Fundamental Frequency Extraction and Tone Recognition of Chinese Continuous Two-character-words

Chinese is a language composed of syllables with tones. Initials, finals, and tones are the three main attributes of Chinese syllables. In this paper, continuous two-character-words are segmented through syllables, and the effective segmentation of syllables is performed by using the characteristics of short-time energy and short-time zero-crossing rate of initial. It is useful to extracting the pitch frequency with the autocorrelation function and average amplitude difference function method . This paper proposes a speaker-independent two-character-words fuzzy identification method and uses the main characteristics of the pitch curve to form a fuzzy membership function. The tone recognition algorithm uses the decision function to classify the tone pattern. The method has simple, reliable, and good real-time characteristics.