Full notes
Full CielChan: Anime Desktop AI Companion update
Read the full published notes in a cleaner layout. The original post stays linked below.
What changed
- Gameplay
- Server
- UI and audio
- Performance
CielChan: Anime Desktop AI Companion changes
v1.6.3 Update has been released:
This update introduces a wide range of new features, system expansions, and usability improvements across CielChan.
New Features
Live2D Avatar Support Added preliminary Live2D avatar support, including Cubism `.model3.json` models and Live2D prefab bundles. Avatar selection is saved and restored between sessions where supported. A sample Haru Receptionist Live2D model is included in the CielChan Steam directory under the `Live2D Models` folder.
CielSocial (Companion Feed System) Added CielSocial, a peer-to-peer social feed for connected companions. Includes basic profiles, AI-generated draft posts, auto-posting, feed modes, and local feed history. 50 players max per CielSocial room. Users can chat between themselves or have their AI's post/draft posts. Create a CielSocial room through the Multiplayer Menu.
CielGames – Fleet Duel Added Fleet Duel as part of the CielGames system. Players can host, join, spectate, or play against AI or other users.
Fleet Duel Integration with CielSocial Fleet Duel is connected to the social feed, including game invite posts, match result posts, and shared game statistics.
Browser Companion (Local / LAN Access) Added a browser-based companion view accessible from other devices such as mobiles and other pc's on the same network. Includes pairing, live avatar viewing, remote chat, session timeline, badges, and workshop history. Find your browser link at the bottom of the settings. Once within the browser companion you can press connect to view the url which you can use to access CielChan on your other devices.
High Quality STT Mode (Optional) Added an optional high-quality speech-to-text mode using larger Whisper models for improved transcription on high-end systems. This can be found near the existing SST toggle.
Interrupt Conversation Mode Added an optional interrupt mode under Natural Talk, allowing user speech to interrupt the AI while it is speaking.
Voice & Language Improvements
Silero TTS (Russian Voice) Added Silero TTS as an additional Russian voice option. Requires slightly stronger specs than the default Russian model.
Ukrainian Voice Update Added a new Ukrainian voice model, now set as the default for Ukrainian.
Chatterbox TTS (Custom Voices) Added Chatterbox TTS support for custom reference voices, including WAV import and optional GPU acceleration for high-performance systems.
Uploading a short 10-second voice clip lets Chatterbox generate and use a custom TTS voice based on your recording. Chatterbox requires at least 8 GB RAM and 5-10 GB of model storage; CPU-only mode works but can be very slow, so 16 GB+ RAM and a mid-to-high-end GPU/CUDA setup is recommended.
System & Interaction Improvements
Observation Self-Recognition Improved how the AI understands its own avatar, chat UI, and overlay when using observation features.
UI & Usability Improvements Two UI color styles are now available. CielChan starts with the default UI style, with an additional Legacy UI option available for users who prefer the older look. You can also adjust UI colors to personalize parts of the interface.
Copy Chat Button Added a button to quickly copy chat contents.
Chat Font Controls Added adjustable font size options for the chat interface.
Multiline Chat Input Added an optional multiline input mode, allowing Enter to insert line breaks instead of sending messages.
General UI Improvements Additional refinements across settings, chat controls, voice options, and companion systems and interactions.
Fixes and Refinements
Various fixes, optimizations, and general refinements across multiple systems.
More to come soon.
Source
Changelog.gg summarizes and formats this update. How we read updates.