基于微型生物雷达的人体语音信号探测及增强关键技术研究

Speech, as one of the important physiological signal of the human body, is the most important and most effective means for human beings to communicate. However, at present, the acquisition of speech signals is almost limited to a single method of collecting by means of microphones. This method of air conduction has the defects of poor directionality and vulnerability to other acoustic signal interference, so its application is obviously limited. On the basis of the pre-study, a new method of non-air conduction speech detection based on microbiological radar is proposed. Because electromagnetic waves are used as media to detect speech, this new speech detection technology can break through the limitations of traditional speech detection methods and expand new ways for humans to obtain speech and acoustic information. However, this new method of speech detection also has problems such as strong electromagnetic noise and poor speech quality. Therefore, this project proposes to use filtering and weighting control strategy to suppress electromagnetic multi-path interference and motion interference in the process of radar voice detection. At the same time, in view of the characteristics of radar speech, such as strong additional electromagnetic noise and missing spectral component, this paper studies the new radar speech denoising enhancement and its missing component compensation method, and extracts and analyzes the acoustic parameters of fundamental frequency, fundamental frequency perturbation and other important values of enhanced speech. The results will provide a new technical method for human speech detection and further improve the practicability of biological radar.

语音作为人体重要的生理信号之一，是人类进行交流最重要和最有效的手段，然而目前对语音信号的获取几乎仅限于借助麦克风采集的单一方式，这种空气传导方式存在方向性差、易受其它声学信号干扰等缺陷，因此其应用场合受到明显限制。本项目在前期预研基础上，提出一种基于微型生物雷达的非空气传导语音探测新技术。由于采用电磁波为介质来探测语音，这种新的语音探测技术能够突破传统语音检测方法的局限，并拓展人类获取语音及声学信息的新途径。然而这种新的语音探测方法也存在电磁噪声强，语音质量差等问题。因此，本项目提出采用滤波加重控制策略抑制雷达语音探测过程中的电磁波多径干扰及运动干扰；同时，针对雷达语音附加电磁噪声强、频谱分量缺失等特点，研究新型雷达语音去噪增强及其缺失分量补偿方法，并对增强语音的基频、基频微扰等有重要价值的声学参数进行提取和分析。研究结果将为人体语音检测提供新的技术方法，进一步提升生物雷达的实用性。

语音作为人体重要的生理信号之一，是人类进行交流最重要和最有效的手段，然而目前对语音信号的获取几乎仅限于借助麦克风采集的单一方式，这种空气传导方式存在方向性差、易受其它声学信号干扰等缺陷，因此其应用场合受到明显限制。本项目在前期预研基础上，提出一种基于微型生物雷达的非空气传导语音探测新技术。由于采用电磁波为介质来探测语音，这种新的语音探测技术能够突破传统语音检测方法的局限，并拓展人类获取语音及声学信息的新途径。然而这种新的语音探测方法也存在电磁噪声强，语音质量差等问题。因此，本项目提出采用滤波加重控制策略抑制雷达语音探测过程中的电磁波多径干扰及运动干扰；同时，针对雷达语音附加电磁噪声强、频谱分量缺失等特点，研究新型雷达语音去噪增强及其缺失分量补偿方法，并对增强语音的基频、基频微扰等有重要价值的声学参数进行提取和分析。研究结果将为人体语音检测提供新的技术方法，进一步提升生物雷达的实用性。