The technology is being developed by the Chinese technology firm Sogou (who are best associated with a search engine). Recently the company has been using artificial intelligence and imaging technology to construct ‘lifelike’ news anchors (these appear on China’s state-run news agency Xinhua). The next stage, the BBC reports, is to harness this technology to promote spoken word recordings.
Sogou has created avatars of two Chinese authors (Yue Guan and Bu Xin Tian Shang Diao Xian Bing), and it is developing technology that will enable the virtual representations of the authors to narrate audiobooks. Trial versions have been produced which combine artificial intelligence, text-to-speech technology and video clips (some made from recording at the China Online Literature+ conference) to fabricate the avatars, according to Engadget.
The new product has been worked on by Sogou in partnership with Zhangyue Technology, the operator of the iReader e-book device and novel-purchasing service.
The key selling point is that the audiobooks will appear to be narrated by authors or well know personalities. It is possible to produce artificial narration from a machine reading text, but the market potential rises if the public are familiar with the person who is doing the talking.
As a next phase, Sogou is considering combining text-to-speech technology with the likeness of different celebrities to create the illusion that audiobooks are being read by real people. Sales success is likely to rest on how lifelike the recording are and whether the public mind how the ‘voices’ have been created. There are also likely to be some copyright issues (and discussions about whether a person has given consent to be digitally converted), plus a discussion around whether recreations of deceased people should be used.
