Speech Segmentation Using Probabilistic Phonetic Feature Hierarchy and Support Vector Machines

We propose a method that combines acoustic-phonetic knowledge with support vector machines for segmentation of continuous speech into five classes vowel, sonorant consonant, fricative, stop and silence. We show that by using a probabilistic phonetic feature hierarchy, only four classifiers are required to recognize the five classes. Due to the probabilistic nature of the hierarchy, the method overcomes the disadvantage of the traditional acoustic-phonetic methods where the error is carried down the hierarchy. On the other hand, the the hierarchical approach allows the use of comparable amount of training data of two classes that each classifier is designed to discriminate. The segmentation method with 13 knowledge based parameters performs considerably better than a context-independent Hidden Markov Model (HMM) based approach that uses 39 mel-cepstrum based parameters. The probabilistic nature of the algorithm allows the method to be extended to phoneme and word recognition with a small number of classifiers.