Smart Speaker Study: When Conversations Are Mistakenly Recorded

[engine]

A study published by Northeastern University tries to understand how and why smart speakers suddenly, mistakenly, record conversations.
It appears that devices wake up as much as 19-times per day, and the triggers are often words sounding like the trigger word.

For instance, with the Google Home Mini, these activations commonly occurred when the dialogue included words rhyming with “hey” (such as letter “A” or “They”) followed by something that starts with hard “G”, or that contains “ol” such as “cold and told”. Examples include “A-P girl”, “Okay, and what”, “I can work”, “What kind of”, “Okay, but not”, “I can spare”, “I don’t like the cold”.

[moniotrlab.ccis.neu.edu...]

[lammert]

Getting older myself, more and more people in my environment are having a diminished hearing ability (I am still in the denial phase :)) One of the symptoms is being less able to distinguish similar sounds.

With the microphones of these smart devices stuffed away to get a pretty design and with cost constraints which may limit the technical abilities of the microphones, I can imagine that the signal coming in under some situations simply doesn't contain all the information to reliably distinguish normal talk from commands to the smart speakers.