Developers making voice apps for Amazons Alexa can now use different voices to power their experiences. The voices are being provided for free through the AWS text-to-speech service Polly. Created in 2016 with the same conversational AI that powers Alexa, Polly can speak in nearly 50 voices in 24 languages, but at its start, only eight Polly voices synthesized to sound more human are being made available in developer preview for English-speaking Alexa skills in the United States. Prior to the introduction of voices from Polly, developers were able to change a voice in a skill experience with the use of pre-recorded MP3 files. Skills developers will be able to add Polly voices using Speech Synthesis Markup Language (SSML). First introduced for Alexa last year, SSML is a set of tags that can be added when constructing an Alexa skill for human-sounding expressions like Boom! and Bada Bing, noises like sighs and expletive beeps, and the ability to lower Alexas voice to a whisper. The ability to use different voices is the latest feature extended to Alexa developers. In-skill purchases opened for Alexa developers in the U.S. earlier this month. The news comes a week after Google gave its Assistant six new voices, plans for an Assistant voice with singer John Legend, and the debut of Duplex, AI that uses WaveNet speech synthesis to generate a human-sounding voice that makes reservations or schedules appointments over the phone. No word yet on when Alexa will get more voice options or the ability to give users the choice to speak with a male or female voice.
You won't have to hear the same voice in every app. Do you wish Alexa would respond to your commands with different voices? You'll get your wish soon. Amazon has released a developer preview that lets Alexa skill developers use a selection of voices from Polly (Amazon's text-to-speech service). An adventure story can give each character a different voice, for instance. Creators could already 'hack' additional voices in through MP3 recordings and similar techniques, but this is decidedly more elegant. The feature currently only allows using eight US English voices (Polly supports roughly two dozen languages), and it should take a while before multi-voice skills arrive. However, the timing couldn't be much better. Google just added six voices to Assistant. While this clearly isn't the same (many skills will just use whatever regional voice you have), it should lead both to some basic variety and voice apps that are considerably more complex.