The construction industry is often affected by unanticipated struck-by accidents, which often cause severe injuries and fatalities to the workers. Therefore, monitoring and tracking struck-by hazards in terms of the spatial relationship between a worker and a heavy vehicle is crucial to prevent such accidents. Current studies focus on using active sensors and implementing computer vision but not on the audibility of their safety signals. To address this issue, this paper utilizes sound, a ubiquitous data source present in every construction site, to track and separate equipment sound into different types and determine the direction of arrival (DOA) using the Open embeddeD Audition System (ODAS) framework. Each circular array performs DOA estimation independently using commercial software on two equipment sound sources, bulldozer (mobile) sound and hammer (stationary) sound. The DOAs are fed to a relational database, pre-processed, and used to perform the source tracking. This process provides a step towards monitoring the spatial relationship between workers and equipment with few labels of source location for calibration. The results of our study showed that this method was effective in identifying activities of multiple pieces of equipment in real-time in construction sites without the need for separating sound signals in advance. Future studies can focus on triangulating the exact location of the sound source with less computation power and monitoring how this helps improve workers' awareness of surrounding equipment.