Amazon’s Alexa personal assistant is able to read the news, but she does so using the same voice users hear when issuing commands and getting search results. That’ll soon change, the company has revealed — it is creating a “newscaster” voice for its assistant that’ll give Alexa the same style and flow heard on an evening news program.
Everyone is familiar with the “newscaster” voice style; it has a unique flow and emphasis atypical in ordinary conversations. The newscaster voice makes it immediately clear that the user is hearing a news report while helping emphasize key points, break up the information, and state everything clearly.
Amazon’s Alexa newscaster voice is no exception. Amazon has conducted surveys that found users prefer the newscaster voice for getting news rather than the ordinary speaking voice, according to The Verge. The newscaster voice is made possible using neural text-to-speech technology called NTTS.
An example of the new voice is available via the audio player above. This is compared to Amazon’s current use of concatenative technology, which is behind the assistant’s present speaking voice. Amazon told The Verge that it plans to switch Alexa to NTTS voice synthesis alongside the new newscaster voice in the near future.
How did Amazon manage to nail the news-speak voice to such a high degree? By using actual news segments combined with machine learning to determine the patterns in how newscasters read text. Though the end result is still a bit robotic compared to an actual human speaking, Alexa’s newscaster impression is still impressive.