Updated on 2025-08-22 GMT+08:00

Features

Huawei Cloud MetaStudio is an AI-infused digital content production pipeline designed to streamline the creation of engaging digital content. The platform leverages advanced image and voice modeling techniques to generate lifelike virtual avatar models. These models can then come into play in various scenarios, including video production, livestreaming, and intelligent interaction.

For details, see Table 1.

Table 1 Features

Feature

Description

Image modeling

Shoots a video of a human to generate a virtual avatar.

The virtual avatar can apply to video production, livestreaming, and intelligent interaction.

Note:

  • You cannot download or export a created virtual avatar model to your local device.
  • Virtual avatar models are not general ones and are, therefore, incompatible with third-party services. Models generated on MetaStudio can be used only within MetaStudio.
  • Virtual avatars do not support clothes or face swap.
  • If a training video used for image modeling includes choreography, you can see on the card of the generated virtual avatar, meaning you can add actions to this virtual avatar.

Voice modeling

Uses a recorded human voice to generate a voice model.

The voice model can be used to dub virtual avatars in video production, livestreaming, and intelligent interaction.

Note:

  • You cannot download or export a created voice model to your local device.
  • Voice models are not general ones and are, therefore, incompatible with third-party services. Models generated on MetaStudio can be used only within MetaStudio.

Video production

Uses a preset or custom virtual avatar image and voice to generate audio/video content.

The generated videos are applicable in a wide range of scenarios, such as teaching and training.

Livestreaming

Uses a preset or custom virtual avatar image and voice for livestreaming.

You can livestream on a platform by:

pushing RTMP streams from Huawei Cloud MetaStudio to a third-party livestreaming platform. You need to obtain an ingest URL from the platform. If the ingest URL is not available, you can start livestreaming through window capture.

Intelligent interaction

Performs interactive Q&A between users and virtual avatars equipped with a third-party brain. Interactive virtual avatars are qualified for many roles, including shopping guide, culture and tourism guide, and customer service personnel.

CAUTION:

The answers are given by the integrated third-party large language model (LLM) or knowledge base.

Asset management

  • You can upload models, PowerPoint files, animations, materials, videos, scenes, images, and music from a local device.
  • Voice and model assets can be transferred to other tenants.

    This function is not available yet. To use it, submit a service ticket.

  • Assets can be deleted.