Models

This module explains how to make the full OpenTalking model chain runnable, not only the talking-head backend. A usable digital-human session depends on five parts:

flowchart LR
    STT[Speech recognition<br/>optional voice input]
    LLM[LLM<br/>decides what to say]
    TTS[TTS<br/>text to audio]
    Avatar[Avatar assets<br/>image / frames / template]
    Head[Talking-head backend<br/>audio to video]
    WebRTC[WebRTC<br/>browser delivery]

    STT --> LLM --> TTS --> Head --> WebRTC
    Avatar --> Head

Recommended defaults

Layer	Default for first run	When to change it
LLM	DashScope OpenAI-compatible endpoint	Use OpenAI, vLLM, Ollama, or DeepSeek when those are already standard in your environment.
STT	DashScope Paraformer realtime	Keep it unless you need a different realtime ASR provider.
TTS	Edge TTS	Use DashScope, CosyVoice, or ElevenLabs for production voices and voice cloning.
Avatar assets	Built-in examples	Prepare model-specific assets before selecting Wav2Lip, QuickTalk, FlashHead, or FlashTalk.
Talking-head backend	mock first, then Wav2Lip compatibility path	Use QuickTalk local, FlashHead direct WS, or OmniRT for heavier models.

Setup order

1. Run Quickstart with mock.
2. Check the Support Matrix to choose the right path.
3. Configure LLM and STT.
4. Choose and verify TTS.
5. Prepare Avatar assets.
6. Start a talking-head model.
7. Verify /models, create a session, and test through the browser.

Keep model execution decoupled from OpenTalking itself: lightweight models should use local or direct_ws where possible, while OmniRT remains the recommended backend for heavyweight, multi-card, remote, or NPU deployments.