QClips - third-party software
============================

FFmpeg (ffmpeg\QClips Engine.exe = ffmpeg.exe, ffmpeg\QClips Probe.exe = ffprobe.exe, and the FFmpeg DLLs)
  QClips runs FFmpeg as a separate program, built from the FFmpeg source with two small patches and only the parts
  QClips uses. Because it includes x264, this FFmpeg build is licensed under the GNU General Public License
  version 2 or later; the license text is in ffmpeg\LICENSE.txt, and ffmpeg\README.txt lists the exact source
  archive, its SHA-256, the library versions and the configure line of the bundled build, and contains the patches
  in full. The source of this build is that archive plus those patches.
  The patches (both to the scale_d3d11 filter):
    0001-scale_d3d11-single-textures.patch: the filter makes its output pictures as single textures, as FFmpeg 8.0
      did; FFmpeg 8.1 makes them as one texture array, which some graphics drivers refuse.
    0002-scale_d3d11-bt709-limited-output.patch: the filter converts RGB pictures to BT.709 limited-range YUV
      instead of leaving the conversion to the graphics driver.
  FFmpeg source code: https://ffmpeg.org/releases/ffmpeg-8.1.3.tar.xz
    (SHA-256 7138d28c96d9d3e3af4ee3d8cad72741f8ffb40da90c1112235dea3ecd3178a3), https://github.com/FFmpeg/FFmpeg
  Built with MSYS2 MinGW-w64 (https://www.msys2.org) and these libraries, linked into the FFmpeg DLLs:
    x264 - GNU GPL v2 or later - https://code.videolan.org/videolan/x264
    dav1d - BSD 2-Clause - https://code.videolan.org/videolan/dav1d
    Intel VPL dispatcher (libvpl) - MIT - https://github.com/intel/libvpl
    NVIDIA codec headers (nv-codec-headers) - MIT - https://github.com/FFmpeg/nv-codec-headers
    AMD AMF headers - MIT - https://github.com/GPUOpen-LibrariesAndSDKs/AMF
    zlib - zlib License - https://zlib.net
    GCC runtime libraries (libgcc, libstdc++, winpthreads) - GPL with the GCC Runtime Library Exception / MIT
  Configure line (FFmpeg 8.1.3):
    --prefix=/ffmpeg --target-os=mingw32 --arch=x86_64 --enable-gpl --enable-shared --disable-static
    --disable-autodetect --disable-everything --disable-doc --disable-ffplay --disable-network --disable-debug
    --enable-w32threads --enable-d3d11va --enable-ffnvcodec --enable-nvenc --enable-amf --enable-libvpl
    --enable-libx264 --enable-libdav1d --enable-zlib --pkg-config-flags=--static
    --extra-ldflags='-L../deps -static-libgcc -static-libstdc++' --extra-libs=-lstdc++ --extra-version=qclips
    plus --enable-<kind>=<names> for every component in the list below and --enable-filter=fsync.
  Components: the filters, encoders, decoders, formats, devices, protocols, bitstream filters, parsers and
  hardware decoders QClips uses (the full list and the exact line are in ffmpeg\README.txt).
  The bundled build's own version line follows at the end of this file.

NAudio - MIT License - https://github.com/naudio/NAudio
Serilog - Apache License 2.0 - https://serilog.net
Microsoft.Data.Sqlite / SQLite - MIT License / Public Domain
Windows App SDK, Win2D, CommunityToolkit.Mvvm - MIT License (Microsoft)
.NET runtime, Microsoft.Diagnostics.Runtime (ClrMD) - MIT License (Microsoft) - https://github.com/dotnet
Velopack (installs and updates) - MIT License - https://github.com/velopack/velopack
Markdig (release notes) - BSD 2-Clause License - https://github.com/xoofx/markdig

Voice commands
  sherpa-onnx (sherpa-onnx.dll, sherpa-onnx-c-api.dll) - Apache License 2.0 - https://github.com/k2-fsa/sherpa-onnx
  ONNX Runtime (onnxruntime.dll) - MIT License - https://github.com/microsoft/onnxruntime
  sherpa-onnx-c-api.dll and onnxruntime.dll are downloaded with the models (into Models\runtime), not installed with QClips.
  The models are downloaded from Settings -> Voice commands into %LOCALAPPDATA%\QClips.Data\Models, each with its license:
  Silero VAD (silero_vad.onnx) - MIT License - https://github.com/snakers4/silero-vad
  Moonshine tiny English - MIT License, Copyright (c) 2025 Useful Sensors, Inc. (dba Moonshine AI) - https://github.com/moonshine-ai/moonshine
  Moonshine base Arabic - Moonshine AI Community License (moonshine-base-ar\LICENSE.txt and NOTICE.txt next to the model):
    "This Moonshine AI Model is licensed under the Moonshine AI Community License, Copyright (c) Moonshine AI Ltd. All Rights Reserved"
    Powered by Moonshine AI.