Automatic Speech Recognition (ASR) is essentially a problem of patternclassification, however, the time dimension of the speech signal hasprevented to pose ASR as a simple static classification problem. SupportVector Machine (SVM) classifiers could provide an appropriate solution,since they are very well adapted to high-dimensional classification problems.Nevertheless, the use of SVMs for ASR is by no means straightforward,mainly because SVM classifiers require an input of fixed-dimension.In this paper we study the use of a HMM-based segmentation as a mean toget the fixed-dimension input vectors required by SVMs, in a problem ofisolated-digit recognition. Different configurations for all the parametersinvolved have been tested. Also, we deal with the problem of multi-classclassification (as SVMs are initially binary classifers), studying two of themost popular approaches: 1-vs-all and 1-vs-1.
展开▼
