Amazon is introducing a brand new talking model for Alexa and including some talking types for different AI voices, the company said today. The longform talking model is offered for US builders who desire a studying voice that sounds extra pure when studying lengthy items of content material, like articles or podcasts.
Amazon says the longform model is “powered by a deep-learning text-to-speech mannequin,” and permits Alexa-voiced gadgets to talk with extra pure conversational pauses. It follows final 12 months’s launch of latest talking types for information and music content material and a November replace that enables Alexa to look “dissatisfied” or “excited.”
(I’ll be trustworthy: the one apparent variations I heard between the 2 samples Amazon offered have been longer pauses between sentences within the second pattern).
Amazon is also including its information and conversational talking types for the Matthew and Joanna voices from Amazon Polly, its neural network-based text-to-speech AWS service, and is including its information talking model to Lupe, its US Spanish voice.
The corporate additionally says builders can now use the information and conversational talking model in 29 languages when constructing new Alexa expertise.