Intelligent Interaction Scenarios

MetaStudio's virtual avatar intelligent interaction can be integrated using one of the following methods:

Method 1: MetaStudio interconnects with third-party intelligent brains and LLMs. Integrators or customers develop intelligent interaction applications using the SDK.
Method 2: The MetaStudio APIs are provided by the real-time inference service (virtual avatars acting as instructed by the input text), without AI assistants. Integrators or customers integrate LLMs and automatic speech recognition (ASR), and develop intelligent interaction applications using the SDK.
- Method 1 requires pre-integrating the ASR service (AI Ask). Customers need to purchase the service and authorize MetaStudio to use it.
- Method 2 does not provide ASR, which should be prepared by customers.

Method 1

MetaStudio interconnects with third-party intelligent brains and LLMs. Integrators or customers develop intelligent interaction applications using the SDK. For details, see Figure 1.

To interconnect MetaStudio with third-party intelligent brains and LLMs, you need to use the LLM callback API of MetaStudio for custom development. For details, see Creating an Interactive Dialog Project (with Third-party Intelligent Brains).

Third-party intelligent brains and LLMs include:

Dify + MaaS + third-party LLMs deployed on Huawei Ascend, such as DeepSeek
Intelligent brains and LLMs deployed by customers

Scenarios: ISVs have the LLM capability and need to overlay virtual avatars.

Figure 1 Method 1

Method 2

The MetaStudio APIs are provided by the real-time inference service (virtual avatars acting as instructed by the input text), without AI assistants. Integrators or customers integrate LLMs and ASR, and develop intelligent interaction applications using the SDK. For details, see Figure 2 and Creating an Interactive Dialog Project (Custom Development).

Scenarios: ISVs need to control virtual avatars' speech.

Figure 2 Method 2