Amazon is developing a voice-mimicking feature for virtual assistant Alexa, that replicates the speech of people alive and dead!
Furthermore, WSJ adds that Amazon is joining other companies that are experimenting with creating digital memories of people after death.
E-commerce giant says its virtual assistant needs less than one minute of an audio recording to mimic someone’s speech
Alex’a tweet on conference call
📣 Amazon's annual Alexa developer conference is now open for registration!!! Get a first look at Alexa’s latest innovations to help grow your business.
— Alexa Developers (@alexadevs) June 2, 2022
Head scientist of Alexa, R. Prasad,
showed a video clip of the new feature in action at Amazon’s re: MARS 2022 conference on artificial intelligence. The clip showed a boy asking Alexa to read him “The Wizard of Oz” in the voice of his grandmother. Alexa then switched from its default voice to a softer voice.
Highlights from re:Mars official clip from Amazon
Mr. Prasad said Alexa is able to mimic voices after accessing less than a minute of recorded audio of a voice.
“While AI can’t eliminate that pain of loss, it can definitely make the memories last,” Mr. Prasad added
An Amazon spokesman declined to say when the new feature will be rolled out to consumers.
He said the new voice-mimicking feature is something the company has been exploring based on recent advancements in text-to-speech technology.
It allows the company to produce a high-quality voice with less data rather than relying on hours of recording in a professional studio, he said.
Using artificial intelligence to duplicate the voice of someone who has already died could raise ethical issues.
Amazon has been working for years on boosting Alexa’s capabilities beyond setting timers and playing music.
Mr. Prasad, who has led efforts to expand Alexa’s capabilities, said he believes that artificial intelligence will soon become part of daily life, adding that ‘we are living in the golden era of AI where our dreams and science fiction are becoming a reality’