基于特征的藏文音节识别算法
|
摘要
为了实现藏文的文语转换(TTS),提出基于字符投影变换特征的藏文音节识别算法。该算法以音节为基元,选择并提取音节中由字符列投影变换组成的特征向量,以此建立音节特征库;通过查表算法对藏文音节进行识别。算法还包括藏文文本的规范化和音节切分两部分内容。通过理论分析和算法测试实验证明:提取的特征向量与藏文音节一一对应,藏文音节识别率达到100%,且特征的提取过程简便易行。该算法已经成功应用于藏文的文语转换系统。
For the purpose of realizing the translation of Tibetan language(TTS),a Tibetan syllable recognition algorithm based on character projection transformation is proposed. The algorithm is on the basis of syllable,selects and extracts feature vectors formed by projection transformation in the character syllables in the hope of setting up syllable feature bank,and it also identifies Tibetan syllable by the method of look-up table algorithm. The algorithm includes two parts: the normalization of Tibetan text and the segmentation of syllables. Theoretical analysis and algorithm test has proved that the extracted feature vectors are each corresponded to Tibetan syllables,and the rate of Tibetan syllable recognition reaches 100%. The process of feature extraction is simple and feasible. The algorithm has been successfully applied to the Tibetan language translation system.
For the purpose of realizing the translation of Tibetan language(TTS),a Tibetan syllable recognition algorithm based on character projection transformation is proposed. The algorithm is on the basis of syllable,selects and extracts feature vectors formed by projection transformation in the character syllables in the hope of setting up syllable feature bank,and it also identifies Tibetan syllable by the method of look-up table algorithm. The algorithm includes two parts: the normalization of Tibetan text and the segmentation of syllables. Theoretical analysis and algorithm test has proved that the extracted feature vectors are each corresponded to Tibetan syllables,and the rate of Tibetan syllable recognition reaches 100%. The process of feature extraction is simple and feasible. The algorithm has been successfully applied to the Tibetan language translation system.
引文
[1]龙从军,刘汇丹,吴健.藏文国际音标(拉萨音)自动转换研究[J].中文信息学报,2016,30(5):203-208,214.
[2] Ogwu E J,Talib M,Odejobi O A.Text-to-speechprocessing using African language as case study[J].Journal of Discrete Mathematical Sciences andCryptography,2006,9(2):365-382
[3] Romportl J,Matou?ek J. Several Aspects of Ma-chine-Driven Phrasing in Text-to-Speech Systems[J]. The Prague Bulletin of Mathematical Linguis-tics,2011,95(1):51-62.
[4]蔡莲红,张维,胡其炜.文语转换系统中汉语韵律的学习和模拟[J].清华大学学报:自然科学版,1998,38(S1):95-98
[5]倪宏.一个实用的汉语文语转换系统[J].小型微型计算机系统,1995,16(11):42-46.
[6]张家騄,吕士楠,齐士钤,等.汉语文语转换系统的研究[J].信号处理,1989,5(1):1-7.
[7]蔡莲红,魏华武.汉语文-语转换系统的研究与实现[J].应用声学,1994,13(6):1-5.
[8]王维兰.藏文基本字符识别算法研究[J].西北民族学院学报:自然科学版,1999,20(3):20-23,51.
[9]王浩军,赵南元,邓钢轶.藏文识别的预处理[J].计算机工程,2001,27(9):93-96.
[10]王维兰,陈万军.基于笔划特征和MCLRNN模型的联机手写藏文识别[J].计算机工程与应用,2008,44(14):91-93.
[11]周纬,陈良育,曾振柄.基于几何形状分析的藏文字符识别[J].计算机工程与应用,2012,48(18):201-205.
[12]蔡晓娟,黄鹤鸣.基于多投影的脱机手写藏文字符特征提取方法[J].计算机技术与发展,2016,26(3):93-96.
[13]Fernando Perez-Tellez,John Cardiff,Paolo Rosso,David Pinto. Weblog and short text featureextraction and impact on categorisation[J]. Journalof Intelligent&Fuzzy Systems,2014,27(5):2529-2544.
[14]YUN Chang,JIA Lee,Omar Rijal,Syed Bakar. Effi-cient online handwritten Chinese character recogni-tion system using a two-dimensional functional re-lationship model[J]. International Journal of Ap-plied Mathematics and Computer Science,2010,20(4):727-738.
[15]才让卓玛,才智杰.现代藏文字构件分解方法[J].青海大学学报:自然科学版,2010,28(4):83-86.
[16]拉巴顿珠,欧珠.现代藏文基字识别的算法设计[J].西藏大学学报:自然科版,2016,31(1):82-88.
[17]Xia Ren,Lian Xiang Ma. The Study of Feature Ex-traction Based on the Optimal Threshold Segmenta-tion Algorithm[J].KeyEngineering Materials,2013,2445(561):652-656.
[2] Ogwu E J,Talib M,Odejobi O A.Text-to-speechprocessing using African language as case study[J].Journal of Discrete Mathematical Sciences andCryptography,2006,9(2):365-382
[3] Romportl J,Matou?ek J. Several Aspects of Ma-chine-Driven Phrasing in Text-to-Speech Systems[J]. The Prague Bulletin of Mathematical Linguis-tics,2011,95(1):51-62.
[4]蔡莲红,张维,胡其炜.文语转换系统中汉语韵律的学习和模拟[J].清华大学学报:自然科学版,1998,38(S1):95-98
[5]倪宏.一个实用的汉语文语转换系统[J].小型微型计算机系统,1995,16(11):42-46.
[6]张家騄,吕士楠,齐士钤,等.汉语文语转换系统的研究[J].信号处理,1989,5(1):1-7.
[7]蔡莲红,魏华武.汉语文-语转换系统的研究与实现[J].应用声学,1994,13(6):1-5.
[8]王维兰.藏文基本字符识别算法研究[J].西北民族学院学报:自然科学版,1999,20(3):20-23,51.
[9]王浩军,赵南元,邓钢轶.藏文识别的预处理[J].计算机工程,2001,27(9):93-96.
[10]王维兰,陈万军.基于笔划特征和MCLRNN模型的联机手写藏文识别[J].计算机工程与应用,2008,44(14):91-93.
[11]周纬,陈良育,曾振柄.基于几何形状分析的藏文字符识别[J].计算机工程与应用,2012,48(18):201-205.
[12]蔡晓娟,黄鹤鸣.基于多投影的脱机手写藏文字符特征提取方法[J].计算机技术与发展,2016,26(3):93-96.
[13]Fernando Perez-Tellez,John Cardiff,Paolo Rosso,David Pinto. Weblog and short text featureextraction and impact on categorisation[J]. Journalof Intelligent&Fuzzy Systems,2014,27(5):2529-2544.
[14]YUN Chang,JIA Lee,Omar Rijal,Syed Bakar. Effi-cient online handwritten Chinese character recogni-tion system using a two-dimensional functional re-lationship model[J]. International Journal of Ap-plied Mathematics and Computer Science,2010,20(4):727-738.
[15]才让卓玛,才智杰.现代藏文字构件分解方法[J].青海大学学报:自然科学版,2010,28(4):83-86.
[16]拉巴顿珠,欧珠.现代藏文基字识别的算法设计[J].西藏大学学报:自然科版,2016,31(1):82-88.
[17]Xia Ren,Lian Xiang Ma. The Study of Feature Ex-traction Based on the Optimal Threshold Segmenta-tion Algorithm[J].KeyEngineering Materials,2013,2445(561):652-656.