声控系统 翻译

声控系统 翻译 Voice control system Automatic speech recognition ASR The concept ，the application and the significance of speech recognition system What is a speech recognition system Known as speech recognition, is the computer through an analysis of speech signal processing so as to identify what types of speech;Speech recognition technology is to make the machine through the recognition and understanding of the speech signal into the appropriate text or command of high technology.Generally speaking, is the use of computer simulation of human auditory function, realize the PC system hasreferred to human auditory intelligence. Application and significance of the speech recognition system PC ASR can be achieved through such as mobile phone,SMS, voice dialing, voice navigation, intelligent Home Furnishing control etc..In order to achieve the liberation of hands, achieve a better human-computer interaction, provide convenience for the disabled, so that it can be completed usually can notcomplete the work. Commonly used classification ASR One, speech recognition according to the voice narration can be divided into isolated word recognition, connected word recognition and continuous speech recognition. 1 isolated word recognition refers to the speaker only saida word, a phrase or a command. 2 connected word recognition generally refers to the ten digital multi connected digit recognition and become,generally used for telephone and database query. 3 continuous speech recognition is used for speaker recognition to the daily natural way; Obviously, continuous speech recognition have been much more difficult, because it exists coarticulation phonetic andword segmentation, search and other issues, in addition tothe acoustic models in speech concerns but also relates to the language model, such as the lexical, grammar etc ASR的分类 Two, from the relationship between the speaker and therecognition system can be the recognition system consists of speech recognition, speaker adaptation and speaker independent speech recognition. 1.Speech recognition systems only consider 1 specific people: to identify special voice, the system itself needs only voice training can be targeted to specific people 2.The speaker independent speech recognition system:voice has nothing to do with the person, usually with a lot ofdifferent speech database to learn recognition system 3.Speaker adaptation speech recognition: between specificand non specific human speech recognition system, the system can adapt to new users. The classification of ASR Three, from the vocabulary size recognition system can bedivided into 3 kinds of recognition system. 1.Small vocabulary speech recognition system.Usually include speech recognition systems of dozens of words; 2.Speech recognition system for mediumvocabulary.Usually includes hundreds of words to the speech recognition systems with thousands of words; 3.Large vocabulary speech recognition system.Usually contain thousands of words to the speech recognition system of tens of thousands of words. With the development of computer and digital signal processor computing power and enhance the accuracy of the recognition system, recognition system based on thevocabulary size classification is also ongoing changes. It isa medium vocabulary recognition system to the future is likely to small vocabulary speech recognition system.These different limit yequedingleyuyinshibiexitong. The realization and the working principle of ASR Training: pre analysis of the speech feature parameters,making speech templates, and stored on the speech parameter library. Recognition: identifying analysis of speech after trainingthe same, get speech parameters. It will be with the libraryreference template one one comparison, and the judgmentmethod to find the nearest feature templates, the recognition results obtained. Distortion: then when compared to a standard, which isbetween the measuring speech feature parameter vector"distortion measure". The main recognition framework: Based on the dynamic time warping method of pattern matching based on Hidden Markov model method and the statistical model. Speech recognition software Windows SAPI SDK Speech recognition and control Speech technology in the software includes two aspects,namely, speech recognition and speech synthesis. These two techniques are needed to the speech engine support.API application programming interface Microsoft launched,although not the industry standard, but more extensive application. SAPI full name The Microsoft Speech API.Relevant SR and SS the engine is located in Speech SDK development kit.The speech recognition engine supports multiple languages and read. SAPIIncludes the following components: (1)Voice commands API.To control the application, generally used for speech recognition system. Identify a specific command, will call the relevant interface is the application to complete the corresponding function. If the program to realize the voice control, must use this set of objects. (2)Voice Dictation API.Dictation, voice recognition interface. (3)Voice Text API.From text to speech, and speech synthesis. (4)Voice Telephone API.Speech recognition and speech synthesis using thetelephone system, can build a telephone answering systemusing this interface, and can control the computer by telephone. (5)Voice Objects API.The pronunciation system encapsulatescomputer. SAPI Is based on the COM ,Microsoft also provides ActiveX control,can also be used for Webpage,VBA even to the extent that EXCEL.