Meeting Data Collection Specifications

This document specifies a process for collecting a new corpus of meetings in the IDIAP Smart Meeting Room. This document is a working draft that is expected to be updated and augmented throughout the data collection process. This follows from an earlier data collection effort that resulted in a corpus of 60 scripted meetings (30 train, 30 test), each of 5 minutes duration (now available at \textsf{}). The current data collection effort aims to address some of the limitations of the previous corpus, as well as to cater for a richer variety of research tasks.

Related material