- What's New
- Service Overview
- Service Brochure
- Getting Started
-
User Guide
- Prerequisites
- Permissions Management
- Image Modeling
- Voice Modeling
- Video Production
- Livestreaming
- Asset Management
- Appendix
-
API Reference
- Before You Start
- API Overview
- Calling APIs
- Asset Management
- Virtual Avatar Video Production
-
Virtual Avatar Livestreaming
-
Intelligent Livestreaming Room Management
- Creating an Intelligent Livestreaming Room
- Querying Intelligent Livestreaming Rooms
- Querying Intelligent Livestreaming Script Details
- Updating Intelligent Livestreaming Room Information
- Deleting an Intelligent Livestreaming Room
- Creating an Interaction Rule Library for Live Rooms
- Querying Interaction Rule Libraries for Live Rooms
- Updating an Interaction Rule Library for Live Rooms
- Deleting an Interaction Rule Library for Live Rooms
-
Livestreaming Task Management
- Starting a Virtual Human Intelligent Livestreaming Task
- Querying Livestreaming Tasks of a Live Room
- Querying Virtual Human Intelligent Livestreaming Task Details
- Ending a Virtual Human Intelligent Livestreaming Task
- Controlling Virtual Human Intelligent Livestreaming
- Querying All Virtual Human Livestreaming Tasks of a Tenant
- Reporting Livestreaming Events
- Live Product Management
-
Intelligent Livestreaming Room Management
-
Image Modeling Management
- Creating a Virtual Avatar Model Training Task
- Querying Virtual Avatar Model Training Tasks
- Querying Details About a Virtual Avatar Model Training Task
- Deleting a Virtual Avatar Model Training Task
- Updating a Virtual Avatar Model Training Task
- Executing a Virtual Avatar Model Training Task as a Tenant
-
Voice Modeling Task Management
- Creating a Voice Training Task (Basic Edition)
- Creating a Voice Training Task (Advanced Edition)
- Creating a Voice Training Task (Premium Edition)
- Querying Voice Training Tasks
- Submitting a Voice Training Task
- Querying Voice Training Task Details
- Deleting a Voice Training Task
- Querying Task Operation Logs
- Obtaining the URL for Uploading a Voice File
- Obtaining the Review Result of a Voice Training Task
- Confirming the Online Recording Result
- Obtaining the Confirmed Online Recording Result
- TTS Management
- Appendix
- Change History
- ssdk
- FAQs
- Videos
- General Reference
Show all
Copied.
Script 1: Introduction to Huawei Cloud MetaStudio
Based on the Pangu virtual human model, Huawei Cloud MetaStudio, a digital content production pipeline, provides a one-stop and full-process virtual avatar livestreaming service. With efficient training and inference, the service reshapes live commerce.
Virtual avatar model generation and control: A virtual avatar model can be generated from just one photo or a three- to five-minute video, which is three times more efficient than the industry average. After the virtual avatar model is generated, the model control and inference service can create lifelike voices, facial expressions, lip movements, and body languages, with a lip sync accuracy higher than 95%.
Intelligent generation of livestreaming script: Based on the pre-training of 100,000 pieces of high-quality livestreaming script, professional script is automatically generated to control virtual avatars so that they can accurately and fluently introduce products.
Intelligent interactive Q&A: Virtual streamers powered by the Pangu foundation model can interact with viewers during livestreams, which helps convert more viewers into buyers.
The virtual avatar livestreaming service built on Huawei Cloud MetaStudio has been widely used across industries. AI helps virtual avatars become professional streamers. Here are some industry cases:
- Huawei Cloud works with the Accessibility Research Association to build virtual avatars for hearing-impaired people to make their voice heard and enjoy better life powered by advanced technologies. The virtual avatar of Shi Xiaohui, a hearing-impaired painter, tells the story behind her works to make her voice heard by more people.
- 491 million people in rural areas in China engage in the production of agricultural products, and there are more than 100,000 rural industry segments with local characteristics. Livestreaming makes these local products known by more people and is unleashing huge market potential for rural areas. However, livestreaming is still too new for many rural citizens, who are camera-shy and need to spend a long time practicing what they will say during livestreams. In addition, it is time- and labor-consuming to set up livestreaming rooms near their houses. This is exactly where virtual avatars come into play. Unlike human streamers, virtual avatars on the cloud can stream for a whole day anytime and anywhere, while farmers can mind their own work. Professional messages are automatically generated and fluently presented by virtual streamers to sell agricultural products to more areas worldwide.
- Wondershare joins hands with Huawei Cloud to develop Wondershare Virbo, an AI video app for international livestreaming by virtual avatars. Based on Huawei Cloud MetaStudio, the app supports end-to-end AIGC creation from scripting, virtual avatar presentation, to automatic video generation. This solution solves many pain points of marketing videos outside China, such as the unavailability of professional promoters, difficult multilingual adaptation, long production period and the consequent high video shooting and livestreaming costs.
- To stay relevant in the snack industry, Xu Fuji embraces new online marketing techniques. Xu Fuji leverages Huawei Cloud MetaStudio to build a virtual assistant, which can stream 24 hours a day anytime and anywhere for over 3,500 suppliers and 300,000 store owners nationwide. Virtual streamers focus on consumers' preference and enable immersive scenario-specific marketing to bring brands closer to their consumers.
Finally, I'd like to introduce Sara, a virtual co-host growing on Huawei Cloud MetaStudio. Virtual humans are qualified for many roles, such as host, idol, live commerce streamer, lecturer, and financial customer service personnel. Traditionally, it takes several months to produce such a 3D realistic virtual human. The multi-modal intelligent modeling and cloud rendering functions of Huawei Cloud MetaStudio shorten the production period to just one week.
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
For any further questions, feel free to contact us through the chatbot.
Chatbot