If you watched Saturday Night Live last week, you saw a skit in which senior citizens routinely call Amazon’s intelligent assistant Alexa names like “Odessa” and “Allegra.” The Amazon Echo Silver, the show jokes, was made to deal with curmudgeons and copious levels of chit-chat.
“It’s super loud and responds to any name even remotely close to Alexa so that they can find out the weather, sports, even local news and pop culture,” the video says.
Well, today Amazon announced plans to do the exact opposite of what that sketch depicts, choosing instead to launch two-step verification of the “Alexa” wake word.
Alexa-enabled devices like the Echo know to start recording audio of voice for speech-to-text translation when they hear a wake word, most commonly “Alexa” or “Echo.”
“With this update, the wake word engine (WWE) on the device handles the initial detection of ‘Alexa’, and then a secondary cloud-based check verifies the utterance,” wrote Alexa evangelist Ted Karczewski in a blog post today. “If a false wake word is detected, the verification process directs the device to close the audio stream and turn off the LED indicator.”
While cloud-based wake word verification could reduce the possibility that we ever see anything that resembles an Echo Silver, it could also reduce the number of unintended recordings by Alexa-enabled devices. Amazon has offered no evidence that two-step verification is more precise, more precise wake word recognition could potentially lead to fewer subpoenas for Alexa recordings like the kind requested in an Arkansas homicide trial late last year.