本文介绍的聋哑人视觉辅助语音训练系统，采用对聋哑人所发语音音素进行识别，并通过显示其发音近似率以及待发标准语音和相似语音的口形图来纠正受训者的发音。与国外现有仅显示语音特征的系统相比，受训者对显示的信息易于理解，因此效果较好。 This dissertation presents a visual aided speech training system for deaf-mute who uses the phoneme of the deaf-mute to identify the phoneme and corrects the pronunciation of the trainees by displaying their phonetic similarities and the oral graphs of the standard pronunciation and similar speech. Compared with the existing systems that only display voice features abroad, the trainees can easily understand the displayed information and thus perform better.