Neural Companion
GitHub

Avatar runtime

MuseTalk

MuseTalk provides local avatar video generation from prepared avatar packs and runs in an isolated environment because it is Neural Companion’s most hardware-sensitive avatar mode.

Requirements

  • An NVIDIA CUDA GPU.
  • The isolated MuseTalk runtime and expected model weights.
  • A prepared avatar pack; large demo packs are distributed separately.

Configuration

  1. Install Neural Companion with the isolated MuseTalk runtime selected.
  2. Install or prepare an avatar pack under avatar_packs/<pack_id>/.
  3. Select MuseTalk as the avatar engine, then use Dry Run and performance profiles to tune the machine.

Verify in the project source →

Feature

AI companion for Windows

Neural Companion is a desktop application whose current primary release targets Windows and provides guided installation, first-run tutorials, and configurable runtime presets.

Explore AI companion for Windows

Feature

Voice AI companion

Neural Companion combines speech input with selectable text-to-speech addons, including Chatterbox, Gemini TTS Preview, PocketTTS, and other addon-provided backends.

Explore Voice AI companion

Feature

AI avatars

Choose no-avatar mode or connect Neural Companion to MuseTalk, Scenic image packs, VSeeFace, or VaM, with each avatar mode configured and enabled independently.

Explore AI avatars