欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
提出了一种基于脉冲耦合神经网络（PCNN）和动态时间规整（DTW）的语音识别方法．首先利用改进后的PCNN提取语谱图图像特征作为语音的特征参数，然后通过DTW来进行语音分类识别。实验表明，论文中所提出的方法与传统的LPCC和MFCC方法相比，所需特征参数量减少约40％，并能达到87．5％识别率，利于系统的硬件实现。
英文摘要：
A method for speech recognition based on Pulse-Coupled Neural Network and Dynamic Time Warping is proposed. Improved PCNN was first used to extract the features of spectrogram as the speech features, then them were divided by DTW. The experiment results show that comparing with the traditional ways as LPCC and MFCC, the number of features needed was decreased by about 40% and it could attain the recognition rate of 87.5% with simple hardware realization.
同期刊论文项目
同项目期刊论文
期刊信息
位置：