中文说明：DTW是较早的一种模式匹配和模型训练技术，它应用动态规划的思想成功解决了语音信号特征参数序列比较时时长不等的难题，在孤立词语音识别中获得了良好性能。虽然HMM模型和ANN在连续语音大词汇量语音识别系统优于DTW，但由于DTW算法计算量较少、无需前期的长期训练，也很容易将DTW算法移植到单片机、DSP上实现语音识别且能满足实时性要求，故其在孤立词语音识别系统中仍然得到了广泛的应用。
English Description：
DTW is an earlier technology of pattern matching and model training. It successfully solves the problem that the sequence of speech signal characteristic parameters is different in time and length by using the idea of dynamic programming, and achieves good performance in isolated word speech recognition. Although HMM model and ANN are better than DTW in continuous speech recognition system with large vocabulary, DTW algorithm is still widely used in isolated speech recognition system because it has less computation and does not need long-term training, and it is easy to transplant DTW algorithm to MCU and DSP to realize speech recognition and can meet the real-time requirements.