基于香橙派的智能语音识别系统的设计  

Design of intelligent speech recognition system based on Orange Pi

作　　者：于镭[1] 林再腾 Yu Lei;Lin Zaiteng(Qingdao University of Science and Technology,Qingdao 266042,China)

机构地区：[1]青岛科技大学自动化学院

出　　处：《电子测量技术》2019年第19期36-40,共5页Electronic Measurement Technology

摘　　要：语音识别技术作为人机交互技术的核心,其发展和应用可以视为人工智能技术发展和应用的前提。同时,随着智能语音技术的商业化,智能语音技术在嵌入式系统层面上的应用也变得尤为重要。在此基础上,旨在香橙派(orange PI)ARM开发板的基础上设计出一款基于Android操作平台下的的智能语音系统,将百度语音提供的开源在线语音SDK与Android相结合,设计出一款能够实现语音识别、语音合成的应用程序,同时将最大期望算法(EM)与高斯混合模型(GMM)相结合,提高语音识别的准确率,最终设计出的智能语音识别系统可以稳定、高效的运行在香橙派开发板上,为智能语音在移动端的应用提出一个更为广泛的应用。As the core of human-computer interaction technology, speech recognition technology can be regarded as the premise of the development and application of artificial intelligence technology.At the same time, with the commercialization of intelligent voice technology, the application of intelligent voice technology at the level of embedded systems has become more and more important.On this basis, based on the orange PI ARM development board, the design of an intelligent voice system based on the Android operating platform combines the open source online voice SDK provided by Baidu Voice with Android. Design an application that enables speech recognition and speech synthesis.At the same time, the maximum expectation algorithm(EM) is combined with the Gaussian mixture model(GMM) to improve the accuracy of speech recognition. Finally, the intelligent speech recognition system can be designed to run stably and efficiently on the orange pie development board for intelligent speech. The application on the mobile side presents a broader application.

关 键 词：香橙派 EM算法 语音识别 ANDROID 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]