Distributed discussion diarisation

In this paper we present Disca, a tool to analyze discussions in terms of which person is speaking at what time. We rely on a set of smartphones collaborating in detecting the most likely speaker at every given moment in real time. Each pair of smartphones observes a time difference of arrival pattern that is caused by the location of the different participants. The set of observations between all pairs of smartphones is then used to identify speakers on-line. To achieve this, clock differences and clock drifts between devices are estimated and compensated. Ultimately, participants are found by clustering time difference of arrival measurements which are unique for distinct speakers. We implement the system as an Android application and show that for more than 90% of time windows the correct speaker can be identified. To cope with heterogeneous hardware of Android smartphones, the computational burden is dynamically distributed among all participating smartphones according to their performance.