Scientists from National Institute of Advanced Industrial Science and Technology (AIST) have developed a novel speech prediction software that can complete half-formed words or sentences.
The new software looks fragments of words and other signs of hesitation such as filler sounds that Japanese speakers use when searching for their next phrase, just as English people use 'um' and 'er'.
"Although the concept of completion is widely used in text-based interfaces, there have been no reports of completion being effectively applied to speech," New Scientist quoted the researchers as saying.
Even if a person asks for a song by 'someone, er, Jackson' it will offer up Michael, Janet, and even Joe.
Though many people are sceptical of speech recognition software because poor past attempts, it is believed that the new programs is actually good and underpins many voice-activated gadgets.