论文部分内容阅读
众所周知,传统算法在实现汉语语音的声韵母分割方面存在着很大的困难,这主要是因其处理方法从本质上讲属于一种线性映射,故在处理声韵母分割这一类非线性问题方面有其局限性.为此本文提出并实现了一种基于BP人工神经元网络的汉语语音声韵母分割算法。计算机模拟实验结果表明,该算法只需对极少数典型音节进行简单训练,便可实现汉语语音的声韵母分割且分割精度远大于传统算法所能获得的精度。
As we all know, the traditional algorithm has great difficulties in the vocal-vowels segmentation of Chinese phonetics, mainly because its processing method is essentially a linear mapping, so it is very important to deal with the non-linear problem of vowel segmentation Has its limitations. Therefore, this paper proposes and implements a Chinese vowel segmentation algorithm based on BP artificial neural network. Computer simulation results show that the proposed algorithm can achieve vocal-vowel segmentation of Chinese phonetic spell with only a few typical syllables and the segmentation accuracy is far greater than that obtained by traditional algorithms.