27 Oct - Conferencia Michael Wagner
Emotion detection using both acoustic and linguistic information in children's speech
Día: Martes 27 de octubre
Hora: 11.30 h
Lugar: Sala de Seminarios D5-007
For obvious reasons, humans have evolved to being quite sensitive to the emotional state of another person, and we make use of cues from all available senses to detect, for example, an angry person. The automatic detection of emotions would be useful in various applications, but research in this field, and the necessary data, are quite limited so far. A recent experiment uses acoustic data from the verbal interaction of German schoolchildren with an Aibo robot dog, in which the robot is controlled by a wizard so as to be disobedient and evoke frustration and anger in the children. The data corpus of the children's utterances is transcribed orthographically and tagged with emotion labels. The experiment explores a large number of acoustic features for their utility in anger detection, and it also uses the emotionally salient words in the children's utterances, either obtained from the transcripts or from an automatic speech recognition system, to detect the anger emotion in those utterances.
----------
Michael Wagner was born in Münster, Germany, studied physics and mathematics at the Universities of Münster and Munich, and in 1973 received his Diplomphysiker degree from Ludwig-Maximilians-Universität in Munich with a thesis on the computer simulation of an elementary particle spectrometer. In 1979 he received his PhD in computer science from the Australian National University with a thesis on the acoustic analysis of speaker characteristics. Dr Wagner held research and teaching positions at the Technical University of Munich, National University of Singapore, Nixdorf AG, the University of Wollongong, the Australian Defence Force Academy, and the Australian National University. Since 1996 he has held the Chair in Computing of the University of Canberra, where at various times he has been Head of the School of Computing, Head of the Discipline of Software Engineering, Director of the Human-Computer Communication Lab and Director of the National Centre for Biometric Studies. He has also been a visiting researcher at the Universities of Amsterdam, Hong Kong and Duisburg, and at Siemens Research Laboratories in Munich. In 2009 he is on study leave at Technical University of Berlin, University of Karlsruhe, University of Avignon and Nanyang Technological University in Singapore. Michael Wagner is the author of more than 120 refereed publications in the field of speech science and technology.