Amazon Alexa gets better at reading audiobooks, articles and generally long content. The new vocal model, created by a deep learning system that focuses in particular on pauses in a conversation, varying them and adapting them to what human beings normally do. As you can hear in this comparison, it is precisely the pauses that make the content less “tiring”. There is also a greater variation in intonation.
They are not the only innovations introduced in the text-to-speech field: Amazon has also updated “Matthew” and “Polly” two vocal models that are part of the AWS offering, with the “newscast” style of speech that was presented on Alexa last November (and allows you to add an excited or serious tone to how some articles are read). Matthey and Polly also receive the style for long articles, while “Lupe”, the voice in American Spanish, only the news item.
All these concerns, for now, exclusively Alexa in American English. The style can be implemented immediately by the developers. No information on the arrival in India was communicated.
SOURCE
