Clone Talking enables real-time voice conversations with AI persona clones through a local web app that runs on PC and phone.

Clone Talking is a voice AI web app that runs on a PC server; users scan a QR code on their phone, hold a button to speak, and receive responses in a cloned voice in real time using Whisper for speech-to-text, OpenRouter for LLM routing, and VoiSpark for text-to-speech with voice cloning.

Clones are defined as Markdown files with YAML frontmatter in a clones/ directory, where each file contains metadata (id, name, description, voiceId) and a system prompt that instructs the AI on persona behavior; supported models include anthropic/claude-sonnet-4-5, anthropic/claude-opus-4, openai/gpt-4o, and others via OpenRouter.

Every conversation is automatically saved to logs/ as structured JSON with session ID, timestamps, and full message history; the setup wizard guides users through environment configuration, lists available clones, and displays a QR code for immediate access without app installation.