多脉冲激励线性预测编码语音合成器  

Multi-Pulse Excited Linear Predictive Coding Speech Synthesizer

作　　者：景为平[1] 李志钧[2] 

机构地区：[1]南通纺织工学院 [2]南京航空学院电子工程系

出　　处：《南京航空航天大学学报》1991年第S1期99-107,共9页Journal of Nanjing University of Aeronautics & Astronautics

摘　　要：与传统的线性预测编码相比,多脉冲激励线性预测编码(MPLPC)的合成语音质量有显著的改善,能在中等或低速率下合成出高质量的语音。本文介绍了用单片TMS32010组成的MPLPC语音合成器,数码率为9.6kbPS,激励脉冲的幅度和位置用最大互相关搜索算法确定。该系统结构简单,操作简便,通用性强,合成语音清晰、自然易懂,可广泛用于各种系统。文中给出了某些应用实例,经用户试用,效果良好,对合成语音质量感到满意。Comparing with the traditional Linear Predictive Coding (LPC) ,the speech quality of the Multipulse Excited Linear Predictive Coding (MPLPC) can be significantly improved. The MPLPC can produce high quality speech at medium to low bit rates. In this paper, a speech synthesizer is described. The speech synthesizer is made up with TMS32010. The bit rate of the synthesizer is 9. 6kbps. The location and amplitude of the pulse in the multipulse excitation are determined by using the maximum cross-correlation search algorithm. The synthesizer has advantages of simplicity, versatility and operational simplicity. The voice of the synthesizer sounds natural .clear and very easy to understand. It can be widely used in various systems. Here,some application examples of the synthesizer are given. The users are very much satisfied with the quality of the synthesized speech and its work.

关 键 词：语音 语音合成 语音处理 语音编码 线性预测编码 

分 类 号：TN92[电子电信—通信与信息系统] TN919.31[电子电信—信息与通信工程]