Novolis 

Rendering composition stack — materials, scene compile, CPU/GPU backends, presentation contracts.

Generated novolis_audio P/Invoke (manifest-driven; do not hand-edit Interop/*.g.cs).

Present CPU ray-traced frames via Raylib textures (host adapter).

KokoroSharp CPU offline TTS adapter for Novolis.Audio.Voice.

Shared path-tracing demo scenes, async workers, and session helpers for Silk and Raylib samples.

Voice synthesizer backend selection and platform speech options for OS TTS adapters.

Inno Setup script generation helpers for Novolis Avalonia installers.

Immutable local-time ranges and duration primitives.

Lightweight multi-track arrangement editor (Music Maker / Audacity–style): library, tracks, fades, mix export.

Browse curated free/CC media collections, download/cache licensed items, and run explore transformers (PCM decode, audio→MIDI sketch). Does not scrape commercial catalogs (Artlist, Epidemic, …).

Game audio output contract (master volume, device probe).

ICAO-style phraseology normalization for voice pipelines.

TUnit ceilings for elapsed time, allocated bytes, and throughput, plus a BenchmarkDotNet JSON highlight reader.

Sherpa-ONNX offline TTS/STT/VAD with bundled Piper and speech models.

Azure Table Storage IRepository provider.

Avalonia-free CadDocument → SceneDocument bridge: tessellate solids/walls/spaces into .nov3djson meshes (Novolis.Modeling.Scene).

Semantic vertical section document model with content, columns, tables, page chrome, and immutable authoring builders.

Typed .NET solution workspace contracts over an I/O workspace root.

UI-neutral contracts and records for Novolis PDF reading.

Named, manually invokable workflow pipelines for .NET applications.