Accuracy of phone base SPL meters?

[mcdn]

In another thread there was some discussion of matching levels for test tracks, and the topic of how to measure levels came up, so I tried a quick comparison. The phone was balanced on top of the UMIK, so it's mic was only about a centimetre away from the UMIK's capsule:

In a quiet room I get:
REW+UMIK-1+calibration file: 36.8dB(A)
Audiotools iOS app on iPhone 11 pro: 27.5dB(A)
iNVH iOS app on iPhone 11 pro: 17.5dB(A)

And with some pink noise playing:
REW+UMIK-1+calibration file: 80dB(A)
Audiotools iOS app on iPhone 11 pro: 68.8dB(A)
iNVH iOS app on iPhone 11 pro: 57.8dB(A)

Question 1: Is it safe to assume the REW+UMIK-1 is correct?
Question 2: How can these results be so different? They're all A-Weighted, Hann windowed, on a very standardised device. Am I doing something wrong?

 

[tmtomh]

In another thread there was some discussion of matching levels for test tracks, and the topic of how to measure levels came up, so I tried a quick comparison. The phone was balanced on top of the UMIK, so it's mic was only about a centimetre away from the UMIK's capsule:

In a quiet room I get:
REW+UMIK-1+calibration file: 36.8dB(A)
Audiotools iOS app on iPhone 11 pro: 27.5dB(A)
iNVH iOS app on iPhone 11 pro: 17.5dB(A)

And with some pink noise playing:
REW+UMIK-1+calibration file: 80dB(A)
Audiotools iOS app on iPhone 11 pro: 68.8dB(A)
iNVH iOS app on iPhone 11 pro: 57.8dB(A)

Question 1: Is it safe to assume the REW+UMIK-1 is correct?
Question 2: How can these results be so different? They're all A-Weighted, Hann windowed, on a very standardised device. Am I doing something wrong?

Apologies if this a dumb question, but are all three devices using the same weighting scheme (A, B, C, or none)?

 

[mcdn]

Apologies if this a dumb question, but are all three devices using the same weighting scheme (A, B, C, or none)?

All A weighted. REW and iNVH are both definitely 20Hz-20kHz, Hann windowing. Audiotools free version doesn't specify the frequency range or the windowing function, but it does lewt you set the weighting to A.

 

[Blumlein 88]

Nearly all phones have badly drooping response below 200 hz. With pink noise weighted to the lower frequencies your phone won't match the Umik.

And beyond that don't use SPL meters for level matching on test tracks. You can use it for level setting. But if you are matching levels to compare one track or device to another do that matching with a volt meter at the speaker leads. Match readings to within 1%.

 

[RayDunzl]

Question 1: Is it safe to assume the REW+UMIK-1 is correct?

The UMIK-1 via REW matches a pair of cheap dedicated SPL meters I have (close enough anyway).

Question 2: How can these results be so different?

They aren't calibrated devices traceable to a standard.

https://www.bksv.com/en/Knowledge-center/blog/articles/sound/sound-level-meter-calibration

 

[mcdn]

Nearly all phones have badly drooping response below 200 hz. With pink noise weighted to the lower frequencies your phone won't match the Umik.

Interesting, I'll try some other frequency ranges

 

[tmtomh]

I have an old Radio Shack SPL meter I've used for years (not for level-matching, but just for rough levels in the room). More recently I got the DecibelX iOS SPL meter for my phone, I realize neither device should be considered super-reliable, but they consistently match up with each other to within 1dB at all frequencies, so I tend to think they are reasonably accurate.

 

[mcdn]

Limiting the pink noise to a range from 200Hz-10kHz gives:

REW+UMIK-1+calibration file: 73.8dB(A)
Audiotools iOS app on iPhone 11 pro: 67.7dB(A)
iNVH iOS app on iPhone 11 pro: 56.8dB(A)

For Android devices the crazy number of SKUs makes it hard, but the Apple devices have so few SKUs that it should be a breeze for the apps to know exactly the microphone characteristics and compensate.

 

[mcdn]

I have an old Radio Shack SPL meter I've used for years (not for level-matching, but just for rough levels in the room). More recently I got the DecibelX iOS SPL meter for my phone, I realize neither device should be considered super-reliable, but they consistently match up with each other to within 1dB at all frequencies, so I tend to think they are reasonably accurate.

Just tried DecibelX (nice looking app) with 200Hz-10kHz pink noise:
REW+UMIK-1+calibration file: 74.5dB(A)
DecibelX iOS app on iPhone 11 pro: 66.4dB(A)

 

[tmtomh]

Just tried DecibelX (nice looking app) with 200Hz-10kHz pink noise:
REW+UMIK-1+calibration file: 74.5dB(A)
DecibelX iOS app on iPhone 11 pro: 66.4dB(A)

Yikes!

 

[andreasmaaan]

The short answer is (as others have already said) that you can't trust your phone's mic as its neither linear across the audio band nor calibrated.

FWIW, though, this is an interesting set of tests on a number of different phones/SPL meter apps.

 

[snowman]

Ears aren’t linear across the audio band either, nor calibratable.

 

[robwpdx]

The short answer is (as others have already said) that you can't trust your phone's mic as its neither linear across the audio band nor calibrated.

FWIW, though, this is an interesting set of tests on a number of different phones/SPL meter apps.

That is why they brought out an accurate one https://www.cdc.gov/niosh/topics/noise/app.html

 

[snowman]

The study is old, software updates occur.

 

[Joachim Herbert]

Ears aren’t linear across the audio band either, nor calibratable.

What do you mean to say?

 

[staticV3]

Very similar test, with vastly different result:

 

[Ifrit]

That is why they brought out an accurate one https://www.cdc.gov/niosh/topics/noise/app.html

Was it here on ASR somebody posted a link to the AES study trying to figure out the most accurate one? With the verdict of that one being the winner?

 

[robwpdx]

Was it here on ASR somebody posted a link to the AES study trying to figure out the most accurate one? With the verdict of that one being the winner?

I posted this article:

The Accuracy of iPhone Applications to Monitor Environmental Noise Levels - PubMed

NA Laryngoscope, 131:E59-E62, 2021.

pubmed.ncbi.nlm.nih.gov

There are more articles listed on that link.

The app found the most accurate is:

Sound Level Meter App | NIOSH | CDC

The NIOSH Sound Level Meter app combines the best features of professional sound levels meters and noise dosimeters into a simple, easy-to-use package

www.cdc.gov

It was the result of NIOSH having the best software ported to the platform with the most accurate microphone. The history is discussed on that app page.

 

[Ifrit]

Ah, yes, that’s the one. Thank you!
I’ve been using the app since… forever it seems like.