Back in November, word surfaced that Amazon was working on a proper newscaster voice for Alexa, which it would be used when reading the news. Unlike the assistant’s regular voice, the newscaster voice features the same intonation and emphasis used by professionals when delivering the news. As of today, Alexa users in the US can now access that newscaster voice.
Amazon first detailed its plans in November on its developers blog, where it explained that its use of a neural text-to-speech system (NTSS) provided the flexibility necessary to adjust Alexa’s speaking style. In this case, Amazon used a few hours of newscaster-style voice recordings to create the newscaster voice, something that wasn’t previously possible with it concatenative system.
Though Alexa being able to read the news is nothing new, the assistant’s regular speaking voice isn’t really suited to the task — its spoken version of written news content sounds off and can be hard to follow at times. The newscaster voice fixes this by putting emphasis on the right spots in a story, helping break up the flow while making it easier for users to keep track of what they’re hearing.
According to its most recent blog post, Amazon’s Neural TTS technology enables Alexa to adjust its voice based on the type of request it receives. The first example of this is the newly launched newscaster voice, which is automatically used when the users asks for a news update.
The same technology could be used in the future for other voice adjustments based on the command Alexa receives. In addition to reading the news, the newscaster voice will also be used when users ask Alexa for information from Wikipedia. The new voice is available to Alexa users in the US starting today.