Onyx Voice speech helper - complete corresponding source ======================================================== The onyx-voice-speech package (in the Onyx Voice 0.2.16 images and as a .deb) holds onyx-speechd, the program that turns text into speech for Onyx Voice. It is licensed under the GNU General Public License version 3 or later (LICENSE in its source). It is a separate program: Onyx Voice starts it and exchanges text and audio files with it. onyx-speechd-0.2.16.tar.gz its source (Program.cs, the project file, LICENSE, NOTICE), the build settings it inherits (Directory.Build.props / .targets) and the package recipe (speech/build-deb.sh, tts-model.txt, kokoro-metadata.py) sherpa-onnx-1.13.8.tar.gz sherpa-onnx 1.13.8 (Apache-2.0), the speech library it uses through the org.k2fsa.sherpa.onnx 1.13.8 NuGet packages espeak-ng-ed530aa113046142eb5115cf2fc9157854d0ffe1.zip espeak-ng (GPL-3.0-or-later) at the commit sherpa-onnx 1.13.8 builds into its native library (cmake/espeak-ng-for-piper.cmake) piper-phonemize-f3ff95afc03640bc1399e113e83361192a2fafb4.zip piper-phonemize (MIT), likewise (cmake/piper-phonemize.cmake) To rebuild: dotnet publish src/Onyx.Voice.Speechd -c Release -r linux-x64 --self-contained (with the .NET 10 SDK), then speech/build-deb.sh on Ubuntu Server 24.04; it downloads the voice model files listed in tts-model.txt and checks their SHA-256. The native library can be rebuilt from the sherpa-onnx source (its build downloads the espeak-ng and piper-phonemize archives above; check them against SHA256SUMS). The voice model (Kokoro-82M, Apache-2.0) and the other components are listed with their licences in NOTICE. Written offer: for at least three years after we last distribute a release, Integrators Resource will give anyone who asks a copy of the complete corresponding source of the GPL software in it, for no more than our cost of providing it. Contact: https://larstrom.com/#contact Integrators Resource, 2210 4th Ave S, Minneapolis, MN 55404, USA