Research And Implementation Of Speech Recognition System Based On HMM

Before to the invention of the computer and other equipment, there is already the existence of speech recognition technology, however, early voice decoder based, it is considered to be the first example of speech recognition technology. However, in recent years, speech recognition technology has also been the development of lightning. Speech recognition technology makes intelligent processing machine to identify the human voice of a new high-tech. At present, to achieve a speech recognition system, there are many existing core technology models, such as: artificial neural network, HMM model, dynamic time planning algorithm, etc..First implementation based on Hidden Markov model(HMM Speech recognition system is Doctor Lee. After this, because of the hidden Markov model in the realization of the speech recognition system has many advantages. For example: mature algorithm, convenient realization, high efficiency, the effect is good. Therefore, in particular, the future of speech recognition technology is mostly based on the HMM model as the framework for the realization of speech recognition system.The realization of speech recognition system the reason is based on Hidden Markov model, because of the hidden Markov model has many advantages, so at the end, the paper realize the HMM based speech recognition system model based on. This system can be roughly divided into three stages.The original voice signal processing stage, this is the basic stage of speech recognition system, of speech signal analysis, convert it to the speech feature parameters, so it can be used in hidden Markov model; second, training phase, the signal training, reference template is established; third, recognition stage, will be the detection of signal processing, and signal in a reference template matching look at success or not. The second and third stage is the focus of speech recognition, speech signal training and recognition, is based on Cin Markoff model to realize the speech recognition system.HMM method is used in the design and implementation of speech recognition system. In this paper, the design and the realization of speech recognition system can accurately identify the different phonetic structure and different environment speech design ability, the design and Realization of speech recognition system and the results are satisfactory, the greater commercial value and broader social application value.