Conversation Finder is based on a decentralized network of body-worn wireless sensor nodes that independently try to figure out with who I am in a face-to-face conversation with.

The way we detect who I am talking to is by looking at alignment of speech—the way we take turns when we talk to each other. Each node has a microphone and sends out messages via RF when I am talking, and in turn receives messages from close-by nodes. It then aggregates this information and continuously updates a list of people it thinks I am talking to.

Eventually, a node can be queried for its list by the agent that controls my communication device. Depending on my conversational status, the agent might or might not interrupt me for a cellphone call.

This work makes mobile communication less aggravating because it makes cellphones sensitive to our face-to-face interactions.

One minute explanatory video [RealVideo] [MPG]