CAJ | 학술논문

语音识别是人机语音通信的关键技术之一,也是难题之一.介绍了一种语音识别系统,主要介绍了该系统的语音处理流程,阐述了系统使用Mel频标倒谱参数作为特征提取的方法,采用隐马尔科夫模型算法的测度估计技术.通过严格测试,该系统达到实用化要求.该语音识别系统较好的实现了在移动电子设备上资源有限条件下方便快捷的汉字语音输入,具有重大现实意义.
어음식별시인궤어음통신적관건기술지일,야시난제지일.개소료일충어음식별계통,주요개소료해계통적어음처리류정,천술료계통사용Mel빈표도보삼수작위특정제취적방법,채용은마이과부모형산법적측도고계기술.통과엄격측시,해계통체도실용화요구.해어음식별계통교호적실현료재이동전자설비상자원유한조건하방편쾌첩적한자어음수입,구유중대현실의의.
Speech recognition is one of the key technologies in human?machine speech communication. A speech recognition system and its voise processing program flow are introduced. The way that Mel frequency cepstrum parameters are used to extract features,and the measurement estimation algorithm based on hidden markov model(HMM)are elaborated. The strict testing results show that the system can meet the requirement of practicability. The speech recognition system realized Chinese characters voice input under the condition of resource limited in electrommunication equipments. This system has great practical significance.