5th European Conference on Speech Communication and Technology

Rhodes, Greece
September 22-25, 1997

FDVQ Based Keyword Spotter Which Incorporates A Semi-Supervised Learning for Primary Processing

Chakib Tadj (1), Pierre Dumouchel (2), Franck Poirier (3)

(1) Ecole de Technologie Superieure, Montreal; Quebec, Canada
(2) Centre de Recherche Informatique de Montréal, Quebec, Canada
(3) Institut Universitaire Professionnalisé, Vannes, France

In this paper, we present a novel hybrid keyword spotting system that combines supervised and semi-supervised competitive learning algorithms. The first stage is a S-SOM (Semi-supervised Self- Organizing Map) module which is specifically designed for discrimination between keywords (KWs) and non-keywords (NKWs). The second stage is an FDVQ (Fuzzy Dynamic Vector Quantization) module which consists of discriminating between KWs detected by the first stage processing. The experiment on Switchboard database has show an improvement of about 6% on the accuracy of the system comparing to our best keyword-spotter one.

Full Paper

Bibliographic reference.  Tadj, Chakib / Dumouchel, Pierre / Poirier, Franck (1997): "FDVQ based keyword spotter which incorporates a semi-supervised learning for primary processing", In EUROSPEECH-1997, 2799-2802.