Voxta

Your speech-driven AI companion backend

Speak to AI characters

Talk with a virtual friend, have fun with a character, and interact with them in virtual reality. Unrestricted.

Bring your own AI

Supports multiple Large Language Model programs, such as:

  • OpenAI GPT-3.5 and GPT-4
  • Oobabooga's Text Generation Web UI
  • KoboldAI
  • HuggingFace Text Generation Inference
  • OpenRouter
  • NovelAI

Speech

Choose your AI voice. You can go with what works best for you and your budget.

  • Azure Speech Service
  • Windows Speech
  • NovelAI
  • ElevenLabs
  • Vosk

Virtual avatar

See, touch, and interact with a virtual 3D model.

Check out our Patreon page for more detail!

Multicultural

There are options for most languages, including speech transcription, speech synthesis, large language models, and character definitions.

Features

  • Interrupt the AI while they speak
  • Tavern character cards support
  • Customizable prompts (Scriban templates)
  • AI text generation adapted for speech
  • Wake word support (Azure)
  • Peek inside the AI using "Analysis mode"
  • Keyword and summarization memory
  • Action inference: Avatars can act

Open

This project exists thanks to passionate individual investing in open source projects.

We will begin rolling out Voxta through Patreon. Long term, the Voxta core system will be open source, and community-driven.

Reach out if you're interested in contributing!

The Team

We are a small team of contributors with a common desire to have a fast and easy to use conversation platform for AI. We have no external financing.

Support us on Patreon