Automatic laughter detection in spontaneous speech using GMM–SVM method

Neuberger, Tilda and Beke, András (2013) Automatic laughter detection in spontaneous speech using GMM–SVM method. In: Text, Speech, and Dialogue, September 1-5, 2013, Pilsen.

Text Automatic laughter detection in spontaneous speech with GMM-SVM.pdf Restricted to Repository staff only Download (536kB) | Request a copy

Abstract

Spontaneous conversations frequently contain various non-verbal vo- calizations (such as laughters). The accuracy of a speech recognizer may de- crease in the case of spontaneous speech because of these non-verbal vocalization phenomena. The aim of the present research is to develop an accurate and efficient method in order to recognize laughters in spontaneous utterances. We used GMM in modeling the data and SVM for differentiating laughters from other speech events. The training and testing of the laughter detector were carried out using the BEA Hungarian spoken language database. The results show that the GMM-SVM system seems to be a particularly good method for solving this problem. This model can be used in speech recognition, speech processing and speaker diarization as well. Keywords: non-verbal communication, laughter, spontaneous speech

Actions (login required)

Edit Item