Alexa Super Bowl Ad Didn’t Wake Up Your Echo and Took Top Honors
Amazon Alexa won the Super Bowl of ad spots during last night’s game according to the USA Today’s Ad Meter. It’s spot with Sir Anthony Hopkins, Rebel Wilson and Cardi B narrowly beat out the NFL Touchdown Celebrations ad. A particularly interesting element of the Alexa ad is that no Echo devices were activated despite the “Alexa” wake word being uttered many times throughout the 90-second television ad. An Amazon blog post discussed this “acoustic fingerprinting” feature in a Friday blog post.
When the 90-second advertisement airs during the game Sunday evening, millions of Echo devices won’t be unintentionally waking up to the Alexa phrase. This is possible because of acoustic fingerprinting technology that can distinguish between the ad and actual customer utterances.
Avoiding the Burger King / South Park Stunts
Many Voicebot readers will recall that Burger King won a top advertising award by deliberately waking up Google Home devices and directly them to recite an altered Wikipedia entry that praised the merits of the Whopper. Google moved quickly to suppress that effort, but did so through manual intervention. Amazon faced a similar challenge when the animated television show South Park incorporated some off-color language into a 2017 episode. In that instance, Echo devices across America added several items to users’ shopping lists. Amazon has now implemented an algorithm to suppress these types of false-triggers automatically. The post states:
“Our advertising, engineering, and science teams are able to anticipate major events like the Super Bowl, but what happens when someone like Tonight Show host Jimmy Fallon does a comedy routine about Alexa, which the team couldn’t anticipate?
“Manoj Sindhwani, director for Speech Recognition, explains that our teams build acoustic fingerprints on-the-fly within our AWS cloud. When multiple devices start waking up simultaneously from a broadcast event, similar audio is streaming to Alexa’s cloud services. An algorithm within Amazon’s cloud detects matching audio from distinct devices and prevents additional devices from responding. The dynamic fingerprinting isn’t perfect, but as many as 80 to 90 percent of devices won’t respond to these broadcasts thanks to the dynamic creation of the fingerprints.”
This is an important technical advance as always-listening voice assistants become more commonplace. Automating this process is critical since the frequency of these types of false-triggers is on the rise. What do you think? Let us know on Twitter.