PRESETKEY

Notices

Updated November 3, 2026.

PresetKey uses preset packs made by other creators, AI models and open-source software. Each one is listed here with who made it and the licence it is used under. Those licences apply to those parts. The rest of PresetKey is covered by the terms.

Preset packs

Xsaandi preset pack by Xsaandi. Presets in PresetKey's profiles. Licence: Included with the creator's permission. Included in the PresetKey download.

Finzar transitions by Finzar. Transition presets. Licence: Included with the creator's permission. Included in the PresetKey download.

Speech models (Auto-captions)

Parakeet TDT 0.6B v2 and v3 by NVIDIA Corporation. Speech recognition: v2 for English, v3 for 25 European languages. Licence: CC BY 4.0. Downloaded from its publisher the first time a feature needs it. Copyright (c) NVIDIA Corporation. Used in the ONNX export made by the k2-fsa sherpa-onnx project, int8 quantised (and, for v2, fp32), with no other change. Original models: huggingface.co/nvidia/parakeet-tdt-0.6b-v2 and huggingface.co/nvidia/parakeet-tdt-0.6b-v3. Source: huggingface.co/nvidia/parakeet-tdt-0.6b-v3.

Whisper (tiny and large-v3-turbo) by OpenAI. Language identification, and the all-languages pack. Licence: MIT. Downloaded from its publisher the first time a feature needs it. Copyright (c) 2022 OpenAI. Used in the sherpa-onnx export. Source: github.com/openai/whisper.

SenseVoice Small by Alibaba Group (FunAudioLLM / FunASR). The Chinese, Japanese and Korean pack. Licence: FunASR Model Open Source License 1.1. Downloaded from its publisher the first time a feature needs it. Converted to ONNX (int8) by the k2-fsa sherpa-onnx project. Source: huggingface.co/FunAudioLLM/SenseVoiceSmall.

Silero VAD v5 by Silero Team. Finding where speech starts and stops. Licence: MIT. Downloaded from its publisher the first time a feature needs it. Source: github.com/snakers4/silero-vad.

pyannote segmentation-3.0 by pyannote. Telling speakers apart. Licence: MIT. Downloaded from its publisher the first time a feature needs it. Source: huggingface.co/pyannote/segmentation-3.0.

TitaNet small by NVIDIA Corporation (NeMo). Telling speakers apart. Licence: Apache-2.0 (the NeMo toolkit licence). Downloaded from its publisher the first time a feature needs it. Copyright (c) NVIDIA Corporation. Used in the sherpa-onnx export, with no other change. Source: catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/titanet_small.

Image models

BiRefNet (general) by ZhengPeng (copyright 2024). Background removal on graphics cards with 9 GB or more. Licence: MIT. Downloaded from its publisher the first time a feature needs it. ONNX file from onnx-community/BiRefNet-ONNX (MIT) or danielgatis/rembg (MIT). Source: github.com/ZhengPeng7/BiRefNet.

IS-Net (isnet-general-use, from DIS) by Xuebin Qin and contributors. Background removal. Licence: Apache-2.0. Downloaded from its publisher the first time a feature needs it. ONNX file from danielgatis/rembg (MIT) or its Hugging Face mirrors. Source: github.com/xuebinqin/DIS.

LaMa by Samsung AI Center Moscow (advimman). Filling in removed areas. Licence: Apache-2.0. Downloaded from its publisher the first time a feature needs it. Source: github.com/advimman/lama.

Real-ESRGAN by Xintao Wang (copyright 2021). Upscaling images. Licence: BSD-3-Clause. Downloaded from its publisher the first time a feature needs it. Models RealESRGAN_x4plus, RealESRGAN_x2plus, realesr-animevideov3 and RealESRGAN_x4plus_anime_6B, in ONNX conversions (BSD-3-Clause). Source: github.com/xinntao/Real-ESRGAN.

