Licenses and attribution
FreeClips is built on open-source software and openly published models. Thank you to their authors.
Software in the app
| Component | Used for | License |
|---|---|---|
| Mediabunny 1.66 | Reading and writing MP4, MOV, WebM, MKV and audio formats; WebCodecs decoding and encoding | MPL-2.0 (unmodified; source at the link) |
| @mediabunny/mp3-encoder with LAME 3.100 | MP3 export | MPL-2.0; LAME is LGPL-2.0-or-later, built unmodified from the official source — the build steps and licence files ship with the package |
| Transformers.js 4.3 | Running the speech, language and speaker models | Apache-2.0 |
| ONNX Runtime Web 1.31 (development build pinned by Transformers.js) | Model inference (WebAssembly and WebGPU) | MIT |
| MediaPipe Tasks Vision 1.2 | Person segmentation for the background remover | Apache-2.0 |
| gifenc 1.0 | GIF encoding | MIT |
| hash-wasm 4.13 | Checking model downloads (SHA-256) | MIT |
| @huggingface/tokenizers, @huggingface/jinja | Model tokenizers and chat templates | Apache-2.0, MIT |
| React 19 | User interface | MIT |
| flatbuffers, long, protobufjs, guid-typescript, platform | ONNX Runtime dependencies | Apache-2.0, Apache-2.0, BSD-3-Clause, ISC, MIT |
Models (each downloaded only when you use its tool)
- Speech: Whisper tiny and base, timestamped ONNX conversions by onnx-community, derived from OpenAI Whisper (MIT). The converted repositories did not declare a license in their metadata when we checked; we link the upstream license and are confirming the conversion's terms before a public launch.
- AI highlights: Qwen3 0.6B and 4B, ONNX conversions by onnx-community of Qwen3 (Apache-2.0 upstream). The conversion repositories did not declare a license in their metadata when we checked.
- Speaker labels: pyannote segmentation 3.0 (MIT) and WeSpeaker ResNet34-LM (CC BY 4.0 — © the WeSpeaker authors; trained on VoxCeleb).
- Background remover: MediaPipe Selfie Segmenter by Google (Apache-2.0 per its model card), hosted by FreeClips unmodified.
- Upscaler: Real-ESRGAN realesr-general-x4v3 and realesr-animevideov3 by Xintao Wang et al. (BSD-3-Clause), converted to ONNX by FreeClips with a reproducible, numerically verified script and hosted by us.
Music
The built-in tracks in Add Music are original compositions generated by FreeClips in your browser. You may use them in any video, including commercial ones; no credit is needed.
Test media
Our automated tests use synthetic speech and test patterns, plus a NASA astronaut portrait (U.S. government work, public domain). None of it ships with the site.
A machine-generated list of every production dependency and its declared license is kept with the source code (docs/dependency-licenses.json).