News

Robot talking heads "a nightmare for DRM" on phones

Close to announcing its first contracts with mobile phone networks, Anthropics has devised a wonderful system for generating animated "talking head" images from any still picture. But who owns the copyright on these "clips"?

Digital rights management with these animated faces could be sensitive, predicts Michael Burling, senior vice president of sales and marketing at Anthropics.

His company is about to sign deals with European phone operators which will initially allow them to generate animated messages. "You take a still photograph, and then with our Face Director software, you can animate it; make it smile, frown, move forward or back - and synchronise its lips with speech."

Which means you can not only have David Beckham talking about football, but you could have Barbie talking about grooming, or a cute hamster talking about The Two Towers. You can see a FacePlayer demonstration of a talking head on Anthropics's own Web site. And that face is NOT a recording of the young woman concerned; it's an animation of a still photo.

And the same technology works just as well for a hamster. Our photograph shows a talking hamster, with the same cute note-wrinkles, snapped on a Sony Ericsson P800 smartphone.

Burling anticipates that these animated clips will be sent as MMS packets to subscribers. "Initially, anything that comes under "broadcast" could do; it could be news, or weather, or sports updates, or horoscopes. We're talking to several content aggregators who like the idea. Or someone like Vodafone could have an animated David Beckham welcoming each customer when they first turn their phone on, without having to get Beckham into the studio to record it."

The face can't yet be one taken on a camera-phone, but that will come soon, says Burling. "We have Face Creator; it's a tool for use only on a PC right now, but it lets you scan any picture you like, and add that face to the face library. Then you can use it for pre-recorded messages. For example, we can do a fully-animated Barbie doll face."

There are also demos of cartoon faces, and film star faces.

There's another tool for text messages. Face-Server text is an extension of text messages, and allows you to nominate a face which will speak incoming texts, using a synthetic voice. The face can respond to normal text abbreviations, pronouncing "l8r" correctly, for example. And it can also respond to emoticons, smiling for :-) and looking sad for :-( icons.

There will be pre-designed messages - jokes, happy birthday notes, and "you are so dumped!" brush-offs.

The problem which Burling foresees, once this is accepted, is of digital rights management. "The issues around that are fairly delicate! - individuals are quite guarded about what you might be able to do; and so we wouldn't want people to have too much freedom to tamper with them," he said. However, he also understands that there's no way of preventing this technology from allowing individual users to create their own MMS clips, scanning celebrity faces in, and recording their own voice track. And some of those will be cleverly mimicked!

Anthropics software responds to phonemes; it takes a small player (50K of code) loaded onto the phone. However, the business of creating the animation isn't done in the phone; it's done at the operator's server system, and then transmitted intact to the phone as an MMS. But it doesn't have to be, and future technology will change that.

"One day, we hope to be able to create a message by voice - that is, just speak a message and build animations around that. Our software we created recognises all facial movements. It trains the software to respond to certain phonemes, it moves the face accordingly. When converts from text, it recognises them from the speech, not from the text. So it's the same thing from speech. It can't be done in today's phones, but the next generation may be able to."