Traditional internet search methods of entering search terms into a search engine have now been supplemented by voice search, in which the searcher interacts with search engines with speech.
Machine learning plays an important role in helping devices understand the requests made of them, enabling the enhancement of search accuracy. Advances in artificial intelligence have made it possible for devices to recognise natural language. The basic process of voice search begins with the translation of human speech into digitised text. This is then analysed to identify questions or commands. The keywords in those questions and commands are then measured against available data sources and search engines to find potential matches. The information thus received is translated into a format that can be relayed back to the user.
Voice search is not infallible so occasionally inappropriate results will be returned but its accuracy is measured at about 80% and it is being improved all the time. It is predicted that voice search will become the dominant means of conducting web enquiries within a few years. The increasing use of smartphones for web interaction is partly responsible, with 56% of all voice searches now being conducted on mobile devices.