Will the real news anchor please stand up? Wait, actually, there’s no need. On Tuesday, Chinese search engine company Sogou and official news agency Xinhua unveiled an “AI composite anchor” that can stand and talk at the same time, multiple Chinese outlets reported.
Previously, the companies worked together to create two sit-down AI television anchors released last November. Based on two real hosts from state media channel China Central Television, they were advertised as the “world’s first” anchors with synthesized voices and appearances. Since that announcement, the AI anchors–one speaking Chinese and the other English–have hosted over 3,400 news reports, clocking in more than 10,000 minutes of screen time, Tencent News reported (in Chinese).
However, clips of those two anchors only showed the upper halves of their “bodies,” omitting hand movements and other gestures. Sogou and Xinhua’s new anchor represents an improvement, adding in more body language for more natural expression. In addition, according to reports, it’ll feature more realistic facial expressions, lip movement, and speech synthesis using Sogou technology.
The new anchor will be featured in reports on the Two Sessions, an abbreviation for the critical annual meeting of the Chinese People’s Political Consultative Congress and the National People’s Congress. This year’s Two Sessions will begin in early March. The AI anchor will also appear on Sogou’s search site and text input software. Sogou could not immediately be reached for comment.
Xinhua and Sogou also announced they would cooperate in the long-term to create more virtual news anchors in the future, to be featured across various Xinhua media platforms.