The 5th World Internet Conference kicked off in Wuzhen, Zhejiang, China today, with state-run Xinhua News Agency and leading search service provider Sogou jointly unveiling the world’s first “AI Synthetic Anchor,” a highly realistic AI-generated newscaster. The presentation showcased Sogou’s advanced voice and video synthesis technology and was a highlight of the conference’s opening day.
The AI anchor delivers the news with the calm fluently of a professional. His lips and mouth appear perfectly synchronized with his speech, while eyebrow twitches, hand gestures and physics treatments add to the realism.
There are actually a couple of AI Synthetic Anchors: One is a clone of the Xinhua News Agency’s Hao Qiu and delivers the news in Mandarin; the other is modeled on Xinhua newscaster Zhang Zhao and performs in English. In their 30-second demonstration clips the AI anchors assure listeners: “I will work tirelessly to keep you informed as texts will be typed into my system uninterrupted.”
Users need only input a text into the AI Synthetic Anchor system and the AI composite anchor will do the rest. Moreover only a small amount of audio and video input data is required to customize the agent.
Sogou began researching voice technology in 2012 and has since developed into the largest speech recognition engine in China. The company receives 500 million voice search requests daily and years of data accumulation and research have driven its multi-language and multi-tone speech synthesis and enabled its complete personalized speech synthesis and emotional migration tech. The team that produced the AI Synthetic Anchor has won challenges in some of the most prestigious competitions in the field of computer vision, including IEEE CVPR WAD, Pascal VOC, Mega Face, and the Blizzard Challenge — the most coveted in speech synthesis.
The AI Anchors not only drew crowds at the World Internet Conference for their technological innovation and breakthrough in AI synthesis, they also pioneered the synthesis of real-time audio and video and AI live-action in the news industry. Both the Hao Qiu and Zhang Zhao clones are scheduled to begin work shortly on a variety of Xinhua’s platforms. With their 24/7 work shifts, they are certainly cost effective.
(If you were wondering which of the above gifs is the AI Anchor, it’s second one.)
Jouarnlist: Fangyu Cai | Editor: Michael Sarazen
Now I am going to do my breakfast, once having my breakfast coming yet again to read more news.