Notice about FFmpeg (GNU GPL v3)
The FFmpeg included in the installer is the “essentials” build from gyan.dev (version 7.1), distributed under the GNU General Public License, version 3 (GPL v3). It ships as a separate program (ffmpeg.exe), which Klicai calls to cut, join, encode, and mix the videos.
The FFmpeg source code is at ffmpeg.org and on the gyan.dev builds page. You may use, study, copy, and redistribute FFmpeg under the terms of the GPL v3.
Voice helper (GPL-3.0)
The local voice uses the voice helper klicai-voz.zip, a separate open-source program under the GPL-3.0 license. It does not come in the installer: it is downloaded separately, from the Klicai Store, only when you ask for the local voice. The package includes the source code, a README, and the license text (LICENSE-GPL-3.0.txt).
Inside it are: phonemizer (GPL-3.0), espeak-ng (GPL-3.0), kokoro-onnx (MIT), ONNX Runtime (MIT), and NumPy (BSD).
Inside the installer
Components that come in the installer of version 1.10.0. The licenses are those stated by each project; the official text of each one is on the project's website.
| Component | Version | What it's for | License |
|---|---|---|---|
FFmpeg (ffmpeg.exe, “essentials” build from gyan.dev) | 7.1 | Cuts, joins, encodes, and mixes the videos | GPL-3.0 (see the notice above) |
| Qt for Python / PySide6 (Qt 6) | 6.12.0 | Interface, player, and PDF | LGPL-3.0 |
| Qt Multimedia FFmpeg libraries | — | Video player | LGPL, according to the Qt project's build |
PyAV (av) | 19.0.1 | Audio for faster-whisper | BSD-3-Clause; the FFmpeg libraries that come with it follow the FFmpeg license, according to the PyAV project's build |
| faster-whisper | 1.2.1 | Local transcription | MIT |
| CTranslate2 | 4.8.2 | Whisper engine | MIT |
OpenCV (opencv-python) | 5.0.0.93 | Video and face detection | Apache-2.0 |
YuNet model (face_detection_yunet_2023mar.onnx) | 2023mar | Face detection | MIT, according to the OpenCV Zoo project |
LiteRT (ai-edge-litert) | 2.3.0 | Runs the YAMNet model | Apache-2.0 |
| YAMNet model (Google) | — | Detects laughter, applause, and shouts | Apache-2.0 |
| ONNX Runtime | 1.31.0 | Runs the U²-Net model | MIT |
U²-Net small model (u2netp.onnx) | — | Character and background cutout | Apache-2.0 |
RNNoise model (sh.rnnn) | — | Removes noise from the voice | RNNoise: BSD-3-Clause; the sh.rnnn model: according to the rnnoise-models project |
| yt-dlp | 2026.8.19 (the app updates it automatically) | Link downloads | Unlicense (public domain) |
| NumPy | 2.5.4 | Calculations | BSD-3-Clause |
| Pillow | 12.3.0 | Images and captions | MIT-CMU (HPND) |
| tokenizers (Hugging Face) | 0.23.3 | Whisper | Apache-2.0 |
| huggingface_hub and hf-xet | 2.2.0 / 1.7.0 | Downloading models | Apache-2.0 |
anthropic SDK (with pydantic, httpx2, and jiter) | 1.13.0 | Claude AI | anthropic: MIT; pydantic: MIT; httpx2: BSD; jiter: according to the project |
| cryptography | 50.0.2 | License signing | Apache-2.0 or BSD |
| certifi | 2026.10.23 | Secure connection (HTTPS) certificates | MPL-2.0 |
| protobuf, PyYAML, tqdm, and click | — | Dependencies | protobuf: BSD-3-Clause; PyYAML: MIT; tqdm: MPL-2.0 and MIT; click: BSD-3-Clause |
| Press Start 2P font | — | Pixel art style | SIL Open Font License 1.1 |
| Python | 3.x | The app's language | PSF License |
PyInstaller (.exe loader) | 6.22.3 | Program packaging | GPL-2.0 with an exception that allows closed-source programs |
| Inno Setup | 6 | Installer | Inno Setup license (allows commercial use) |
Downloaded later, separate from the installer
The app downloads these items separately, on its own or when you ask:
| Component | When it is downloaded | License |
|---|---|---|
| Whisper models (tiny, base, small, and medium; faster-whisper by Systran, on Hugging Face) | On their own, at the first transcription | MIT |
Voice helper klicai-voz.zip (Klicai Store, about 60 MB) | When you download the local voice | GPL-3.0 (see above) |
| Kokoro-82M voice model (GitHub, kokoro-onnx project, about 120 MB) | When you download the local voice | Apache-2.0 |
| yt-dlp updates (GitHub) | On their own, every 3 days | Unlicense (public domain) |
Photos, videos, and sounds
- Pexels and Pixabay: the photos and videos from the banks follow each bank's licenses (commercial use is allowed, with rules). The app stores the author's name and puts the credit in the description. Check the rules at pexels.com/license and pixabay.com.
- Sounds: the sound effects are made by code, inside the app itself, with no third-party sound bank.
Questions
To request a copy of a license or ask questions, contact us at contato.klicai.com.br. See also the Terms of Use and the Privacy Policy.
Back to KlicTube
From scratch to a monetized channel — in a Windows app.