Minimum Bayes Error Feature Selection for Continuous Speech Recognition
George Saon, Mukund Padmanabhan · 2000
We consider the problem of designing a linear transformation \t 4 of rank , which projects the features of a classifier onto such as to achieve minimum Bayes error (or probability of misclassification). Two avenues will be explored: the first is to maximize the -average divergence between the class densities and the second is to minimize the union Bhattacharyya bound in the range of . While both approaches yield similar performance in practice, they outperform standard LDA features and show a 10% relative improvement in the word error rate over state-of-the-art cepstral features on a large vocabulary telephony speech recognition task. 1