摘要
从提高单词层的语音识别精度到提供个性化的发音训练 ,音节切分都有着广泛的应用领域。该文提出了一个利用声音特性对英文单词进行快速音节切分的算法。该方法首先通过对能量和过零率参数的分析 ,划出粗略的音节边界 ,然后检测峰值点 /谷值点的基音周期参数来做修定。实验结果显示 。
Syllable identification has wide uses from increasing the accuracy of word-level recognition to personalizing pronunciation trainning. This paper describes a syllable identification method by using acoustic properties. This method analyses energy and zerocrossing properties to decide a rough boundery of syllables first, then uses pitch properties of peak/valley to refine it. Experimental results show that this algorithm has good performance both in speed and accuracy.
出处
《计算机仿真》
CSCD
2005年第2期86-88,共3页
Computer Simulation
关键词
音节切分
声音特性
方法
Syllable identification
Acoustic properties
Method
