语音识别技术主要包括特征提取技术、模式匹配准则及模型训练技术三个方面。此外，还涉及到语音识别单元的选取。选择识别单元是语音识别研究的第一步。语音识别单元有单词（句）、音节和音素三种，具体选择哪一种，由具体的研究任务决定。-Speech recognition technology, including feature extraction techniques, pattern-matching criteria and the three aspects of model training techniques. In addition, it involves the selection of speech recognition modules. Choose identification unit is the first step in speech recognition research. Speech recognition module has the word (sentence), three types of syllables and phonemes, the specific selection of which, by the specific research mission decide.