The Study And Implement Of Speech Recognition Based On LonWorks

Speech Recognition Technology is related to Signal Processing, Pattern Recognition, Probability and Information Theory, Speech Subject and Artificial Intelligence and so on. Nowadays with the study of Speech Recognition Theory and the development of Digital Signal Processing's technology, Speech Recognition Technology is gradually having the abilities of walking out of lab and serving for society. Speaker Recognition is one kind of Speech Recognition. It can be divided into Speaker Verification and Speaker Identification. Different from Speech Recognition, Speaker Recognition systems choose the parameter or parameters which had great difference between speakers and little difference about itself.Later in 80's, Echelon designed LonWorks technology for Control Network's universal standard. It completes tasks using LonTalk protocol. Now LonWorks has become the universal standard of Control Network. Considering practicality and standard disposal, Echelon designed Neuron Chip. Neuron Chip supports the former six layers of ISO/OSI model. So users only need support the APP's programme and configuration. This makes the implement of protocol standard, also makes easier for development and configuration. In some complex applications, the,speed of CPU and the maximum memory can't catch up with the LonWorks nodes' demand. In order to meet these applications, some kinds of Neuron Chips designed with High Speed parallel interface. So any kind of MCU can use it. Furthermore LonTalk protocol can be transplanted in any other MCU.This article mainly talks about the design and implement of Speaker Recognition based on LonWorks. Many intelligent nodes compose the LonWorks network through twisted-pair. Each node is a Speaker Recognition system. We chose Motorola's MC143120DW as our Neuron Chip. There are three CPUs inside it. One of them is used to deal with user's application routine, other two are for network's communication. MC143120DW uses Node Builder as it's software develop environment, this environment integrates Neuron C's edit compile link and debug function. User can easily develop many kinds of application system. Each node's Speech Processing chip is TFs TMS320VC5402, character parameter is Fundamental cycle Linear Predict Cepstrum Coefficient and the combination of them. Under the control of the keyboard linked with neuron chip, we test the parameters with dynamic time warping. The result is displayed by LED linked with neuron chip.