Research On Speech Recognition For Background Of Warship Command Module

Speech recognition is to let the machine know what people say,that means the computer should accurately identify the content of the speech in all situations and carry out various intentions according to the information.Without speech,the civilization can not be contained for so long time.Besides,speech is a sign of culture and it distinguish us from animal.It is a very wide range of interdiscipline,including the computer,voice communications,linguistics,besides,it has a close connection with mathematical statistics and artificial intelligence.The process of human society can not be alone.Computer technology offer help to solve more and more complex problems.Meanwhile,communications between man and computer has become much more flexible.The purpose of the research of automatic speech recognition is to enable computer to automatic translate natural languages.Speech recognition can be divided into three areas:1)isolated words recognition;2)continuous speech recognition,that are continuous sentences spoken by the speakers;3)semantic understanding,its aim is not the purpose of translation but the understanding of the meaning of system and making the right feed back to instructions and requirements.Command module is a complex environment,there is lots of noise.Speech recognition is also influenced by the environmental noise,such as engine noise,industrial noise and even the natural noise.Modern ship has already been a national maritime power of the important performance.It plays an important part in the modern sea war.Ships can not survive without a outstanding command recognition performance.Our country owns quite vast territory,and it needs our protect.Sailors communicate by the system of speech in the ship.Only if the commands are conveyed quickly and accurately can they carry out the correct performance.All of these affect the receiving and transmitting of the combat unit's information.How to reduce the noise interference of the command system,improve the identification efficiency of instructions directly influence the execution of ships.In addition to military purposes,in the aspect of civil it also plays an increasingly important role,our daily life is connected with it.So improving the performance of ship instructions recognition system has great realistic significance.In this paper,the main work includes:1)Studying the characteristics of speech signal,identification technology and content,theoretical basis for analyzing speech recognition technology,studying and analyzing the existing research methods,understanding the characteristics of the speech recognition technology of the traditional recognition method.Analyzing ship engine noise environment,then carry out the simulation algorithm under this circumstance.2)Learning support vector machine statistical theory which is based on statistical learning theory.Studying the advantages of its application,combining with the traditional classification algorithm,comprehensive evaluation of its performance,and what's more,further research of support vector machine algorithm principle,observing its speech recognition effect under various kernel functions.3)The research on relevance vector machine algorithm.Introducing relevance vector machine algorithm principle,mathematical model,and voice recognition experiment simulation.Comparing with the support vector machine algorithm about the recognition accuracy and efficiency.Finally analyzing the experimental results.