The meta universe is still on the way, and the virtual human has arrived. "Liu night Xi" the first voice of the first show attracted more than 14 million people to watch, tiktok stage "Luotian" advertising received soft, Hunan satellite TV "Xiao Yang" has become the first permanent resident virtual host, CCTV AI sign language anchor and Zhu Guang partner in the Winter Olympic Games rush to search. With the rapid iteration of hardware such as AR and VR and the improvement of computing power, the development of virtual human industry is entering the fast lane, and the application scenarios are constantly landing. However, at the same time, factors such as insufficient computing power, fragmented industrial chain and imperfect laws and regulations still restrict industrial upgrading. Virtual human who "crush" human beings "Welcome the Winter Olympics, welcome the winter Paralympic Games, and usher in the Shenzhou new year. Sidaokou, Wudaokou and Liudaokou lead to Zhangjiakou..." during the Winter Olympics, Zhu Guangquan, a member of CCTV, gave a tongue twister, which made the newly appointed "Ai sign language anchor" out of the circle. According to reports, this sweet looking AI sign language anchor can automatically translate words, audio and video through AI technologies such as speech recognition and machine translation, and interpret sign language into facial expressions in real time. By studying the standard of national general sign language dictionary and the data of millions of standard sign language parallel corpus, the accuracy of sign language movement can reach more than 95%. The moving picture comes from CCTV news More than 100 events are broadcast in sign language at the same time. What humans can't do is easy for virtual humans. In fact, in addition to sign language broadcasting, virtual people comprehensively "crush" many people in the workplace. In 2021, the excellent newcomer award of Vanke headquarters was awarded to a virtual person named "Cui Xiaopan". Yu Liang, chairman of the board of directors of Vanke Group, explained that "the write off rate of prepaid accounts receivable / overdue documents urged reached 91.44%". Last September, the virtual human "ayayi" unveiled Alibaba's work card. Her first little red book was read by 2.24 million on the day of release, and the makeup effect attracted many bloggers to imitate. Virtual human "ayayi" In the campus, the virtual person also got an exclusive student card. On June 15 last year, the computer department of Tsinghua University announced that the virtual student "Hua Zhibing" was officially enrolled. She wore white board shoes with ponytail and studied in the computer department laboratory of Tsinghua University. On July 13 last year, Xia Yubing, a graduate virtual human of the Central Academy of fine arts, held an art solo exhibition of "alternative worlds". 3.0 era of thousands of people and thousands of faces Although the definition of "virtual human" has not been unified, it is generally believed in the industry that virtual human (digital human) refers to a comprehensive product that exists in the non physical world, is created and used by computer means such as computer graphics, graphics rendering, motion capture, deep learning and speech synthesis, and has multiple human characteristics. It seems to take advantage of the east wind of the meta universe to grab the limelight, but virtual human is no longer a new thing. It was produced before the rise of the meta universe, and even dates back to the beginning of the birth of computer animation technology. As early as the 1980s, the virtual song Jilin Mingmei in the Japanese animation "fortress in time and space" has gained a large number of fans. The world's first virtual idol may be the "first sound future" born in 2007, and her "song of throwing onions" is also the youth memory of a generation. Virtual idol concert With the continuous maturity of AI technology, more and more virtual people appear on the screen, and finally ushered in an explosive singularity in the past two years - virtual people have moved from two-dimensional cartoon to super simulation three-dimensional rendering, from synthetic sound performance to action expression and emotional full-dimensional interactive coverage. Some head virtual anchors are close to the top in traffic and income. Li Shiyan, head of Baidu Intelligent Cloud AI Human-Computer Interaction Laboratory, said that virtual human has gone through the 1 stage represented by "paper player". The 2 era represented by Vtuber (virtual anchor) anchors has evolved to the 3 stage, that is, virtual human 3 with the characteristics of high precision and artificial intelligence drive has become the mainstream of the industry, and its modeling and content production are all involved in AI. The production is more efficient and intelligent. It can also face more application scenarios. At the same time, it can face the whole group of users. It can be carried by live video and other carriers, and has the ability of interaction between thousands of people and thousands of faces. Chen Rui, CEO of BiliBili, introduced last year that more than 32000 virtual anchors started broadcasting at station B in a year, with a year-on-year increase of 40%. Virtual anchors have become the fastest-growing category in the field of live broadcasting at station B. According to the data of qixinbao, from 2020 to 2022, the number of virtual human