Concepts and API

This section is for developers and integrators. It explains OpenTalking concepts, API boundaries, and extension points. If the goal is to run the project first, start with Tutorials. If the goal is to understand business scenarios, start with Use Cases. If the goal is model serving, start with Model Deployment.

Understand Concepts

• Architecture — components, deployment topologies, session lifecycle, event bus, and backend boundaries.
• Render Pipeline — how LLM, TTS, synthesis backends, and WebRTC fit together.
• Avatar Format — avatar bundles, manifest fields, and model matching rules.

Query the API

• API Interfaces — Base URL, error conventions, endpoint table, and common request sequence.
• Health and Models — /health, /models, and queue status.
• Sessions — sessions, chat, WebRTC, recording, and offline bundles.
• Events and Streaming — SSE and audio WebSocket protocols.

Extend Implementation

• Model Adapter — integration contract for new talking-head backends.
• Developing — local execution, tests, debugging, and repository layout.

Debug and Contribute

Before contributing, run python -m mkdocs build --strict --clean and update the API docs, model deployment docs, or Benchmark records according to the change. See Community for contribution paths.