Development of an intelligent agent for speech recognition and translation

Abstract

Nowadays, speech researchers and software engineer make an effort to integrate speech function into internet applications. Intelligent Agent for Speech Recognition and Translation prototype has been build to achieve that purposed. This prototype also proposed a new framework for speech recognition, word translation and mobile agent technology. Intelligent Agent for Speech Recognition and Translation is a system that recognized human voice in spoken English, then translate English word to Malay word. Generally, this prototype is a merge of two areas, speech recognition and word translation and using mobile agent technology to integrate them. This prototype is a system executed in Local Area Network (LAN) environments and also called Distributed Speech Recognition (DSR) system. DSR system divided into two parts, front-end processing (feature extractions) and back-end processing (recognition). There is a server which connected with a number of client devices. At the client side, the system will received human voice in spoken English and front-end processing will be done here. Then the feature files will be transmits to the server side which executed the back-end processing. When the voice has been recognized, the result will be translated into Malay word in text. After translation process is done, the Malay word in text will be send back to the client computer. All these processes will be handled and monitor by agent. With contribution from mobile agent technology, the efficiency of speech recognition server can be improved compared with traditional distributed speech recognition server in term of time processing. By using mobile agent technology, speech recognition server can perform their task as parallel processing. Beside that, this approach also can reduce network traffic while handling the transfer data.