Author:
Tudor Barbu
Affiliation:
Institute of Computer Science, Romania
Keyword(s):
Speech recognition, Discrete speech, Vocal sound, Mel cepstral analysis, Hausdorff metric, Feature vectors, Supervised classification, Training set.
Related
Ontology
Subjects/Areas/Topics:
Multimedia
;
Multimedia and Communications
;
Telecommunications
;
Traffic Modeling for Multimedia
Abstract:
In this work we provide an automatic speaker-independent word-based discrete speech recognition approach. Our proposed method consist of several processing levels. First, an word-based audio segmentation is performed, then a feature extraction is applied on the obtained segments. The speech feature vectors are computed using a delta delta mel cepstral vocal sound analysis. Then, a minimum distance supervised classifier is proposed. Because of the different dimensions of the speech feature vectors, we create a Hausdorff-based nonlinear metric to measure the distance between them.