Voxta What is Voxta
A brief tour of Voxta — what it does, who it's for, and how the pieces fit together.
Voxta is an AI orchestration platform for natural voice and text chat. It connects the things you already have — local LLMs, cloud APIs, voice services, your hardware — into a single runtime that powers real-time conversations with AI characters.
You author the character. You pick the brains (LLM), the voice (TTS), the ears (STT). Voxta handles the plumbing.
What Voxta is not
- Not a single chatbot. Voxta is an engine. The characters you build with it are yours.
- Not tied to one AI provider. Swap LLMs, voices, and transcription services freely. Local llama.cpp this week, Anthropic next week, Voxta Cloud after that — same characters, same chats.
- Not cloud-locked. Voxta runs on your machine. If you choose to use Voxta Cloud for AI, that's an option, not a requirement.
How it fits together
Voxta Server is the orchestration core. It runs as a local desktop app and exposes:
- A web UI at
http://127.0.0.1:5384/for chatting, character authoring, and configuration. - A WebSocket / SignalR API that other apps (Voxy, the VAM plugin, your own integrations) connect to.
- A plugin system that loads each AI service as a module — OpenAI, Anthropic, ElevenLabs, llama.cpp, Vosk, and dozens more.
The orchestrator pulls together a turn: STT transcribes your voice, the LLM module generates the reply, the TTS module speaks it back, augmentations (vision, memory, search) feed extra context in along the way. Each piece is swappable per character or scenario, so you can run a fully-local stack one day and a Voxta-Cloud-backed one the next without rebuilding anything.
What people use it for
- Voice-first AI companions — speak to a character, hear it speak back.
- Voxta-driven VAM scenes — see Voxta for Virt-A-Mate.
- Desktop avatar companions — see Voxy.
- Custom integrations — game mods, home automation, agentic assistants. Anything that wants natural conversation in the loop.
Local-first by default
Voxta is built local-first. By default your prompts and conversations never leave your machine. You can opt into cloud services (Voxta Cloud, OpenAI, ElevenLabs, etc.) per service — and even when you do, Voxta itself never sees the content.
See the Privacy Policy for the full data-handling story.