论文部分内容阅读
提出了一种基于统计和规则的混合方法来实现汉语音字转换。利用汉语的语法规则，在统计语言模型中采用了两种基于词和词性的混合语言模型。在实验中，将这两种混合语言模型与基于词的语言模型进行了比较。实验证明，在语言模型中引入词性后，提高了音字转换正确率。考虑了出现拼音错误时的音字转换问题，提出了一种拼音纠错方法来纠正错误。实验证明，当拼音正确率高于８５％时，这种带纠错的音字转换方法可以提高音字转换正确率。
A mixed method based on statistics and rules is proposed to realize Chinese pronunciation conversion. Using Chinese grammar rules, two mixed language models based on words and parts of speech are used in the statistical language model. In the experiment, these two hybrid language models were compared with the word-based language model. Experiments show that after the introduction of part-of-speech in the language model, the correct conversion rate of the pronunciation is improved. Considering the problem of phonetic conversion when phonetic errors occur, a method of phonetic correction is proposed to correct the errors. Experiments show that when the accuracy of pinyin is higher than 85%, this error-correcting pronunciation conversion method can improve the accuracy of pronunciation conversion.