An objective method, using a multi-band analysis technique, was proposed for analyzing plosive consonants in cleft palate speech. At first, the speech signal is decomposed in frequency domain using an auditory filter-...An objective method, using a multi-band analysis technique, was proposed for analyzing plosive consonants in cleft palate speech. At first, the speech signal is decomposed in frequency domain using an auditory filter-bank. Then, the sample-based features, namely cumulative energy and its increment speed, in each band were computed. Finally, using principle component analysis, these features were fused into one combined feature vector for assessment. Since the algorithm is based on perceptual properties of human auditory ear using non-uniform and multi-band analysis, the improvements of the consistence between the proposed approach and subjective evaluation are obtained.展开更多
It is conventionally believed that,once one has acquired his native phonemes,the acoustic realizations are relatively stable since the later stage of childhood,and of high intra-speaker consistency throughout adulthoo...It is conventionally believed that,once one has acquired his native phonemes,the acoustic realizations are relatively stable since the later stage of childhood,and of high intra-speaker consistency throughout adulthood.With evidence from plosives produced in connected speech in Standard Chinese,the present study shows that even middle-aged speakers show gradually increased voice onset time(VOT)in the production of plosives in general.Therefore,at least for phonemes such as plosives,the acoustic realizations are further developed and in an ongoing,dynamic process of change even in adulthood.Furthermore,by considering both raw VOT and VOT ratios,the present study also shows that the increase in VOT cannot be completely due to decrease in speech rate as speakers age,but could be related to speakers’adjustments to the physiological changes due to aging in speech production.展开更多
基金supported by the National Natural Science Foundation of China(60875014,60772039)
文摘An objective method, using a multi-band analysis technique, was proposed for analyzing plosive consonants in cleft palate speech. At first, the speech signal is decomposed in frequency domain using an auditory filter-bank. Then, the sample-based features, namely cumulative energy and its increment speed, in each band were computed. Finally, using principle component analysis, these features were fused into one combined feature vector for assessment. Since the algorithm is based on perceptual properties of human auditory ear using non-uniform and multi-band analysis, the improvements of the consistence between the proposed approach and subjective evaluation are obtained.
基金funded by the National Social Science Foundation of China(Grant No.19CYY021)
文摘It is conventionally believed that,once one has acquired his native phonemes,the acoustic realizations are relatively stable since the later stage of childhood,and of high intra-speaker consistency throughout adulthood.With evidence from plosives produced in connected speech in Standard Chinese,the present study shows that even middle-aged speakers show gradually increased voice onset time(VOT)in the production of plosives in general.Therefore,at least for phonemes such as plosives,the acoustic realizations are further developed and in an ongoing,dynamic process of change even in adulthood.Furthermore,by considering both raw VOT and VOT ratios,the present study also shows that the increase in VOT cannot be completely due to decrease in speech rate as speakers age,but could be related to speakers’adjustments to the physiological changes due to aging in speech production.