基于语音识别技术的生命探测设备设计与实现

发布时间：2018-04-15 03:05

本文选题：音频生命探测器 + 语音端点检测　；参考：《成都理工大学》2014年硕士论文

【摘要】：本论文来源于四川省芦山地震灾后恢复重建项目——四川省防震减灾社会服务工程。论文在探讨现有的音频生命探测器的基础上，结合音频生命探测器智能化，高效化的发展趋势，浅析语音识别在地震救援中的应用并给出了相应的设计方案。当灾害降临，生灵涂炭，建筑倒塌，交通阻断，通讯中断，以地震为首的自然灾害具有破坏性强，瞬时发生、波及范围广，监测预报困难、社会影响深远等特点，，近几年我国发生的5.12汶川地震和4.20芦山地震给国家经济建设和人民生命财产安全造成重大损失，目前我国在地震应急救援技术方面相对发达国家还较为落后，相关的地震救助技术研究较少。近年来地震灾害依然频繁，地震救援工作仍然受到党中央国务院的高度重视。目前，中国的减灾救助行业仍然需要大量的救援设备的支持，然而高技术救援设备却相对的缺乏自主知识产权。防震减灾作为公共安全的重要组成部分，是关乎民生的公益性事业，是关乎社会稳定造福于民的事业，是关乎人民生命财产安全和社会经济的可持续发展的事业，公共安全领域已经被列入《国家中长期科学和技术发展规划纲要（2006-2020）》的重点领域，其中的优先主题中的重大自然灾害监测与防御明确指出，重点研究开发地震、台风、暴雨、洪水、地质灾害等监测、预警和应急处置关键技术[9]。随着我国经济的不断发展和科技的不断进步，从“十五”国家科技攻关项目到“十一五”预研项目都提出了相关的研究课题，越来越多的专家学者投入到研究行列中也取得了不少研究成果。但许多自主研制的救援设备由于材料选用欠佳，结构考虑不周，导致在实际救援应用中效果并不理想，怎样结合实际救援环境研制出具有一定运用价值的救援设备成为了主导方向。语音信号作为有效的生命体征信号，在救援过程中可以被用来作为有效营救信号，目前在地震救援中音频生命探测仪被广泛的使用，但大部分音频声明探测仪需要通过人来识别信息，救援时需要在废墟中寻找空隙伸入探头，或是在地势相对平坦的地带安放传感器，容易受到现场噪音的影响，影响辨识。 FPGA（现场可编程逻辑阵列）是在CPLD的基础上发展起来的新型高性能可编程逻辑器件[10]，用户可以个性化的配置所需的总线接口，凭借其良好的稳定性与低功耗的特点在工业控制与高速信号处理方面得到了广泛应用，并且借助自身基于IP（Intellectual Property）的SOC设计可以大幅提高研发进度。研究路线：研究并设计一套基于FPGA的集合语音采集、语音识别的装置，该装置通过对周围环境中的音频信号采集，由自适应滤波算法完成语音消噪和增强的功能，再对增强后的语音信号提取Mel频率倒谱系数作为特征向量，再用特征向量对BP神经网络进行训练，最后用训练好的神经网络完成语音识别。
[Abstract]:This paper comes from Sichuan Province Lushan earthquake disaster recovery and reconstruction project-Sichuan Province earthquake disaster reduction social service project.Based on the discussion of the existing audio life detector and the development trend of intelligent and high efficiency audio life detector, this paper analyzes the application of speech recognition in earthquake rescue and gives the corresponding design scheme.When disasters come, life is ruined, buildings collapse, traffic is blocked, communications are interrupted, natural disasters headed by earthquakes are characterized by strong destruction, instantaneous occurrence, wide spread, difficulty in monitoring and forecasting, and far-reaching social impact, etc.The 5.12 Wenchuan earthquake and 4.20 Lushan earthquake occurred in China in recent years have caused great losses to national economic construction and the safety of people's lives and property. At present, our country is still relatively backward in earthquake emergency rescue technology.There are few related seismic rescue techniques.Earthquake disaster is still frequent in recent years, earthquake rescue work is still attached great importance by the State Council of the CPC Central Committee.At present, China's disaster reduction and rescue industry still needs a large number of rescue equipment support, but high-tech rescue equipment is relatively lack of independent intellectual property rights.As an important part of public safety, earthquake prevention and disaster reduction is a public welfare undertaking related to people's livelihood, a cause related to social stability and benefit to the people, and a cause related to the safety of people's lives and property and the sustainable development of social economy.The field of public security has been included in the key areas of the National Medium- and Long-Term Science and Technology Development Program (2006-2020). The monitoring and prevention of major natural disasters in the priority themes clearly indicate that the key research and development of earthquakes, typhoons, rainstorms,Key technologies for flood, geological hazard monitoring, early warning and emergency management [9].With the development of our economy and the progress of science and technology, relevant research topics have been put forward from the National Science and Technology Research Project of the Tenth Five-Year Plan to the 11th Five-Year Plan project.More and more experts and scholars have also made a lot of research achievements.However, due to poor selection of materials and inadequate structural considerations, many self-developed rescue equipment are not effective in practical rescue applications.How to combine the actual rescue environment to develop a certain value of rescue equipment has become the leading direction.As an effective vital sign signal, voice signal can be used as an effective rescue signal in the rescue process. At present, audio life detector is widely used in earthquake rescue.However, most audio declaration detectors need to identify information by people, search for gaps in the debris to reach the probe, or install sensors in relatively flat terrain, which are vulnerable to the impact of field noise and affect identification.FPGA (Field Programmable Logic Array) is a new type of high performance programmable logic device based on CPLD.Because of its good stability and low power consumption, it has been widely used in industrial control and high-speed signal processing, and its own SOC design based on IP(Intellectual property can greatly improve the progress of research and development.Research route: research and design a set of FPGA based set of voice acquisition, speech recognition device, the device through the surrounding environment of audio signal acquisition, by the adaptive filtering algorithm to complete the speech noise reduction and enhancement function,The frequency cepstrum coefficient of Mel is extracted from the enhanced speech signal as the feature vector, and the BP neural network is trained with the feature vector. Finally, the speech recognition is completed with the trained neural network.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.34

【参考文献】