Study Of Speech Recognition Algorithm Based On ANN And Realization Of Android-based Speech Assistant Software

With the intelligentialize of portable devices, mobile phone has become an indispensable partner to our life. Among all the software stacks for mobile devices,Android Operating System is the world’s best-selling smart phone platform because of its powerful function and convenience. Android is also one of the best choices for developing mobile applications with wonderful system performance, the humanized UI design and the Android Open Source Project(AOSP).The technique of speech recognition has important theoretical significance and broad practical value. It has been paid more and more attention and made a great progress in resent years. The research of Artificial Neural Network, especially Deep Neural Network, has made speech recognition more accurate and more effective. In these circumstances, the technique of speech recognition is commercially available and plays an important role in improving people’s life and our society. In practical applications at present, most of the Artificial Neural Network models are using the BP net or its change form, as BP net is the essence of Artificial Neural Network.Therefore, a mobile application which can directly understand what we say and do what we wish to do without manual operation will be one of the most important trends in software development. By the speech interface, a speech assistant software can realize or replace some functions that people used to do with hands on telephone. A typical application is speech input software, it improves input efficiency, liberate the hands and satisfy the demands of input in some specific environment. This paper makes an mobile application which can dial telephones, answer incoming calls, send messages and receive messages through listening people’s speech commands. It hopes to solve the problems that mobile phone can not be used in driving or other no-hand situations in some extent.This paper first briefly introduces the subject background, including the history of speech recognition and its current development status in China and abroad. After that this paper discusses the basic structure and algorithm principle of Artificial Neural Network in detail, and developing an algorithm to recognize isolated words based on the BP net. Then in accordance with Software Engineering, the requirement analysis and design scheme of the assistant software is discussed.Recording by phone and Bluetooth Headset, restoring files, Bluetooth communication in Android, service and client interface, searching and improving the database and operation in telephone and messages is included. At last, I finish the software combined with Bluetooth Headset and my algorithm on the basis of Android.