嵌入式语音识别系统的设计与实现

介绍一种基于FPGA的嵌入式语音识别系统设计与实现，系统采用线性预测倒谱系数（LPCC）算法和动态时间规整（DTw）算法，该系统的核心部件采用目前流行的Virtex—Ⅱ Pro系列FPGA芯片，使用的工具为业界领先的嵌入式设计套件Xilinx ISE Design Suite10．1，并且运用现代电子系统软硬件协同设计、协同验证和协同工作的方法完成设计。通过实验表明，对于小词汇量、特定人识别系统，具有很好的识别效果，识别准确率达到95．2％以上。该系统的识别性能可以满足基本的嵌入式设备需要，具有广阔的应用价值。

Design and Implementation of Embedded Speech Recognition System

The design and implementation of FPGA - based embedded speech recognition system are introduced, the system uses Linear Prediction Cepstral Coefficients （LPCC） algorithm and Dynamic Time Warping （DTW） algorithm,the key components of the system use current popular Virtex - II Pro Series FPGA chip,and use the industry - leading embedded design kit Xilinx ISE Design Suite 10.1, apply modern electronic systems＇ hardware and software co- design and co- verification to complete the design. Experiments show that for a small vocabulary and a specific recognition system,it has very good recognition results,recognition accuracy rate is above 95. 2%. Recognition performance of this system can meet the basic needs of embedded devices, it has broad application value.