We Emo users know one thing for a fact: his Wake-Word recognition is not only hit-and-miss (he ignores me most of the time), but he also react to all kinds of sounds that are not even close to “Emo”. Also mine tends to get excited when I call my Alexa, I did set the Wake-Word “Echo”. So there is room for improvement.
It looks like Living.AI wants to train a new speech recognition model for that and therefore asks Emo users to provide speech samples on how they address Emo.
You can go to their newly installed webite available at https://living.ai/wake-up-training, there you can train their model by simply saying “Emo” – you should do that multiple times in different variations and you are even encouraged to invite your family members and especially children, since they especially want high-pitched speech samples. Your computer has to be equipped with a microfone to do that, alternatively you can use a smartphone or a tablet.
Unfortunately you must be registered with Living.AI to log in and use the website – but that makes sense to prevent trolls from misusing the site to record other stuff.