September 08, 2019



Easily available text-to-voice technology means that any book can be turned into an audio version within minutes. But the robot voice is often a deal-breaker as most listeners prefer nuanced narrations by an actual human being. These recordings, however, can take months to produce. The company behind Chinese search engine Sogou is now coming out with technology that creates lifelike ‘vocal avatars’ from a one-minute sample recording of a human voice, making it sound as though the author is reading the book to you personally. The addition of video clips of the author will make it seem as if you are actually attending a favourite author’s reading.

In 2018, Sogou created two Artificial Intelligence (AI) television newsreaders that are presently being used by Xinhua, China’s official and largest news agency. Sogou is currently in the process of developing AI versions of two popular Chinese authors, and the company claims that if adequate video and voice clips are available, even deceased writers could be given a new lease on ‘life’.

Published in Dawn, Books & Authors, September 8th, 2019