摘要
藏语语音在识别过程中会受到各种信号的干扰,从而影响语音识别的质量。在进行语音识别预处理过程中,语音信号的降噪技术是一个关键的环节。文章通过分析各种降噪手段,重点研究了藏语语音信号降噪方法,提出了符合藏语语音信号降噪的小波包降噪算法。其研究结果为进一步的藏语语音识别技术的研究提供了基础。
Pronunciation Tibetan text in the recognition process will be subject to a variety of signal interference, thus affecting the quality of speech recognition. The noise reduction technology of speech signal is a key processing part in the course of speech recognition pretreatment. In the present paper,a noise reduction method for Tibetan speech signal focused is studied by analyzing the various noise reduction methods, and proposed wavelet packet noise reduction algorithm in accordance with noise reduction of Tibetan speech signal. The result of this study can be used in Tibetan speech recognition system and provide a foundation for further study on Tibetan speech recognition technology.
基金
科技部973计划前期研究专项项目"藏文语音识别技术研究"阶段性成果
项目号:2009CB326201
关键词
小波包
降噪
藏语语音
算法
wavelet packet
noise reduction
Tibetan speech
algorithm