related enterprises showed a gradual and rapid growth trend. Among them, the number of virtual human related enterprises increased by 36080 in 2020 and soared to 66293 in 2021. According to media statistics, the maximum single financing amount of a single virtual human project last year exceeded hundreds of millions of yuan. There are still difficulties in large-scale implementation After nearly 40 years of silence, why does virtual human suddenly usher in an explosive growth period? Li Shiyan believes that it is inseparable from its core driving force, the "three-stage rocket". "First stage rocket" is the integration of user needs and technological upgrading. At present, China has more than 400 million pan secondary users and more than trillion enterprise intelligent market. The blessing of artificial intelligence enables virtual people to meet the needs of all parties; "Second stage rocket" is the support of policy and the influx of capital. The state has issued a large number of policies to support the development of digital related industries in the past four years; The "third stage rocket" is an iteration of the computing platform. Users can enter the metauniverse through AR and VR devices, especially in the social field. Although there are many difficulties in the implementation of virtualization, many people still face the prospect of commercialization. "Each node of the virtual human's industrial chain is relatively fragmented and cannot cooperate efficiently, resulting in high barriers in the production and optimization of virtual human. At present, most companies in the industry are only one or several links in the whole process of production and operation." Li Shiyan said that there is no efficient connection between the service scene and the performance scene. The performance digital people do not have the business capabilities required by customers. However, the service digital people lack human facilities and are difficult to communicate with users. More importantly, the cost of virtual people with high mobility and high-frequency needs is still very high. In this regard, analysts of Huachuang securities also pointed out some industry risks, such as the concept of virtual human is relatively advanced, the future evolution process is less than expected, the business model of virtual human is not clear, resulting in the decline of the overall scale of the industry. In addition, the relevant legal system of virtual human is not perfect, and there are policy or regulatory risks. Within 10 years, it will meet the 100 billion market Compared with the landing of the meta universe, the current development of virtual human is more "visible to the naked eye", and many existing occupations may be replaced. In the short term, virtual people are generally divided into service type and identity type. The former can provide services in specific scenes to replace the social roles of many service industries, such as enterprise employees, hosts, housekeepers, etc., while the latter is more inclined to be applied in the field of entertainment, such as virtual idols, virtual anchors, etc. According to the prediction of the virtual digital human in-depth industry report of qubit, a research report and Market Research Institute of CITIC Securities, the market scale of virtual digital human in China is expected to reach 270 billion yuan in 2030, of which the market scale of identity virtual human is expected to be 175 billion yuan, and the total scale of service virtual human will also exceed 95 billion yuan. In the "14th five year plan for the development of science and technology of radio, television and network audio-visual", the State Administration of radio, film and television proposed to "accelerate the technological upgrading of the production and broadcasting system, and promote the wide application of virtual anchor and animation sign language in the production of news broadcasting, weather forecast, variety science and education and other programs". It can be seen that virtual anchor will usher in a bright spring in the radio and television industry. Communication University of China's "China virtual digital human influence index report" divides the most widely used and popular virtual human in China into three categories: virtual idol, virtual employee and virtual anchor. With the expansion of application scenarios, the classification will be further expanded and refined in the future. Virtual human will become a new interface for human-computer interaction and carry the immersive experience of the digital world. Virtual employees will play a role in consumer goods, finance, real estate, property, education, culture and tourism and other service industries. In the long run, analysts at CITIC Securities concluded that there are four business models for virtual people in the future: image virtual people can take the star route to speak for brands; Content-based virtual people pay more attention to interaction and IP core, which can be realized in the form of concert, live broadcast, surrounding and so on; Functional virtual human is limited by technology and cost in the short term, and the commercial realization space is limited, but it will be widely used as the infrastructure of meta universe in the future; The virtual avatar has a large coverage, strong willingness of users to pay, and broad imagination space in the era of meta universe. (Xinhua News Agency)