基于Kinect多生物识别技术的智能同声音译翻译器交互系统研究  

Research on an Intelligent Same Voice Interpreter Interaction System Based on Kinect Multi biometric Recognition Technology

作　　者：王萨 李田 WANG Sa;LI Tian(Sichuan University of Science and Technology,Meishan Sichuan 620500,China)

机构地区：[1]四川科技职业学院,四川眉山620500

出　　处：《自动化与仪器仪表》2025年第2期219-222,共4页Automation & Instrumentation

基　　金：《以“文化自信”引领高职院校大学英语课程思政教学—用英语讲中国(川蜀)故事》;四川省民办教育协会《课程思政视域下民办高职院校大学语文教学改革探索》(NH2023032)。

摘　　要：为解决传统英语同声音译翻译器在语音信号、外界噪音等多种因素影响下识别精度问题,提升语音交互性及翻译质量。研究提出基于多Kinect多生物识别技术的交互系统。将语音识别特征与唇型特征相结合,并借助语音增强算法、交互识别算法和Kinect传感器搭建翻译器交互系统。结果表明,研究提出的算法能有效减少噪声环境干扰,合成语音波形曲线衔接流畅,翻译平均准确率超过96%,语境翻译质量明显优于其他比较模型。研究提出的交互系统能显著提高英语翻译器的翻译效果,可为英语语音交互实现提供可行参考思路。To solve the recognition accuracy problem of traditional English simultaneous transliteration translators under various factors such as speech signals and external noise,and to improve speech interactivity and translation quality.Research proposes an in-teractive system based on multi Kinect and multi biometric recognition technology.Combine speech recognition features with lip shape features,and build a translator interaction system using speech enhancement algorithms,interaction recognition algorithms,and Ki-nect sensors.The results show that the proposed algorithm can effectively reduce noise environment interference,synthesize speech waveform curves smoothly,achieve an average translation accuracy of over 96%,and the contextual translation quality is significantly better than other comparative models.The interaction system proposed in the study can significantly improve the translation perform-ance of English translators and provide feasible reference ideas for the implementation of English voice interaction.

关 键 词：Kinect传感器 语音识别 英语 同声音译 翻译器 交互系统 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]