【作者】 岳玮宁； 王衡； 汪国平；

【Author】 YUE Wei-Ning+, WANG Heng, WANG Guo-Ping1(Department of Computer Science & Technology, Peking University, 100871, China)

【机构】 北京大学计算机科学技术系；

【摘要】 汉字是结构和读音的统一体,且结构变化多样且同音字众多.根据汉字的这些特点,提出了通过语音对笔进行同步辅助校正的非连续汉字手写输入方法.对同时发生的语音输入和笔输入识别后进行整合,使用户以”边说边写”的方式完成输入.本文首先从理论上分析了该方法的可行性,而后对其原理和其中的核心技术,多通道信息整合,进行了深入讨论.在模拟器上进行的实验表明,该方法有效的提高了非连续汉字的输入效率.更多还原

【Abstract】 Pen is the most commonly-used interaction facility on desktop and mobile devicesbeyond traditional keyboard and mouse, and the pen-based Chinese character input is used bymore and more users. However, handwriting by pen in a mobile environment faces a greatchallenge because the structural relationships of strokes are often error-prone in mobile input,especially when a character contains many strokes. Speech is the other important interactionmodality of the handheld mobile devices. Being the most natural and efficient facility in ourdaily lives, speech is the most potential modality on the handheld mobile devices. But speechrecognition technique, especially the Chinese speech recognition, is not very reliable atpresent. Besides, most Chinese characters have many homophones. Therefore, when a userinputs a character by speech, s/he often has to select the result from homophonic candidates,which is a time consuming process. In this paper, speech and pen are combined to address thisproblem. Users are permitted to utilize the features of pronunciation and strokessynchronously to input each character. In other words, the user can speak the pronunciation ofa character while writing its first few strokes at the same time. By checking 1860 mostcommonly-used Chinese characters, we find that for each pronunciation, the number ofhomophones which have the same first three strokes is small than three. It means that the usercan select what s/he wants more easily. Experiment shows that it does achieve theimprovement of input speed.更多还原