摘要
以波形拼接式维吾尔语音合成系统研发为背景,在已建立的维吾尔语最小发音单位音节和音素作为合成基元的语音库基础上,对语料库中的所有音节、音素进行无损压缩,选择了运算速度快,便于实现的哈夫曼压缩。在解压过程中只解压人们所需的语音单元,而不需要解压整个语料库。实验结果表明,通过哈夫曼压缩算法对语料库进行压缩和解压,减小了语料库的占用空间,同时解压后的语音不失真,解压速度快。
This paper takes waveform concatenation Uyghur speech synthesis system as the research background,based on the syllable and phoneme corpus,which had been already established,takes the smallest pronunciation unit in Uyghur as the basic syntheses unit,selected the Huffman compression as the lossless compression algorithm,which is fast,easy to achieve,makes a compression to all the syllables and phonemes in the corpus.During decompression,it did not decompress the entire speech corpus but what we really want.The experimental results show that,this algorithm reduced the space and realized the decompression of voice without distortion and with a high speed.
出处
《信息技术》
2012年第10期11-14,共4页
Information Technology
基金
国家自然科学基金资助项目(61062008)
关键词
维吾尔语
语音合成
语料库
哈夫曼编码
压缩
解压
Uyghur language
speech synthesis
corpus
Huffman Code
compression
decompression
