Pre-release list. The final component list and source links will be published with the download.

Clipper X

Third-party licenses and source notices

Clipper X is built with the third-party software below. Each component stays under its own license.

This is not the final list. It covers the main components planned for release and leaves out smaller supporting packages. The complete list for the released version will be published with the download.

Exact timing: non-commercial only

Exact timing is an optional download that is not selected by default. It uses the MMS_FA word-timing model, licensed CC BY-NC 4.0, so it may not be used for commercial purposes. Clipper X being free does not remove that condition, and it applies to everyone who uses the model. Whether a clip on a monetized channel counts as commercial use is something each user needs to consider. If in doubt, leave Exact timing off.

Attribution: Scaling Speech Technology to 1,000+ Languages, Vineel Pratap et al., Meta AI, 2023. Published through the fairseq project and accessed, unmodified, through torchaudio.

FFmpeg source notice

The Windows build includes FFmpeg, unmodified, as separate programs. This FFmpeg build is licensed under the GNU GPL version 3 or later. See FFmpeg’s legal page, the FFmpeg source and the build recipe project.

The exact source archive and build recipe for the released version will be linked here with the download. PyAV’s bundled libraries, which include GPL components, will be covered in the same notice.

Included in the app

ComponentLicense
FFmpegGPL v3 or later
PyAVBSD-3-Clause; bundled libraries include GPL components
faster-whisperMIT
CTranslate2MIT
ONNX RuntimeMIT
tokenizersApache 2.0
NumPyBSD-3-Clause, 0BSD, MIT, Zlib and CC0-1.0
opencv-python-headlessApache 2.0, with bundled third-party notices
PillowMIT-CMU
yt-dlp (Python package)Unlicense
yt-dlp-ejsUnlicense, MIT and ISC
Node.jsMIT, with bundled licenses
Anthropic Python SDKApache 2.0
pywebviewBSD-3-Clause
Python.NET (pythonnet)MIT
certifiMPL 2.0
PyInstallerGPL v2 or later, with an exception for packaged apps

Speech models

ModelLicense
distil-large-v3 (Hugging Face distil-whisper), CTranslate2 conversion by SystranMIT
Whisper small.en (OpenAI), CTranslate2 conversion by SystranMIT

Optional Quality Pack

These parts are downloaded only if you choose them in the app.

ComponentLicense
PyTorchBSD-3-Clause
torchaudioBSD-2-Clause for code
WAV2VEC2_ASR_BASE_960H word-timing model (Accurate timing)MIT
MMS_FA word-timing model (Exact timing)CC BY-NC 4.0, non-commercial only
NVIDIA CUDA, cuBLAS and cuDNN runtime packagesNVIDIA proprietary license

The Accurate timing model comes from wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations, Alexei Baevski et al., 2020, and was trained on LibriSpeech (CC BY 4.0). See the torchaudio documentation and the fairseq MIT license.

Fonts

All eight fonts are available as caption fonts in the app. This website serves a Latin subset of Archivo Black, and the font list on the home page shows each name as a vector drawing of its typeface.

FontLicenseCopyright
Archivo BlackSIL OFL 1.1The Archivo Black Project Authors
AntonSIL OFL 1.1The Anton Project Authors
Bebas NeueSIL OFL 1.1Dharma Type
BangersSIL OFL 1.1The Bangers Project Authors
Lilita OneSIL OFL 1.1Juan Montoreano, Reserved Font Name “Lilita”
Titan OneSIL OFL 1.1Rodrigo Fuenzalida, Reserved Font Name “Titan”
Russo OneSIL OFL 1.1Jovanny Lemonad, Reserved Font Name “Russo”
Luckiest GuyApache 2.0Astigmatic

Questions

Create a support ticket in our Discord.