Sound Recognition for Civil Infrastructure Construction
In a dynamic construction site, sound information involving onsite human communications and working/equipment sounds conveys crucial insights into construction progress, work performance, and safety. The reflection of these interactive dependencies implies that a construction project entails the logical interrelationships of a series of construction activities, onsite communications, and work/operation sounds. This research study will investigate a speech and sound recognition framework that can capture and monitor onsite two-way radio communications and work/equipment sounds.
Preliminary Results of Ongoing Studies
Examples in the Figure below demonstrate the preliminary results for classifying the construction work/equipment sounds. The comparative study can be found as listed in Table 1. Dr. Yong-Cheol Lee, has been involved in research projects regarding construction work sound data recognition. His one study about sound recognition of multi-layered construction activities was selected for the 1st Place Best Paper Award at the 2017 International Workshop on Computing in Civil Engineering (IWCCE) (Cho and Lee 2017). In addition, one conference paper regarding a supervised machine learning-based sound identification was accepted for 2018 CRC conference (Zhang et al. 2018) and one journal paper about sound classifiers analyses is under-review.
1st Place Best Paper Award, Sound Recognition Techniques for Multi-layered Construction Activities and Events, International Workshop on Computing in Civil Engineering (IWCCE) 2017
Data Archive for Construction Sound
The data archive named DSpace, which is a digital repository software package, was established for this project. The website contains more than 100 sound libraries and conversation corpus of construction activities and equipment collected in this project.
K-12 STEM EDUCATION AND OUTREACH ACTIVITIES
The goal of this research study has aligned with K-12 STEM education that can allow K-12 students to be motivated and inspired in STEM areas. As shown in the pictures below, the researchers have already conducted the sound recognition study with one 7th grade African-American woman student named Jnea Steiner attending Kenilworth Science and Technology Charter School. She collected diverse construction sounds at the Our Lady of the Lake Regional Medical Center at Baton Rouge and learned the automated sound identification method for competing in the regional science fair.