China has a virtual nieuwsanker. An AI system that is eerily realistic the news read. A technical feat. It Is also a world first? Not at all. We look back on virtual presenters, MTV icon Max Headroom to the green-haired Ananova.
The Chinese news agency Xinhua announced this week a new recruit: a virtual nieuwsanker. Or actually two: there is an English and a version in Mandarin. The English-Speaking …
The Chinese news agency Xinhua announced this week a new recruit: a virtual nieuwsanker. Or actually two: there is an English and a version in Mandarin. The English-Speaking has the face of a human reader to Zhang Zhao, but has apparently not been named. The images are created on the basis of video recordings of Zhang Zhao, combined with a computer voice. In a video of 30 seconds he introduces himself: “Hello everyone, I am an English-AI-anchor. This is my very first day in the Xinhua-news agency.’ He will continually check the latest messages ‘read’ that human journalists typing. In the meantime, he has already some bulletins of two minutes ‘presented’.
And, how does he fix them? That is a matter of taste. The video is sharp and the ‘man’ creates natural-looking head movements. But yet you can see already after a few seconds that something is wrong. With the movement of his mouth is something strange going on: his lips speak the words, but his chin moves just barely. It strongly reminds of the ‘deepfake’ videos a few months ago, the round did, and in which the face of well-known persons on a other actor ‘attached’ to that person, other things to do or say. That resemblance is no accident: Xinhua makes use of the same technology, so-called Generative Adversarial Networks (GAN). That is the latest wrinkle on Artificial Intelligence.
But here is that technology is linked to a text to speech system, a synthetic voice so that a (man-written) news article reads. The English computer voice is unfortunately much less convincing than the video – a lot of robotachtiger than recent versions of Siri or Google Assistant, for example.
The two unnamed Chinese anchors in the footsteps of some illustrious predecessors.
Max Headroom (1985): plastic pack and sarcasm
There can only be one first, and that was Max Headroom. On april 4, 1985, the virtual presenter making his debut in the British tv program Max Headroom: 20 minutes into the future. Then he got a job as a veejay in the music program , The Max Headroom Show. Already, many people mainly remember from the video for Paranoimia by The Art of Noise.
Okay, Max Headroom was not really: he was played by actor Matt Frewer, hidden under all kinds of rubber fittings and a plastic pack. Frewer inspired on Ted Baxter, the smug news reader in the tv series , The Mary Tyler Moore show. The ‘choppiness’ of the computer that belonged to the character, but the ‘computerlijnen ” in the background were drawn by hand. In the for other uses, see Pixels from 2015 turned Frewer back as Max Headroom.
Ananova (2000): the green-haired vamp
The dotcomzeepbel of the late 90s was on the verge to explode, in april 2000, when Ananova her debut. A moving 3D cartoonfiguurtje that, just as its Chinese imitators eighteen years later – written news could read. What they on its own website, Ananova.com non-stop did. The appearance of Ananova was partly inspired by Victoria Beckham (the Spice Girls were still in the charts!), with a little bit of Kylie Minogue and Carol Vorderman (British tv-face). Also she was actually coined by a news agency: Press Association.
Ananova got a bio nevertheless: it was proposed as a 28-year-old uitgaanstype, 1,73 great and a fan of Oasis and The Simpsons. At that time, it was a real feat of computer graphics and speech technology, there was a team of 80 people. A few months after her debut she became for a mind-boggling 95 million pounds were sold to the British mobile phone operator Orange, who then bitter little started. After a career of four years was the figure in 2004 retired. Her digital voice resonates almost twenty years later is still striking, of course.
Erica the robot (2015):
The Japanese robotmaker and weirdo Horoshi Ishiguro leurt for nearly four years with the robot, Erica. She was designed as an artificial receptionist, but at the beginning of 2018, it was announced that, from april, the nieuwsanker would be of a Japanese tv program on Nippon TV. A message that worldwide was picked up, but that was premature and found to have been. Ishiguro is also the first of his exaggeration is not dead – he also claims to like Erica a ‘soul’.
Erica looks amazingly lifelike, with an extremely realistic skin, eyes and hair. But her movements are stilted, and the facial expressions she has hardly. They can be a person looking straight into the eyes and simple to answer questions. But most of her public ‘appearances’ seem to be largely or completely pre-programmed.