该系统能够识别连续说出的数字串和若干组姓名。建模是针对子词(sub-word, eg. 音

素)，具有一定的可扩充性。当加入一个新名字时，只需修改发音词典和任务语法即可。模

型为连续混合高斯输出，运用语音决策树聚类形成的绑定状态式三音素。-The system is able to recognize continuously spoken digit string and the number of groups name. Modeling is a sub-word (sub-word, eg. Phonemes), with a certain degree of scalability. When adding a new name, simply modify the pronunciation dictionary and task grammar can be. Model triphone continuous mixture Gaussian output the binding state formula, the use of the voice decision tree clustering formation.