论文部分内容阅读
为了实现高速语音特征参数的提取,在分析了美尔频率倒谱特征参数提取算法的基础上,提出了算法的硬件设计方案,介绍了各模块的设计原理。该方案增加了语音激活检测功能,可对语音信号中的噪音帧进行检测,提高了特征参数的可靠性。最后将设计实现于Stratix II系列FPGA上,仿真结果表明设计可以实现高速、高精度的MFCC特征参数提取。
In order to extract high-speed speech feature parameters, an algorithm of hardware design is proposed based on the analysis of the frequency-cepstrum feature extraction algorithm. The design principles of each module are introduced. The scheme adds the voice activation detection function, which can detect the noise frame in the voice signal and improve the reliability of the characteristic parameters. Finally, the design is implemented on Stratix II series FPGAs. The simulation results show that the design can extract high-speed, high-precision MFCC feature parameters.