摘要
美尔倒谱系数(MFCC)基于人耳的听觉特性,在语音识别实际应用中可取得较高的识别率.介绍了MFCC的基本原理以及提取MFCC的过程和方法,给出了一种硬件实现模型,并详细描述了各个子模块的设计方法.以VerilogHDL作为设计语言,在ISE、Modelsim开发工具下完成综合与仿真,最后在Xlinx Spartan-3e FPGA目标板上实现设计.
MFCC（Mel frequency cepstral coefficient） is based on the human ears characteristic and perform a high recognition rate in practical speech recognition.This paper introduces principle of MFCC,process and method of extracting MFCC.The hardware implementation model is proposed and the design of each sub-model in detail is described.The design is described with Verilog HDL,synthesis and simulation is completed by ISE and Modelsim development tools.Finally,the design has been implemented in Xilinx Spartan-3e FPGA board.
出处
《沈阳工程学院学报（自然科学版）》
2009年第4期370-372,共3页
Journal of Shenyang Institute of Engineering:Natural Science
