论文部分内容阅读
维吾尔语属阿尔泰语系突厥语族,由于其构词法的特点,八个元音的声频特性在语音识别中,尤其是识别基元选取中有重要作用,其共振峰频率参数也是语音识别和语音合成的重要依据。运用实验语音学的基本理论和方法,在维吾尔语综合语音数据库的办公环境语料条件下,对维吾尔语八个元音进行了声频特性统计分析,给出了维吾尔语元音共振峰频率参数和分布规律,并通过八个元音的语音识别实验结果,验证了其共振峰频率分布规律的正确性。实验证明:维吾尔语在排除元音和谐情况下,其声频特性具有很强的可区分性,对于实现语音信息的传送接受正确性很高。
Uyghur language belongs to the Turkic family of Altaic languages. Due to the characteristics of its word formation, the audio features of eight vowels play an important role in speech recognition, especially in the selection of recognition primitives. The formant frequency parameters are also important for speech recognition and speech synthesis in accordance with. Using the basic theories and methods of experimental phonetics, the author analyzes the audio features of eight Uyghur vowels under the corpora of office environment of Uyghur Integrated Speech Database, and gives the frequency parameters and distribution of Uyghur vowel formant The correctness of the frequency distribution of formant was verified by the experimental results of speech recognition of eight vowels. The experiment proves that the Uyghur language can eliminate the vowel harmony and its audio characteristics are highly distinguishable, and the correctness of the transmission of the received voice information is high.