Archidexter/adlindel-assets

AD.Lindel assets

Public download mirror for AD.Lindel — a launcher that sets up the AI-dialogue mods SkyrimNet and Mantella on Skyrim builds. Copies of original files (Whisper, all-MiniLM, Ollama) are a fallback: the launcher tries the original source first and comes here only when it is down, throttled or hits a daily quota. Ready-to-run bundles that have no single original download (the voice servers, the Piper voice pack, our Mantella build) are downloaded from here. Everything here is free to download, no registration.

Each file stays under its own authors' license, listed below. All credit goes to them.

Files

File	What it is	Source	License
SkyrimNet_XTTS.7z	Local XTTS voice server. The skyrimnet-xtts.exe build and its files are the unmodified release v0.1.10 of langfod/skyrimnet-tts (SHA-256 of the exe: db0782a05c30db265d521870f56694d20a9d9f9e032969b62fc71a55aadd9f9b, identical to the author's release). Added: the Coqui XTTS-v2 model weights, voice latents for Russian speech, and Start_RU.bat (the same start with a UTF-8 console).	langfod; Coqui	Server: based on coqui-ai-TTS (MPL-2.0). Model: Coqui Public Model License 1.0.0 — non-commercial
SkyrimNet_OmniTTS.7z	Local OmniTTS voice server: jagerius/OmniTTS, the k2-fsa/OmniVoice model, an embedded Python 3.12 and our small bridge script to SkyrimNet (adl_tts_proxy.py).	jagerius; k2-fsa	OmniVoice code Apache-2.0, model weights CC BY-NC (non-commercial). Python — PSF License
SkyrimNetPiperModels.zip	Russian voices for SkyrimNet's built-in Piper TTS, packaged as a Mod Organizer 2 mod. Stock rhasspy/piper-voices ru_RU models under SkyrimNet voice names: FemaleEvenToned = irina, MaleCommoner = ruslan. Includes espeak-ng data.	rhasspy	irina — voice data from RHVoice (personal, non-commercial); ruslan — RUSLAN corpus, CC BY-NC-SA 4.0; espeak-ng — GPL-3.0; piper — MIT
mantella/SSE_Lindel_Mantella.7z	Our build of Mantella for Skyrim from open sources. Game part from art-from-the-machine/Mantella-Spell with script sources; SKSE_HTTP.dll from Leidtier/SKSE_HTTP; the Mantella program from art-from-the-machine/Mantella (commit c0c1ae6) with our changes and Piper voices. The complete program sources and the list of our changes (НАШИ ПРАВКИ.txt) are inside the archive, in the sources folder next to Mantella.exe.	art-from-the-machine; Leidtier	Program — AGPL-3.0; game part and SKSE_HTTP — MIT
models/ggml-small.bin, models/ggml-large-v3.bin, models/ggml-large-v3-turbo.bin	Whisper speech-recognition models, unmodified copies of ggerganov/whisper.cpp (SHA-256 identical to the original).	OpenAI; ggerganov	MIT
models/all-MiniLM-L6-v2.onnx	Embedding model, unmodified copy of onnx-models/all-MiniLM-L6-v2-onnx model.onnx (SHA-256 identical), from sentence-transformers/all-MiniLM-L6-v2.	sentence-transformers	Apache-2.0
tools/ollama-windows-amd64.zip	Unmodified copy of the Ollama v0.33.2 Windows archive (SHA-256 identical to the GitHub release asset). License text: tools/ollama-LICENSE.txt.	Ollama	MIT

Non-commercial materials

The XTTS-v2 model, the OmniVoice model and the Piper voices irina and ruslan allow non-commercial use only. They are offered here free of charge; nobody is charged for them. Use them and the speech they produce for personal, non-commercial purposes. The Coqui Public Model License text is in LICENSE.

For rights holders

If you are the author of a file mirrored here and do not want it hosted, open an issue at github.com/Archidexter/AD.Lindel/issues or write on Boosty — the copy will be removed and the launcher will use your source only.