Evaluation of Smartphone-based Sound Level Meters

Abstract

Free, widely-available smartphone-based sound level meters have been utilized to collect large quantities of distributed data in short time periods for the efficient creation of crowd-sourced noise maps. However, the accuracy of these apps can vary greatly as previous studies have shown. In this study, four smartphone-based sound level meters were tested to evaluate their agreement. Four experiments were conducted to test the impact of different apps, operating systems, smartphone hardware, and microphones on app measurements at different sound levels. A combination of four apps, four smartphones, two operating systems, and two microphone types were used in the tests, as well as a hardware-based sound level meter. Errors were evaluated based on two evaluation methods— root mean square error and linearity. The experiment results show that all of the apps produced different readings with respect to the same input stimulus. In other words, each of the apps, operating systems, smartphone hardware, and external microphones influenced the accuracy of smartphone-based sound level meters. Due to the wide variation in measurements, the usage of uncalibrated smartphone-based sound level meters seems to be unacceptable for serious noise assessments. However, the high linearity displayed by some apps indicates the potential for increased accuracy through calibration by professional-grade instruments.