Open-source notices

VerseCut 3.0 · September 2026 · Stateam LLC

VerseCut's own code is © Stateam LLC. It is built on open-source software, and we thank the people who make it. This page lists every third-party component, how VerseCut uses it, and how to get its source.

The video engine (FFmpeg)

Windows macOS
What ships ffmpeg.exe, ffprobe.exe and their libraries, in setup\ffmpeg ffmpeg and ffprobe (Apple silicon and Intel), inside VerseCut.app
Build BtbN FFmpeg-Builds ffmpeg-n9.0-latest-win64-lgpl-shared-9.0, FFmpeg version n9.0.2-3-ga5923073bf FFmpeg 9.0.2 with LAME 3.100, built by Stateam LLC with the script published beside the source
Licence GNU LGPL version 3 (no GPL or non-free components) GNU LGPL version 2.1 or later (no GPL or non-free components)
Not included x264, x265, fdk-aac or any other GPL or non-free library the same; network protocols are also switched off

FFmpeg is included in the VerseCut download and runs as a separate program. It is not linked into VerseCut and it is not modified. You may replace it with your own build of the same libraries by putting your ffmpeg and ffprobe in VerseCut's runtime/ffmpeg folder.

Source code. The complete corresponding source for both builds is published with VerseCut:

They are at https://downloads.versecut.net/source/index.html (linked from the VerseCut download page as Source code for open-source components). We will also send any of it on request, at no charge beyond the cost of media and postage, for at least three years after we last distributed this version: email bjohn@gostateam.com. FFmpeg is at https://ffmpeg.org. Full licence texts are in docs/licences inside VerseCut.

H.264 and AAC

VerseCut encodes H.264 video and AAC audio only with the encoders that are part of your operating system, which the operating system vendor licenses:

VerseCut does not include or use x264 or FFmpeg's own AAC encoder. MP3 export uses LAME.

Speech-to-text and other components

These are downloaded by the installer from the Python Package Index at the exact versions listed in setup/requirements.txt, and installed only inside VerseCut's private folder. faster-whisper is installed without its PyAV dependency, which VerseCut does not use.

Component Version Licence Used for
faster-whisper (SYSTRAN) 1.2.1 MIT Offline speech-to-text
OpenAI Whisper model small.en (converted by SYSTRAN) — MIT Speech model, downloaded at setup
CTranslate2 4.8.2 MIT Runs the speech model
ONNX Runtime 1.19.2 MIT Voice-activity detection
tokenizers 0.22.2 Apache 2.0 Speech model text
Hugging Face Hub, hf-xet 1.8.0, 1.6.0 Apache 2.0 Downloads the model once
NumPy 2.0.2 BSD 3-clause Audio arrays
Pillow 12.3.0 MIT-CMU (HPND) Draws burned-in captions
certifi, tqdm 2026.7.22, 4.70.1 MPL 2.0 (tqdm: MPL 2.0 and MIT) Unmodified; source at https://pypi.org
anyio, annotated-doc, coloredlogs, filelock, h11, httpx*, humanfriendly, markdown-it-py, mdurl, PyYAML, rich, typer pinned MIT (*httpx: BSD 3-clause) Supporting libraries
flatbuffers, packaging, protobuf, fsspec, httpcore, idna, mpmath, Pygments, sympy, shellingham, typing_extensions pinned Apache 2.0, BSD, ISC or PSF Supporting libraries
Python 3.12 PSF License Runs VerseCut's engine (Windows: from python.org; macOS: python-build-standalone, inside VerseCut.app)
Source Sans 3 3.052 SIL Open Font License 1.1 Caption lettering (bundled)
Lora, Poppins, Source Sans 3, JetBrains Mono — SIL Open Font License 1.1 Brand typography on the website

The licence text of each Python package is in its folder inside VerseCut's private Python after setup.

Trademarks

VerseCut is a trademark of Stateam LLC. FFmpeg is a trademark of Fabrice Bellard. Other names are trademarks of their owners.