While current digital assistants require a wake word such as “Alexa” or “OK Google,” the patent describes how this word could come at any point during a command, including the end. Given examples include: “Play some music, Alexa,” and “Play some music, Alexa. The Beatles, please.”
In order to perform this function, devices such as the Echo speaker line would need to be constantly recording and storing audio so they can “look backwards” should they hear the word “Alexa.”
While the system has obvious privacy implications, the patent states that all captured speech doesn’t need to be sent to Amazon. The process could also be configured to store between only 10 seconds and 30 seconds of audio at a time.
Despite these assurances, the idea of a smart speaker that is constantly recording conversations is unlikely to sit well with owners. Last month’s report of thousands of Amazon workers around the world listening to voice recordings didn’t help the Echo devices’ reputation for respecting privacy. We’ve also seen murder cases where Echo recordings have been examined for evidence.
Another Amazon patent from last year described how Alexa could be constantly listening for certain trigger words, such as “love,” “like,” and “hate,” that will help it ascertain a user’s preferences and dislikes, which could then be used for targeted ads.
As with all patents, there’s no guarantee this one will become a real-world feature. An Amazon spokesperson told Engadget: "Like many companies, we file a number of forward-looking patent applications that explore new scientific ideas that may not make it into customer-facing products. Patents take multiple years to receive and do not necessarily reflect current or near-future state of products and services."