Page 17 - Summer2017

measurements to be made with the built-in microphone after the user has performed his/her own calibration rela- tive to a more accurate sound level meter. It also highlights one of the challenges faced by app developers, which is the need to determine reasonably accurate nominal sensitivities for any smartphone device to be supported by a sound mea- surement app. This presents a more daunting challenge for developers of Android apps because so many smartphone manufacturers offer so many different devices with poten- tially modified versions of the Android operating system.
External Microphones
Although the previous work clearly demonstrated the po- tential for reasonably accurate sound level measurements with smartphone apps (Kardous and Shaw, 2014), at least in a well-behaved environment, NIOSH researchers were in- terested to see what could be accomplished with relatively inexpensive, highly portable, external measurement micro- phones. In their follow-up study, Kardous and Shaw (2016) used similar tests to examine two different external micro- phones that may be connected to the analog headset jack of a smartphone. The study examined the Dayton Audio iMM-6 and the MicW i436. The i436, although more expensive, is claimed by the manufacturer to comply with the IEC 61672- 1 Class 2 specification. It should be noted, however, that the IEC standard applies to the SLM as a complete measurement system, not to the microphone, alone (IEC, 2013). To inves- tigate the relative performance of these microphones, the four iOS apps that were found to be most accurate in the previous study were used once again to measure sound lev- els ranging from 65- to 95-dB SPL in 5-dB increments. Fig- ure 5 offers a look at the test configurations used to compare the performance of various external microphones and apps.
The mean sound level differences for the external micro- phones, relative to a type 1 SLM, were much better than those previously obtained with the built-in microphones of the smartphones. The mean and standard deviation for ex- ternal microphones were −0.023 and 0.530 dB, respectively, in contrast to 1.646 and 3.795 dB, respectively, for internal microphones. Both external microphones performed well in these tests, which suggests that even a very inexpensive microphone, such as the iMM-6, can be used for reasonably accurate measurements with a smartphone. The more robust construction of the i436 as well as its ability to fit a standard 0.25-inch adapter for an acoustic calibrator, may make the i436 more reliable in changing environmental conditions and easier to calibrate, but its higher cost will be justified ac-
Figure 6. Statistical distributions of differences between reference sound level meter (SLM) levels and app measurements with external and internal microphones are represented with box plots. The hori- zontal line inside the box represents the median value. The horizontal lines above and below the box represent maximum and minimum values, respectively. Extremely high or low values may be considered outliers and are represented as dots above or below the box plot. Re- sults are shown by app (top) and by nominal sound level (bottom). Republished from Kardous and Shaw (2016), with permission.
cording to the user’s needs. A visual comparison of the rela- tive performance of the internal and external microphones as well as of the four selected smartphone apps is shown in Figure 6.
In another study, Roberts et al. (2016) also concluded that it may be possible to measure valid occupational noise ex- posure levels with smartphones and other “smart” mobile devices, with suitable apps and external measurement mi- crophones. (See article by Enda Murphy in this issue of Acoustics Today.) The need remains to conduct additional research to better understand the suitability of smartphone SLMs for calibrated measurements in real-world conditions outside the laboratory. Even within the laboratory, there are other issues to investigate, such as frequency dependence, directional response, signal path linearity and distortion, and even user behavior. An investigation by Robinson and Tingay (2014) found that with carefully selected compo-
Summer 2017 | Acoustics Today | 15