The news is decidedly rich on the Seattle side! After the launch of Favourites Smart Home and the revelation of the release date of the new Echo Show 15, Amazon announces today to deploy the conversation mode ofAlexa. A novelty already mentioned last year which uses artificial intelligence and will allow you to interact more naturally with Alexa.
A new conversation mode with Alexa
Rohit Prasad, vice-president ofAmazon and chief scientist ofAlexa, had described in September 2020 three major advances coming soon for the voice assistant, in particular a mysterious mode allowing to converse in a more fluid way. Today the team Alexa Speech announces the deployment, on Echo Show 10 only, of the new conversation mode with Alexa.
Concretely, you will not have no need to repeat the wake-up word Alexa to place all your voice commands. Thanks to a combination of visual and acoustic cues, the AI ofAlexa will let it know when your speech is directed at the device and if you are expecting a response.
It will also be possible to invoke the Conversation mode saying « Alexa, join the conversation " and leave it by simply saying " Leave the conversation ". Alternatively, the assistant will exit the mode on her own if there is no interaction for a short time.
In addition, the conversation mode will offer you the possibility ofinteract with several and simultaneously with Alexa. This may be useful when requesting suggestions from Alexa or for skills requiring it, or to intervene in a conversation.
Le deep learning at the service of our connected homes
If this new conversation mode ofAlexa may seem like a small evolution to you, you have to understand that it is actually a major breakthrough. A feat that required innovations in several areas, in particular the orientation detection visual devices, voice activity detection based on audio, but also the fusion of audiovisual functionalities.
The scientific teams ofAmazon must indeed have resulted in models of deep neural networks to determine the orientation of the user's head, or to know when the speech is directed towards the device. This also results in a significant reduction of false alarms triggered by ambient noise… Another great news!