作　　者： (卓嘎）; (董志诚）;

机构地区： 西藏大学工学院,西藏拉萨850000

出　　处： 《现代电子技术》 2017年第18期20-22,共3页

摘　　要： 藏语语音参数提取是藏语语音识别的关键技术之一,参数提取的精确度直接影响语音识别的效果。线性预测系数(LPC)是语音信号的重要频域参数,是目前语音信号处理过程中比较重要的处理技术,广泛应用于语音压缩、语音声学建模、语音合成、语音识别等过程中。首先介绍了线性预测算法原理,然后提出了藏语语音LPC参数提取的方案,最后在Matlab平台上实现了藏语语音LPC参数的提取和仿真分析,研究结果对完善藏语语音合成技术和提高藏语语音识别效率有一定的研究参考价值。 Tibetan speech parameter extraction is the one of the key technologies of Tibetan speech recognition. The accura- cy of parameter extraction directly affects the effect of speech recognition. Linear prediction coefficient （LPC） , as an important frequency domain parameter of speech signals, is currently an important processing technology in speech signal processing and widely used in the process of speech compression, speech acoustic modeling, speech synthesis, speech recognition and so on. In this paper, the principle of linear prediction algorithm is introduced. A scheme of LPC parameter extraction from Tibetan speech signals is proposed. LPC parameter extraction from Tibetan speech signals and simulation analysis are conducted on the Matlab platform. The results of the research provide some references for improving the Tibetan speech synthesis technology and the efficiency of speech recognition.

关 键 词： 藏语语音 参数提取 语音信号