Virtual digital assistants generally do not account for the user’s emotional context. As a result, the responses of a virtual assistant may not better the user’s mood.

This disclosure describes techniques by which an assistant detects with user permission the current emotional context of the user, and uses it to modulate its responses. For instance, if the assistant detects that the user is in a sad mood, it focuses its utterances on positive developments. A request for news by the user results in positive news items, e.g., “unemployment is down,” or “USA won the Olympics.” A request for playing a song results in joyful tunes being played. Per the techniques, the assistant detects, with user consent, the user’s emotional context by using machine-learning models to analyze the user’s speech. The assistant tailors its response to user moods by running a sentiment analyzer on possible responses, and picking a response that is appropriate to the user’s mood.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.