AI Features
MiuAi, voice features, and local AI.
11 articles
What AI features are available in the launcher? The launcher includes several optional AI systems you can set up from the AI hub: • Text-to-Speech (TTS) - High-quality AI voice synthesis powered by IndexTTS-2. • Speech-to-Text (Whisper) - Real-time voice recognition powered by faster-whisper. • Local AI Chat (Ollama) - Download and run large language models (LLMs) entirely on your own device. • MiuAi (beta) - Cloud AI chat hosted by us, included with Tier 4 and higher subscriptions - no local GPU needed. • Bring Your Own Key (BYOK) - Connect your own AI provider account and let Miu's brain run in the cloud on your terms. Everything is optional - the local features run 100% on your own device, and the cloud options only ever talk to the service you chose. What is MiuAi? MiuAi is our hosted cloud AI for Miu, currently in beta . Instead of downloading models and running them on your own hardware, Miu's text conversations are powered by our servers - no API key and no powerful GPU required. (Voice features are separate and still run on your own PC.) MiuAi is included with Tier 4 and higher subscriptions at no extra cost. Just enable it in the launcher under Settings → MiuAi and start playing. Who can use MiuAi? MiuAi is available exclusively to members with an active Tier 4 or higher subscription (Tier 4, Gold, or Diamond) with their Patreon account linked to their Miu's World account. Not on Tier 4 yet? You can upgrade your pledge at any time - or use the free local AI chat (Ollama) or the Bring Your Own Key option instead. Does MiuAi have usage limits? MiuAi comes with a generous fair-use allowance that resets every month - most players will never notice it. If you do reach it, MiuAi is temporarily paused and becomes available again at the start of the next month. While paused, you can keep chatting with Miu through the local AI (Ollama) or your own provider key - your game is never locked out of AI features. Can I use my own AI provider instead? (Bring Your Own Key) Yes! With BYOK (Bring Your Own Key) you can connect your own account from a supported AI provider and have Miu's conversations run through it. Your API key is stored securely in your operating system's keychain - never in plain text and never sent to our servers. Since you're using your own account, you stay in full control of the provider, model, and spending. Can the voice features run in the cloud too? Yes. If your PC isn't powerful enough for local voice AI, Cloud Voice lets you run both sides of it remotely: • Voice synthesis (TTS) - Miu's voice is generated by a hosted voice engine instead of your GPU. • Speech recognition (STT) - your microphone input is transcribed by a cloud provider using your own key. You can mix and match - for example local voice output with cloud speech recognition, or the other way around. What is the Text-to-Speech (TTS) feature? TTS uses IndexTTS-2 , a state-of-the-art voice synthesis engine, to generate natural, expressive speech. It supports custom voice prompts - including built-in MiuTraining voices - so you can choose how characters sound. NVIDIA GPU acceleration (CUDA) is supported for fast generation, but the feature also works on CPU. What is the Speech-to-Text (Whisper) feature? The Speech-to-Text feature uses faster-whisper - an optimised build of OpenAI's Whisper model - to transcribe your voice to text in real time. It requires Python 3.12 (the launcher can install this automatically) and supports NVIDIA GPU acceleration for best performance, with a CPU fallback. What is the local AI chat / Ollama feature? Ollama is an open-source runtime that lets you download and run large language models (LLMs) entirely on your own machine - no internet connection required once a model is downloaded. The launcher lets you browse available models, pull them to your device, see how much disk space they use, and remove them - all without leaving the app. Do I need an NVIDIA GPU for the AI features? No - all local AI features (TTS, Whisper, and Ollama) fall back to CPU automatically if no compatible GPU is detected. However, an NVIDIA GPU with CUDA support is strongly recommended for significantly faster performance, especially for voice synthesis and speech recognition. Want to skip local AI chat entirely? MiuAi (Tier 4+) and BYOK run Miu's text conversations in the cloud instead of on your PC. Note that this covers text only - the voice features (TTS and Whisper) still need to be installed and run on your own PC , unless you set up Cloud Voice with your own provider key. Does the AI send my data to external servers? That depends on which features you enable - and it's always your choice: • Local features (TTS, Whisper, Ollama) - run entirely on your own machine. Voice recordings, text, and conversations never leave your device; the only network activity is the one-time download of model files. • Cloud features (MiuAi, BYOK, Cloud Voice) - these are optional and off by default. When enabled, only the data needed to generate a response is sent - and only for the feature you enabled. MiuAi and BYOK send just Miu's text conversations (to our servers for MiuAi, or directly to your configured provider for BYOK) - voice still runs locally on your PC. Audio only leaves your device if you explicitly set up Cloud Voice with your own provider key.