Scrubbing Complex Sound Sources for Factory Situational Awareness

根据CDC NIOSH网站的说法，“职业性听力损失是最常见的听力损失之一， 美国的职业病”。这部分是由于工人人数 在工作场所暴露于危险噪音水平，不使用听力保护装置 （HPD）。国家卫生研究所的一项调查估计，超过2200万美国工人 每年在工作中暴露于有害的噪音水平。另一项调查显示， 工人报告未使用HPD。在工厂车间不使用HPD通常不是因为 这些设备的成本或缺乏可用性。在许多情况下，可以看到耳塞和其他HPD 挂在工人的脖子上或塞在口袋里！问题是，除了 除了屏蔽有害声音外，HPD还屏蔽环境中的所有其他声音， 在这些行业中，报警器的可靠性对人身安全和工作绩效至关重要。我们提出了一个 一种新颖的听力保护系统，提供了对“情景”声音的增强访问， 如警报器、叉车和声音，同时抑制有害的巨大噪音。通过 克服了情境意识的障碍，我们相信工人们会更愿意穿 包括我们系统的HPD。 我们的解决方案采用创新的信号处理技术， 机器，每个机器都被相邻的麦克风（麦克风）“标记”， 其他的混合物，“情境”的声音放置在更不同的位置，在 工作空间。该擦洗过程从通信设备的响应信号中去除标记的噪声。 情景对话，从而增强了其他重要声音的可听性， 环境与现有的产品不同，这种擦洗与频率无关， 信号的振幅被擦除。 基于成功的结果擦洗噪音从小来源，这个项目的重点是两个 我们目前的系统的局限性：1）声学传感器用作标记， 除了不必要的噪音。这种标签污染使得 从情景麦克风中清除不需要的噪音。我们建议研究替代传感器 减少污染的技术。2)嘈杂的机器通常有多个底层 噪声源（如电机、皮带轮、轴承）和相对较大的外壳， 源成功的清理需要尽可能多的标记噪声源。我们 建议制定和测试一个程序，以确定标签的数量和位置 需要清理一台有多个内部源的机器。

According to the CDC NIOSH website “Occupational hearing loss is one of the most common work-related illnesses in the United States.” This is due, in part, to the number of workers exposed to hazardous noise levels in the workplace who don’t use hearing protection devices (HPDs). A National Health Institute Survey estimated that over 22 million U.S. workers are exposed to hazardous noise levels at work annually. Another survey indicated over half of those workers reported non-use of HPDs. Non-use of HPDs on factory floors is often not due to the cost of such devices or lack of availability. In many cases, ear plugs and other HPDs can be seen hanging around a worker’s neck or stuffed in a pocket! The problem is that, in addition to blocking hazardous sounds, HPDs also block all other sounds in the environment but awareness of alarms is critical for physical safety and job performance in these industries. We propose a novel hearing protection system that provides enhanced access to “situational” sounds such as alarms, forklifts, and voices, while simultaneously suppressing hazardously loud noises. By overcoming the situational-awareness obstacle, we believe workers will be more willing to wear HPDs that include our system. Our solution uses innovative signal processing technologies to “scrub” hazardously loud machines, each of which has been “tagged” by adjacent microphones (mics), from the response mixtures of other, “situational” mics placed in more acoustically diverse locations in the workspace. This scrubbing process removes the tagged noises from the response signals of the situational mics, thereby enhancing the audibility of the other important sounds in the environment. Unlike existing products, this scrubbing is independent of the frequency or amplitude of the signal being scrubbed. Based on successful results scrubbing noise from small sources, this project focuses on two limitations of our current system: 1) Acoustic sensors used as tagging mics pick up situational sounds in addition to the unwanted noise. This tagger contamination makes it more difficult to scrub just unwanted noise from situational mics. We propose to investigate alternative sensor technologies to reduce contamination. 2) Noisy machines typically have multiple underlying noise sources (e.g. motors, pulleys, bearings) and relatively large shells encasing the noise sources. Successful scrubbing requires as many tagging mics as distinct noise sources. We propose to develop and test a procedure to identify the number and location of tagging mics needed to scrub a machine with multiple internal sources.