Recognition Using Classification and Segmentation Scoring

Recognition Using Classification and Segmentation Scoring* Owen Kimball t, Mari Ostendorf t, Robin Rohlicek t Boston University :~ BBN Inc. 44 Cummington St. 10 Moulton St. Boston, MA 02215 Cambridge, MA 02138 ABSTRACT Traditional statistical speech recognition systems typically make strong assumptions about the independence of obser- vation frames and generally do not make use of segmental information. In contrast, when the segmentation is known, existing classifiers can readily accommodate segmental infor- mation in the decision process. We describe an approach to connected word recognition that allows the use of segmental information through an explicit decomposition of the recog- nition criterion into classification and segmentation scoring. Preliminary experiments are presented, demonstrating that the proposed framework, using fixed length sequences of cep- stral feature vectors for classification of individual phonemes, performs comparably to more traditional recognition ap- proaches that use the entire observation sequence. We expect that performance gain can be obtained using this structure with additional, more general features. 1. INTRODUCTION Although hidden-Markov-model (HMM) based speech recognition systems have achieved very high perfor- mance, it may be possible to improve on their perfor- mance by addressing the known deficits of the HMM. Perhaps the most obvious weaknesses of the model are the reliance on frame-based feature extraction and the assumption of conditional independence of these features given an underlying state sequence. The assumption of independence disagrees with what is known of the ac- tual speech signal, and when this framework is accepted, it is difficult to incorporate potentially useful measure- ments made across an entire segment of speech. Much of the linguistic knowledge of acoustic-phonetic properties of speech is most naturally expressed in such segmental measurements, and the inability to use such measure- ments may represent a significant loss in potential per- formance. In an attempt to address this issue, a number of mod- els have been proposed that use segmental features as the basis of recognition. Although these models al- low the use of segmental measurements, they have not yet achieved significant performance gains over HMMs *This research was jointly funded by NSF and DARPA under NSF grant number IRI-8902124. because of difficulties associated with modeling a vari- able length observation with segmental features. Many of these models represent the segmental characteristics as a fixed-dimensional vector of features derived from the variable-length observation sequence. Although such features may work quite well for classification of individ- ual units, such as phonemes or syllables, it is less obvious how to use fixed-length features to score a sequence of these units where the number and location of the units is not known. For example, simply taking the product of independent phoneme classification probabilities using fixed length measurements is inadequate. If this is done, the total number of observations used for an utterance is F x N, where F is the fixed number of features per segment and N is the number of phonemes in the hypoth- esized sentence. As a result, the scores for hypotheses with different numbers of phonemes will effectively …

Content truncated to 500 words