语音技术在矿山运输车辆中的应用研究

Research and Application of Voice Technology in Mine Transport Vehicles

在露天矿山运输车辆智慧化过程中,为使司机的工作更加便捷和安全,文章提出一种将语音技术应用于矿山运输车辆司机室及地面指挥中心的方案。其包含基于阵列麦克风语音降噪、语音唤醒、语音识别和语音合成等功能;并依托车辆显示器平台及地面服务器平台,开发了语音交互组件、语音播报组件,实现了矿卡车辆驾驶员关键词识别和危险情况紧急播报。实验室及矿山现场测试结果显示,采用该方案后,驾驶室回声抑制前后满足20 dB的抑制效果,声谱图底噪可控制在200~300 sample内,满足后续对音频质量的识别需求;在正常车载环境下,拾音距离为80 cm,模型识别准确率大于85%,实时率小于500 ms,验证了其语音降噪、语音识别和语音合成功能及性能均满足矿山运输车辆技术要求,表明语音技术的应用提升了司机工作中与车辆设备交互的便捷性及危险状况预警的及时性。

In order to develop more intelligent transport vehicles for surface mines to enable the drivers work more efficiently and safely,this paper puts forward a solution of applying voice technology to the cab of mine transport vehicle and ground command center.This approach encompasses an array of functions,including noise reduction based on array microphone pickup,voice wakeup,speech recognition,and speech synthesis.Relying on the on-board display platform and ground server platform,two modules were developed respectively for voice interaction and voice broadcasting,to enable keyword recognition and emergency broadcasting of dangerous situations.According to the results of laboratory test and field test at mines,the application of this solution managed to meet the echo suppression effect of 20 dB reduction in the cab,and control the bottom noise of spectrograms within 200-300 sample,meeting the subsequent demand for audio quality recognition;the model demonstrated a recognition accuracy greater than 85%and a real-time rate less than 500 ms,in the normal on-board environment and at a pickup distance of 80 cm.The results verifies that the functions and performances of the voice noise reduction,speech recognition and speech synthesis meet the technical requirements for mine transport vehicles,and shows that the application of voice technology improves the convenience of interaction between drivers and vehicle equipment and the timeliness of warning in dangerous situations.