Automatic discrimination system for Chinese bilabial aspirated and unaspirated sounds pronounced by Japanese students
Akemi Hoshino · 2016
Chinese aspirates are usually difficult to pronounce for Japanese students learning Chinese. In particular, discriminating between utterances of aspirated and unaspirated sounds is very difficult. For self-learning and practice, we developed an automatic judgment system that enabled students to check their pronunciations of bilabial sounds using a computer. We extracted the features of correctly pronounced single-vowel bilabial aspirated sounds pa[p'a], pi[p'i], po[p'o], and pu[p'u], and bilabial unaspirated sounds ba[pa], bi[pi], bo[po], and bu[pu], by observing the evolution of the breathing power spectrum during both the voice onset time (VOT) and the voiced period of the sounds, uttered by 50 native Chinese speakers. We developed a 35-channel computerized filter bank to analyze the evolution of the breathing power spectrum and then automatically evaluated the utterances of 50 Japanese students. Our system rejected several samples that passed the screening of native Chinese speakers. The success rates of the system were higher than 93% and 95% for aspirated and unaspirated sounds, respectively.