Abstract: Based on the biomimetic pattern recognition theory，this paper proposes an improved algorithm for gender identification.Firstly，the pitch of each training data will be extracted as the feature we used to train and test samples.Then，cover the high dimensional space by some neurons，according to the distances between training samples of each kind of patterns.Experiments show that the biomimetic pattern recognition theory is validity in classification problems.And using only 50 training samples，the result shows better improvement in correct recognition rate.

Key words: biomimetic pattern recognition, gender identification, pitch, high dimensional space

摘要： 基音频率是语音信号中的重要参数，也蕴涵了区分性别的主要特征。在仿生模式识别的理论基础上，提出一种高维空间连续神经元覆盖算法，以基音频率为特征，将该算法应用于语音信号的性别识别。实验表明，仿生模式识别理论在进行划分时取得了很好的效果，只需要50个训练样本就可以得到较高的正确识别效果。

关键词: 仿生模式识别, 性别识别, 基音频率, 高维空间