Amazon’s Alexa could soon understand requests without initial wakeword

  • TECH
  • Saturday, 25 May 2019

An Amazon patent has been published by the US Patent and Trademark Office describing a system that would allow Alexa to record audio without the utterance of a "wakeword", bringing users a more natural voice assistant experience. 

Instead of preceding a voice assistant command with a wakeword or phrase like "Hey Alexa", Amazon filed a patent which was made public on May 23 that outlines the potential of the voice assistant to understand and respond to requests like "Play some music Alexa" or "Play some music Alexa, the Beatles please". 

While this type of phrasing is more natural for people, it would require Alexa to record audio at all times: "When a wakeword is detected, the system may look backwards to the most recent marker that preceded the wakeword." 

Though users will still be required to say Alexa (or the respective wakeword for the voice assistant being used), it could potentially be placed in a much more natural position – a format that today's speech processing systems are not yet configured to handle – rather than always at the beginning of a command. 

Naturally, perpetual audio recording could raise privacy concerns among users; to answer this, Amazon suggested in the patent that the system could be programmed to only hold a limited amount of speech, like 10-30 seconds, at a given time. Because of general uneasiness around major tech companies recording personal data, the technology has not yet been implemented and may never be. 

In any case, this type of surveillance is necessary to create more natural communication experiences with voice assistants, and companies like Amazon will need to be responsible in deciding where the balancing point is between collecting data and respecting user privacy. – AFP Relaxnews

Article type: metered
User Type: anonymous web
User Status:
Campaign ID: 1
Cxense type: free
User access status: 3

Did you find this article insightful?


Across The Star Online