This dissertation investigates statistical optimization of acoustic models in speech recognition. Two new optimization methods are proposed for phonetic decision tree (PDT) search and Hidden Markov modeling (HMM)-the knowledge-based adaptive PDT algorithm and the HMM gradient boosting algorithm. Investigations are conducted to applying both methods to improve word error rate of the state-of-the...

Morphologically derivative languages form words by fusing stems and suffixes, are important to be extracted in order make cross lingual alignment knowledge transfer. As there phonetic harmony disharmony when linguistic particles combined, both morphological changes need analyzed. This paper proposes a multilingual stemming method that learns morpho-phonetic automatically based on character embe...

Coarticulation is an important phenomenon that affects the realization of phonetic segments in fluent speech. The effects of coarticulation are prominent in both spectral and temporal domains. Various effects of phonetic contexts on the duration of the adjacent phonetic segments have been previously reported based on individual distinctive features (e.g., voiced stops lengthen and unvoiced stop...

This instrumental study of the phonetic contrast between English /ptk/ and /bdg/ produced by Saudi Arabians reveals that native-language phonetic nonTis may carryover to production of target-language sounds. Despite the existence of phonetic interference, however, the present cross-sectional study suggests that Saudi learners gradually approximate the phonetic norms of English, at least insofar...

Over the last decades, a considerable body of evidence has accumulated that speakers have detailed phonetic knowledge of a type which is not readily modelled using the categories and categorical rules of phonological theory. One line of evidence is systematic differences between languages in fine details of pronunciation. For example, it is known that Spanish and English differ systematically i...