'Red Dwarf' Talking Head Could Be A Future PA
A virtual 'talking head' called Zoe that expresses a range of human emotions could become the friendly face of future computers.
The disembodied head generates voice and facial expression from typed text. But it could be used as a digital personal assistant similar to Holly, the computer on the sci-fi comedy Red Dwarf.
The head is so lifelike that it is being used to teach autistic children to recognise emotion and deaf children to lip-read.
Professor Roberto Cipolla, an engineer at Cambridge University, said: "This technology could be the start of a whole new generation of interfaces which make interacting with a computer much more like talking to another human being."
The face is that of actress Zoe Lister from Hollyoaks. Researchers recorded her speech and facial expressions over several days and then super-imposed her face over a computer generated template.
In future it will be possible for users to upload their own faces and voices to create a personal digital assistant on their smartphone.
They could instruct the system to send a 'face message' that they are running late, with a frustrated expression.
Bjorn Stenger, another of the developers, told Sky News: "You can't convey emotion in a normal text message. But with our system people were able to read emotion much better."
Zoe's voice has six basic settings - happy, sad, tender, angry, afraid and neutral. Users can adjust the settings to different levels, as well as changing the pitch, speed and depth of the voice itself.
Tests show that 77% of volunteers were able to correctly recognise the emotion when they could see and hear the avatar - compared to 73% who identified the right emotion conveyed in a video of the real actress speaking.
The engineers are now trying to make the generated speech smoother to make the avatar even more realistic.
what do you think?
Please can I have one that looks like Carol Kirkwood?
Shame that the talking head 'ananova' that orange purchased as a news site and then abandoned several years ago had better speech synthesis than this..
Allan Booth - Staff