Two discriminative methods for solving tone problems in Mandarin speech recognition are presented. First, discriminative training on the HMM （hidden Markov model） based tone models is proposed. Then an integration t...Two discriminative methods for solving tone problems in Mandarin speech recognition are presented. First, discriminative training on the HMM （hidden Markov model） based tone models is proposed. Then an integration technique of tone models into a large vocabulary continuous speech recognition system is presented. Discriminative model weight training based on minimum phone error criteria is adopted aiming at optimal integration of the tone models. The extended Baum Welch algorithm is applied to find the model-dependent weights to scale the acoustic scores and tone scores. Experimental results show that tone recognition rates and continuous speech recognition accuracy can be improved by the discriminatively trained tone model. Performance of a large vocabulary continuous Mandarin speech recognition system can be further enhanced by the discriminatively trained weight combinations due to a better interpolation of the given models.展开更多
The brain localization debate of lexical tone processing concerns functional hypothesis that lexical tone, owing to its strong linguistic features, is dominant in the left hemisphere, and acoustic hypothesis that all ...The brain localization debate of lexical tone processing concerns functional hypothesis that lexical tone, owing to its strong linguistic features, is dominant in the left hemisphere, and acoustic hypothesis that all pitch patterns, including lexical tone, are dominant in the right hemisphere due to their acoustic features. Lexical tone as a complex signal contains acoustic components that carry linguistic, paralinguistic, and nonlinguistic information. To examine these two hypotheses, the current study adopted triplet stimuli including Chinese characters, their corresponding pinyin with a diacritic, and the four diacritics representing Chinese lexical tones. The stimuli represent the variation of lexical tone for its linguistic and acoustic features. The results of a listening task by Mandarin Chinese speakers with and without aphasia support the functional hypothesis that pitch patterns are lateralized to different hemispheres of the brain depending on their functions, with lexical tone to the left hemisphere as a function of linguistic features.展开更多
When scientific research began in early twentieth-century China,a key issue was the acquisition of reliable empirical information through objective and precise observations.This article examines a specific case where ...When scientific research began in early twentieth-century China,a key issue was the acquisition of reliable empirical information through objective and precise observations.This article examines a specific case where a scientist grappled with such an issue:the linguist Chao Yuen Ren’s application of mechanical means in his phonetic studies.In the 1920s–1930s,Chao conducted a series of field and lab studies on the dialects in southern and central China.In contrast to traditional scholars’exclusive reliance on sharp ears and rhyme books,Chao employed mechanical devices to inscribe and analyze the spectrographs of dialectical tones and used phonographs to record the articulations of his subjects.It is demonstrated that Chao’s machines not only provided a new method of observation;they also altered the theoretical understanding of certain fundamental categories in Chinese phonology,such as tones.Moreover,Chao did not aim to replace human perception with automatic mechanisms in empirical investigations.Rather,the use of machines in his research called for an active and engaged scientific persona.展开更多
文摘Two discriminative methods for solving tone problems in Mandarin speech recognition are presented. First, discriminative training on the HMM （hidden Markov model） based tone models is proposed. Then an integration technique of tone models into a large vocabulary continuous speech recognition system is presented. Discriminative model weight training based on minimum phone error criteria is adopted aiming at optimal integration of the tone models. The extended Baum Welch algorithm is applied to find the model-dependent weights to scale the acoustic scores and tone scores. Experimental results show that tone recognition rates and continuous speech recognition accuracy can be improved by the discriminatively trained tone model. Performance of a large vocabulary continuous Mandarin speech recognition system can be further enhanced by the discriminatively trained weight combinations due to a better interpolation of the given models.
文摘The brain localization debate of lexical tone processing concerns functional hypothesis that lexical tone, owing to its strong linguistic features, is dominant in the left hemisphere, and acoustic hypothesis that all pitch patterns, including lexical tone, are dominant in the right hemisphere due to their acoustic features. Lexical tone as a complex signal contains acoustic components that carry linguistic, paralinguistic, and nonlinguistic information. To examine these two hypotheses, the current study adopted triplet stimuli including Chinese characters, their corresponding pinyin with a diacritic, and the four diacritics representing Chinese lexical tones. The stimuli represent the variation of lexical tone for its linguistic and acoustic features. The results of a listening task by Mandarin Chinese speakers with and without aphasia support the functional hypothesis that pitch patterns are lateralized to different hemispheres of the brain depending on their functions, with lexical tone to the left hemisphere as a function of linguistic features.
文摘When scientific research began in early twentieth-century China,a key issue was the acquisition of reliable empirical information through objective and precise observations.This article examines a specific case where a scientist grappled with such an issue:the linguist Chao Yuen Ren’s application of mechanical means in his phonetic studies.In the 1920s–1930s,Chao conducted a series of field and lab studies on the dialects in southern and central China.In contrast to traditional scholars’exclusive reliance on sharp ears and rhyme books,Chao employed mechanical devices to inscribe and analyze the spectrographs of dialectical tones and used phonographs to record the articulations of his subjects.It is demonstrated that Chao’s machines not only provided a new method of observation;they also altered the theoretical understanding of certain fundamental categories in Chinese phonology,such as tones.Moreover,Chao did not aim to replace human perception with automatic mechanisms in empirical investigations.Rather,the use of machines in his research called for an active and engaged scientific persona.
