 | QR code scanner |
| jsvectormap | 1.5.3 | MIT | admin map |
| MSAL.js (@azure/msal-browser) | 3.30.0 | MIT | OneDrive picker sign-in |
AI models (run on your device)
| Model | License | Used for |
|---|---|---|
| Whisper (tiny, base, small, large-v3-turbo) | MIT | speech to text |
| Kokoro-82M | Apache-2.0 | text to speech |
| MODNet | Apache-2.0 | background remover, portrait blur |
| HTDemucs (Meta AI), browser ONNX build monteslu/htdemucs-web-onnx | MIT | vocal remover & stem splitter |
| U²-Net (u2net) (notice) | Apache-2.0 | background remover (objects, products and pets) |
| Swin2SR (lightweight x2) | Apache-2.0 | AI image upscaler |
| Real-ESRGAN General x4v3 (notice) | BSD-3-Clause | AI image upscaler (Fast model) |
| MediaPipe BlazeFace face detector | Apache-2.0 | face detection (including Studio auto-reframe) |
| MediaPipe Selfie Segmenter | Apache-2.0 | Studio: remove, blur or replace a person's background |
| Tesseract trained data | Apache-2.0 | OCR languages |
Fonts
| Font | License |
|---|---|
| Inter | SIL Open Font License 1.1 |
| Noto Sans and Noto fonts | SIL Open Font License 1.1 |
| Roboto (bundled with pdfmake) | Apache-2.0 |
| Great Vibes, Allura, Sacramento, Anton, Archivo Black | SIL Open Font License 1.1 |
| Space Grotesk, DM Serif Display, JetBrains Mono (self-hosted, promo video templates) | SIL Open Font License 1.1 |
| Fonts from Fontsource used by Add Text to Image | SIL Open Font License 1.1 (each font's own license) |
Questions
If you think a license notice is missing or wrong, please email support@grabcast.click and we will fix it.