摘要： 分析语音信号声道特征参数提取问题，针对自相关法的缺陷，提出声道特征参数提取的改进算法。介绍其运算步骤和流程，考虑FPGA适于短期开发及高速性的优点，设计Finite State Machine来控制复杂运算操作及对寄存器的频繁访问。利用Cyclone EP1C6 FPGA实现语音信号声道特征参数提取算法。

关键词: 语音信号, 声道模型, 特征参数, 协方差斜格法

Abstract: This paper analyzes the problem of sound track characteristic parameters’ extraction of speech signal. According to the disadvantage of self-correlation, an improved algorithm of the sound track characteristic parameters’ extraction of speech signal is presented to introduce its operational procession. In view of the high speed and suit for short term exploitation of FPGA, it designes the Finite State Machine so as to control the complicated operations and frequent visit to the registers for the FPGA design principles. The algorithm is implemented successfully into the single Altera Cyclone EP1C6 FPGA to fulfill the algorithm of the sound track characteristic parameters’ extraction of speech signal.

Key words: speech signal, sound track model, characteristic parameters, covariance lattice method

中图分类号: 

• TN912