Hello Humans! Patch 3 is here. We've made some critical adjustments to our voice services to ensure Neko sounds as sweet as ever:
📢 Voice Service Status
Non-Mainland Users: Starting today, free TTS for users outside Mainland China will be routed through USA-based Gemini TTS. You should experience better stability and multilingual support, nyan!
Mainland China Users: We are aware of widespread 503 errors affecting TTS services in the Mainland region, originating from our upstream provider (StepFun). The Dev Nekos are currently in communication with them to resolve this as soon as possible.
🛠️ Core Fixes & Enhancements
Kernel i18n Overhaul: Refactored our internationalization logic for a more consistent language experience (no more accidental language switching during chats!). - We'll watch the impact closely!
Screen Resolution & "Emergency Summon": Continued fixes for non-1080p displays. If Neko gets lost, Right-click the Tray icon -> Select "Reset Position" to bring her back.
Agent Performance Boost: Resolved a critical bug affecting Agent features for paid API users. We are also improving the responsiveness of task cancellation to prevent background processes from blocking the UI.
✨ New Features & Interactions
Eye Tracking Toggle: You can now choose whether Live2D and VRM models should track your mouse cursor.
Voice Cloning Upgrade: Added an "Auto-Trim Audio" option to remove long silences from reference clips, ensuring better cloning quality.
Interrupt Logic Fix: Polished the "Interrupt Voice" feature in text mode for a smoother experience.
🚀 Experimental & WIP (Work In Progress)
Multi-platform Support: Full platform builds (Win/Mac/Linux) are now available for testing on GitHub Nightly!
Workshop Loading Fixes: We’ve implemented several fixes to address model loading issues and are monitoring the results, nyan.
Personalized Feeds Status: Integration for social media personalized streams is still in progress. Thank you for your patience!