Software

AutoHotkey 2.0.19 by AutoHotkey contributors (Chris Mallett, Steve Gray and others). Runs PresetKey's keyboard helper as a separate program. Licence: GPL-2.0. Included in the PresetKey download. helper/AutoHotkey64.exe, unmodified, with its licence beside it. Its source code is at the link below. For three years from the day you got PresetKey, PresetKey sends a copy on request: email ibra@presetkey.com with your PresetKey version. Source: github.com/AutoHotkey/AutoHotkey/tree/v2.0.19.

yt-dlp 2026.08.19 by yt-dlp contributors. Downloading media from a link. Licence: The Unlicense, with parts under their own licences. Included in the PresetKey download. bin/yt-dlp.exe, the official Windows release, unmodified. It contains Python, OpenSSL, certifi, mutagen (GPL-2.0-or-later), requests, urllib3, websockets, brotli, PyCryptodome, curl_cffi and others, listed at the licence link. The same source-code offer as AutoHotkey covers its GPL parts. Source: github.com/yt-dlp/yt-dlp/tree/2026.08.19.

Node.js by Node.js contributors. Runs parts of the installer and helper. Licence: MIT. Included in the PresetKey download. The official Windows release, unmodified. It also contains V8, libuv, OpenSSL, ICU, zlib and other parts under their own licences, listed at the licence link. Source: nodejs.org.

sherpa-onnx by k2-fsa (Next-gen Kaldi). Runs the speech models. Licence: Apache-2.0. Downloaded from its publisher the first time a feature needs it. Source: github.com/k2-fsa/sherpa-onnx.

ONNX Runtime by Microsoft. Runs the image and voice models. Licence: MIT. Downloaded from its publisher the first time a feature needs it. Source: github.com/microsoft/onnxruntime.

NumPy by NumPy developers. Audio processing. Licence: BSD-3-Clause. Downloaded from its publisher the first time a feature needs it. Source: numpy.org.

Python by Python Software Foundation. Runs the Auto-captions engine. Licence: PSF License. Downloaded from its publisher the first time a feature needs it. Source: www.python.org.

FFmpeg 9.0.1 by The FFmpeg developers, essentials build by Gyan Doshi. Reading audio from your media. Licence: GPL-3.0 (this build). Downloaded from its publisher the first time a feature needs it. Downloaded unmodified from gyan.dev by the captions setup, and run as a separate program. Source code is at ffmpeg.org. Source: ffmpeg.org.

pip by The pip developers. Installs the captions engine's packages. Licence: MIT. Downloaded from its publisher the first time a feature needs it. Source: github.com/pypa/pip.

NVIDIA CUDA runtime, cuBLAS, cuFFT and cuDNN by NVIDIA Corporation. GPU speech recognition, only with the NVIDIA card option. Licence: NVIDIA software licence agreements. Downloaded from its publisher the first time a feature needs it. Installed from PyPI by the captions setup.

Media

Red Eclipse 1.5 gameplay by Premitre. Sample footage in the tutorial and the preset previews. Licence: CC BY 3.0. Included in the PresetKey download. Excerpted, cropped, scaled and re-encoded for the tutorial clips, the preset previews and the hover card. Red Eclipse is a free and open-source game. Source: commons.wikimedia.org/wiki/File:Red_Eclipse_1,5_Gameplay_1.webm.

Fonts

Anton, Bangers, Bebas Neue, Inter, Montserrat, Poppins and Roboto by their authors, named in each font's OFL.txt. Caption fonts. Licence: SIL Open Font License 1.1. Included in the PresetKey download. Unmodified. Montserrat comes from github.com/JulietaUla/Montserrat. Source: github.com/google/fonts.

Geist by The Geist Project Authors (copyright 2024). PresetKey's wordmark and headings. Licence: SIL Open Font License 1.1. Included in the PresetKey download. Source: github.com/vercel/geist-font.

Questions

If you made something listed here and want the credit changed, or think something is missing, email ibra@presetkey.com.

Back to PresetKey