论文部分内容阅读
根据汉语语音信号的特点．该文提出一种简化的自相关基音提取的方法，算法具有简便、快速、准确和易于实现等特点。此方法不仅可用于孤立字．更适用于提取二字词的声词。实验表明，所得到的１５种二字词声调的基本模式和轻声变调等都与已有的研究报导相符。含有清声母音节时，二字词声调能提供音节分割和声韵分离的准确信号．基音提取成功率高于９０％。
According to the characteristics of Chinese voice signal. This paper presents a simplified method of auto-correlation pitch extraction. The algorithm is simple, fast, accurate and easy to implement. This method can be used not only for isolated words. More suitable for extracting two words of sound words. The experiment shows that the basic pattern of the 15 kinds of two-word tone and the soft tone change are consistent with the existing research reports. When a clear vowel tone is included, the two-word tone provides an accurate signal of syllable segmentation and phonological separation. Pitch extraction success rate higher than 90%.