PLM 2005 ABSTRACTS VAULT
see http://elex.amu.edu.pl/ifa/plm
for further details
This talk describes work carried out over the past 5 years to collect and annotate a very large corpus of spontaneous conversational speech. Our most surprising finding concerns the amount of non-verbal information that is transmitted (as a side-channel?) along with the traditional linguistic information in speech. We propose a framework for the modelling of such information and present some examples of how it can be used in speech technology.