1·Improve on optimized method of parameter B in the process of HMM training.
对隐马尔可夫模型训练过程中参数B的优化方法进行改进。
2·The HMM based multiple frequency line tracking algorithms of work well under very low SNR, except for a high complexity of computation and for generally unnegligible quantization error.
基于隐马尔可夫模型的多频率线跟踪算法，能在很低的SNR环境下工作，但量化误差较大，和计算量大。
3·A recognition method based on HMM and K-means cluster is proposed through extracting LPC characteristic from acoustic target.
提出一种隐马尔可夫模型和K -均值聚类混合模型的声目标识别方法。
4·At last the experiment of the improved smoothing algorithm and improved HMM between traditional algorithms shows the amelioration of new system.
最后在实验部分对改进的平滑算法以及改进的隐马尔可夫模型做了相对原算法的对比实验。