by Andrew Cochran

A 3D AI news anchor is the latest in a steady evolution of ‘vocational avatars’ in China. The AI figure is produced by Sogou, a Chinese search and internet company, which has pioneered AI news figures in association with Zinhua TV, the state-run news channel.

Like its predecessors, this AI presenter is modelled after a well-known newsperson.

It replicates the appearance and mannerisms of Zhao Wanwei, a Zinhua reporter. All of the AI news anchors work from input text. They are promoted for their ability to ‘work tirelessly.’

Hand-out video from Sogou

Information provided by the company is unclear on how the figure is executed, for example, as a 3D rendering or using robotics. A company release says its 3D anchor uses ‘cutting-edge algorithms and high-fidelity 3D technologies used in movie productions to create the 3D avatar.’

  • The company says it uses ‘multi-modal recognition and synthesis, facial recognition and animation, and transfer learning, (which) allow the avatar to intelligently imitate human voices, facial expressions, lip movements, and mannerisms with only text inputs.’
  • Elsewhere it says the figure can ‘walk around during a broadcast’ and ‘in the future, I will walk out of the studio.’ No first-hand accounts of the figure are evident in online reporting.

‘Compared with previous generations of Sogou’s AI News Anchors, the 3D AI News Anchor communicates visual and audio information more naturally and smoothly in real time and in a real-life setting’

Sogou press release