That's better than I would have thought, and in some cases on par with very low cost (sub-$50) dedicated SPL meters.
I have a side business that does audiovisual services, and I've been meaning to get the other folks in the crew to test their phone with one of these apps against my SPL meter (which is a ~$80 unit that I believe is pretty accurate in most circumstances), so they'd have a ballpark idea of how loud they're running the gear (there are noise ordinances that we sometimes have to take responsibility for following).
I'd be curious how well a phone could act as an RTA, as well. I used to use my Amiga to do analysis (non-realtime, but reasonably useful in a time when having a high resolution spectrum analyzer was amazing), and standard PCs have been able to act as RTAs for at least 15 years, so I'm absolutely certain the processing power needed is there in modern phones and tablets, but the mic and audio interface would be a limiting factor.
Maybe you could get an idea of the accuracy by e.g. feeding white noise to the phone and seeing how flat the spectrum is? And though I've never used them personally, I hear there are some pretty good clip-on microphones for smartphones on the market.
AS a former Sound Engineer and Studio owner I am blown away at what a phone can do now a days. It really is asy to have someone down load an app and talk them through live sound and EQ with the visual (accurate or not) to show what happens with sound when they play with cheap EQ (AKA cheap EQ is always for REDUCTION and never amplification)
I think if I was doing any sound work anymore I would quickly pick up an external mic and just always use pink noise and the meter to rough in a system.
I've been using SPLnFFT for a while now, with consistent results and I'm happy to see a formal evaluation.
The problem that remains is that there are differences in microphones between iOS devices. SPLnFFT author should probably calibrate the app on every device that is out there on the market. I'm not sure if that's at all feasible.
"NIOSH does not recommend nor endorse a particular commercial product. The data from our study “Evaluation of smartphone sound measurement applications” is available in the journal article referenced at http://scitation.aip.org/content/asa/journal/jasa/135/4/10.1... . The discussion section (page 189) highlights two apps — SoundMeter and SPLnFFT — that had the best accuracy over our testing range. There were two other apps — Noise Hunter and NoiSee — that were within ± 2dB of the reference sound source. Once again, note that the data results in the study do not constitute as NIOSH endorsement of any of these products."
It's not surprising that a device around a mobile device can produce decent results. The Bruel and Kjaer 2250 [0] and 2270 sound meters (probably considered to be the highest-end sound meter on the market) is built around Windows CE.
That said, the JASA paper give this very huge caveat near the end: Furthermore, this study examined these apps in a controlled noise environment. Field measurement results may vary greatly due to the effect of temperature, humidity, long-term use, object interference, and overall stability of the microphone and electronics in these devices.
Furthermore, they tested the devices with relatively high SPL signals (65 to 95 dB), the only tested overall noise levels (no frequency response testing) and in a reverberant field so these results may not be applicable to many other situations (temperature extremes, lower sound levels, etc)
The week point of using a smartphone as a sound meter is going to be the microphone. If you can pair the smartphone with a precision microphone (using something like Studio Six Digitals iAudiointerface [1] for example) you can have the equivalent of a IEC Class 1/ANSI Type 1 microphone (I and colleagues have used this setup for sound measurement projects involving litigation). But if you're depending on the device's internal microphone, be very careful at how you apply those results, especially given the vacates listed in the paper.
That said a smartphone sound meter apps will get you in the right ball park, and as alluded to by SwellJoe they can be at least as accurate (if not more accurate) than some of the $50-$100 Type 2 meters on the market. Just don't rush to any conclusions based solely on data obtained from those setups.
I'd like to have seen a few tests with Windows Phone (somebody must care about us!). My Lumia Icon has 4 "high def" microphones with great claims of accuracy and background noise reduction from Nokia. In my experience they've worked well but I don't have an SPL to compare it against.
Pity it only mentions iOS apps and not the ones I use on Android. I use two apps and they read radically different values (unless I am reading them wrong - that might be the case).
The hardware is at least as important as the software (I'm amazed to see so much scatter between apps), so the homogeneity of the iOS ecosystem is a boon to this sort of measurement. Figure three in their paper shows that the Apple hardware is quite consistent across all models tested.
They did test Android phones and apps; the final paragraph from their results section makes a general statement:
A similar examination of Android apps and devices was not performed because of the low number of apps with similar functionality, and the lack of conformity of fea-
tures between devices. Testing conducted with the four Android smartphones also revealed a high variance in measurements of similar apps between different devices.
Apologies - I did go and take a look at the paper but examined the graphs and scanned the last paragraph but saw iOS everywhere; a more concentrated reading would have stopped me making a fool of myself.
>the homogeneity of the iOS ecosystem is a boon to this sort of measurement.
It was a boon to the researchers because it made their job more simple. Given equivalent or better hardware quality, the same measurements could have been taken without regard to which OEM produced the phone; albeit with more effort on the part of the researchers.
The biggest problem with Android apps is calibration. There's hundreds of models, each with different microphones, available. Even the hardware used on seemingly the same phone models often varies depending on the batch and country.
I have an app (Sound Meter) which apparently downloads calibration data for a number of popular phone models. This is probably the only realistic solution when you're faced with such a wide variety of hardware. I have no idea how accurate the calibration is though.
It's a byproduct of Apples total ecosystem control that developers have a very good idea of what hardware users have. This has upsides and downsides - with other phone ecosystems you lose accurate sound meters, but you gain a wider variety of devices running the same operating system.
> A total of 62 Android apps were examined and downloaded, however, only 4 apps partially met our criteria and were selected for additional testing.
> Android-based apps lacked the features and functionalities found in iOS apps. This is likely due to the iOS advanced audio capabilities compared to other platforms, the open ecosystem of the Android platform, and having so many different Android device manufacturers using different suppliers and components.
I have a side business that does audiovisual services, and I've been meaning to get the other folks in the crew to test their phone with one of these apps against my SPL meter (which is a ~$80 unit that I believe is pretty accurate in most circumstances), so they'd have a ballpark idea of how loud they're running the gear (there are noise ordinances that we sometimes have to take responsibility for following).
I'd be curious how well a phone could act as an RTA, as well. I used to use my Amiga to do analysis (non-realtime, but reasonably useful in a time when having a high resolution spectrum analyzer was amazing), and standard PCs have been able to act as RTAs for at least 15 years, so I'm absolutely certain the processing power needed is there in modern phones and tablets, but the mic and audio interface would be a limiting factor.