No abstract

One of the approaches to automatic speech recognition is a distinctive feature-based speech recognition system, in which each of the underlying word segments is represented with a set of distinctive features. This thesis presents a study concerning acoustic attributes used for identifying the place of articulation features for stop consonant segments. The acoustic attributes are selected so that they capture the information relevant to place identification, including amplitude and energy of release bursts, formant movements of adjacent vowels, spectra of noises after the releases, and some temporal cues.An experimental procedure for examining the relative importance of these acoustic attributes for identifying stop place is developed. The ability of each attribute to separate the three places is evaluated by the classification error based on the distributions of its values for the three places, and another quantifier based on F-ratio. These two quantifiers generally agree and show how well each individual attribute separates the three places.Combinations of non-redundant attributes are used for the place classifications based on Mahalanobis distance. When stops contain release bursts, the classification accuracies are better than 90%. It was also shown that voicing and vowel frontness contexts lead to a better classification accuracy of stops in some contexts. When stops are located between two vowels, information on the formant structures in the vowels on both sides can be combined. Such combination yielded the best classification accuracy of 95.5%. By using appropriate methods for stops in different contexts, an overall classification accuracy of 92.1% is achieved.Linear discriminant function analysis is used to address the relative importance of these attributes when combinations are used. Their discriminating abilities and the ranking of their relative importance to the classifications in different vowel and voicing contexts are reported. The overall findings are that attributes relating to the burst spectrum in relation to the vowel contribute most effectively, while attributes relating to formant transition are somewhat less effective. The approach used in this study can be applied to different classes of sounds, as well as stops in different noise environments.

Most of the researches in synchronization of audio and text have been focusing on the synchronization at the level of utterance. However, to generate audio books in unstructured language like Thai from live speech, a finer level of synchronization is necessary. We propose an algorithm to synchronize live speech with its corresponding transcription in real time at syllabic unit. The proposed algorithm employs the syllable endpoint detection method and the syllable landmark detection method with bandlimited intensity as features. The experiment was conducted with LOTUS datasets and the results were compared with baseline ASR-based syllable detection. We evaluated our algorithm by measuring its error through error aberration, which is the difference of the actual number of syllables and the detected syllables for each phrase, and found average total error aberration of the proposed algorithm to outperform that of the baseline. The average total error aberrations are 11.54 and 34.21 for the proposed method and the baseline respectively. We also found the reference deviation from our method to be better than that of the baseline as well.

No abstract

In Thai, tonal information is a crucial component for identifying the lexical meaning of a word. Consequently, Thai tone classification can obviously improve performance of Thai speech recognition system. In this article, we therefore reported our study of Thai tone classification. Based on our investigation, most of Thai tone classification studies relied on statistical machine learning approaches, especially the Artificial Neural Network (ANN)-based approach and the Hidden Markov Model (HMM)-based approach. Although both approaches gave reasonable performances, they had some limitations due to their mathematical models. We therefore introduced a novel approach for Thai tone classification using a Hidden Conditional Random Field (HCRF)based approach. In our study, we also investigated tone configurations involving tone features, frequency scaling and normalization techniques in order to fine-tune performances of Thai tone classification. Experiments were conducted in both isolated word scenario and continuous speech scenario. Results showed that the HCRF-based approach with the feature F_dF_aF, ERB-rate scaling and a z-score normalization technique yielded the highest performance and outperformed a baseline using the ANNbased approach, which had been reported as the best for the Thai tone classification, in both scenarios. The best performance of HCRF-based approach provided the error rate reduction of 10.58% and 12.02% for isolated word scenario and continuous speech scenario respectively when comparing with the best result of baselines.