The Improvement of HMM Algorithm using wavelet de-noising in speech recognition

This paper proposes a multi-dimensional time series data mining model for the meteorological data, In this model the dimensions redundant reduction algorithm is used for reducing the redundant dimensions and the complexity of data mining, the extremum slope piecewise linear fitting method is used to implement multi-dimensional meteorological time series segmentation, data compression and feature value extraction, reduce the difficulty of data mining, then use k-means cluster to make the symbols of sequence; final rule extraction is used for getting useful rules in experiments. The results of experiment show that this model has a great practicability.