Starting today, if you ask Alexa on your Amazon Echo speaker any question, you may notice its voice sounds more natural or different than before. Don’t be surprised; Amazon has just rolled out a fresh new speaking style to Amazon Alexa users in the US.
The Alexa new style of speaking is modeled on the way newscasters deliver the news. This was first announced in November 2018, when developers at Amazon utilized developments in Neural TTS (aka Text To Speech) technology to create the updated voice, which allows Alexa to tailor its speaking style to different requests and commands.
Now, Alexa’s newscaster-style voice has clearer intonation and now sounds less robotic than it used to be. Before Amazon rolled the new-styled sound out, it has conducted several experiments to know how important the natural voice is to the consumers.
Amazon found that people preferred Alexa’s newscaster voice when hearing the news. However, despite Alexa’s newscaster voice, consumers still prefer voices that sound like their own.
Alexa is a fast learner
Amazon also noted that researchers were able to train Alexa to use the newscaster voice after a few hours of feeding it phonemes (the most basic unit of language that distinguishes one word from another).
According to the statement released in November 2018, Amazon said, “this advance paves way for Alexa and other services to adopt different styles of speaking contexts to improve customer experiences.”
Alexa can already read your child a story and even personalize it to include the name of your child. But how nice would it be Alexa could read in dramatic voices and even impersonate different characters?
Amazon Alexa can lower its voice when you don’t want it to disturb the people around you late at night. If you enable Whisper mode on it, it will respond to your hushed tones in a whisper. Unfortunately, Whisper Mode is only currently available in the US just like the newscaster’s voice and we are yet to hear anything concerning when the new features will be rolled out for the use of other regions.