Research And Implementation Of SAPI Engine Based On Speech Interaction

Speech Recognition is a kind of technology that convert speech information to text or other information which computer can deal with. It appears lots of methods of speech recognition during years of exploration and research in this area, including statistics probability model. The typical method is Hidden Markov Model, artificial neural network.This paper starts with the structure of the speech recognition, discusses method of speech recognition technology based on Hidden Markov Model. Including speech acoustic analyzing, acoustic modeling and recognition strategy.Speech is the most suitable communication way to human. With the development of speech recognition technology, it is being used widely in the human-computer interface and in multimedia application areas. The input and output interface becomes more and more important after the enhancement of computer's calculator speed and storage capacity. The interface of human-computer is being a focus on computer research. Because the speech has the abundant information, it contains mankind's intelligence and it's the most normal way in human's everyday communication. Once the computer has this performance, it can be applied in every activities of human's society, it can change the world. In a word, Speech Recognition is a comfortable interaction between human and machine. It can be combined with other technology to be used in many domains, such as automatic telephone system, synchronous meeting translation system, intelligent multimedia language teaching system. The microsoft's ms-agent is the most popular of all.This paper devised an speech interface of human-computer using ms-agent after research of the speech recognition technology based on Hidden Markov. The interface of human-computer divide the system into two parts, one is speech input, that is hearing, the other is speech output, that is speaking. Those functions are based on speech recognition engine and text-to-speech engine. Those engines make the ms-agent have the function of speaking and hearing and make the PC have great communication capability between the mankind and PC.Speech recognition technology has a bright future, though this technology has lots of work to modify and improve. It has already being used in many areas. To master the approach of the speech recognition development is of benefit to apply this technology.