General description and recognition of 37 Chinese speech sounds

Abstract

The frequencies and levels of first three formants of a11 37 Chinese speech sounds, their time varying wave-forms, and their spectrograms were measured using a D.G.C. NOVA 2 Digital Computer, H.P. 7210A X-Y Digital Plotter, BARCO CTVM 2/51 or H.P. 131OA. Screen Display and some interface circuits. Using Fensom’s program (Reference 9 and. Appendix 6), the frequencies and. levels of these first three formants were displayed on the H.P. 1310A. For the sake of brevity, only the results for five vowel sounds and the first two formant frequencies of eleven other vowels were listed. Useful theories in the classification of speech sounds have been obtained by previous workers using formant frequencies F1 and. F2 (i.e. Reference 2), and the author used this approach for the Chinese speech sounds. Maximum likelihood regions (refer (2)*) were computed and. used to classify these phonetic sounds. Spectrum variations of vowels affected by consonants and tones are mentioned in Chapter I.4. The tones of those phonetic sounds can be observed (i.e. TabLe 4) as pitch (i.e. fundamental frequency) is varied.