Tutorials¶
Tutorials are organized by task path. Use them to run OpenTalking from scratch, connect a real model, prepare custom avatars, or enter the development workflow.
First Run¶
- Installation — check Python, Node.js, ffmpeg, DashScope keys, and hardware requirements.
- Quickstart — run the
mocksynthesis path through browser, API, LLM, STT, TTS, and WebRTC. - Configuration — understand
.env, YAML configuration, providers, and backend precedence. - Mock E2E case — validate the local environment with the smallest full path.
Connect a Real Model¶
Verify orchestration with mock first, then connect a real talking-head backend:
| Scenario | Tutorial |
|---|---|
| First real lip-sync model | Wav2Lip integration |
| High-quality FlashTalk/OmniRT path | FlashTalk integration |
| Model, weight, and deployment selection | Model deployment |
These pages focus on low-level integration steps. If you want to start from business scenarios, see Use Cases.
Custom Avatar¶
Start with Custom Avatar to learn how images, videos, and
manifest.json form an avatar bundle that OpenTalking can discover. Field details are in
Avatar Format.
Development and Debugging¶
| Goal | Entry |
|---|---|
| Modify API, session, or Worker behavior | Developing |
| Understand components and data flow | Architecture |
| Add a new talking-head backend | Model Adapter |
| Investigate performance and first-frame latency | Benchmark |