2
\$\begingroup\$

I am a computer-science student working on a signal processing project on smartphones based on recorded microphone signals. For my project, I need a really "reliable" recorded signal. The system works perfectly when I put phones on a table and don't touch it. However, when I touch the phone by my hands (e.g., press a button when phone is static in the table), the recorded data in microphone seems changed dynamically for unknown reason. For example, at certain time period, the received audio (sent from my laptop) is louder.

Since I am not an expert of electrical theory, I would like to know if there is any theory about this phenomenon (such as touching make the diaphragm hard to vibrate) I know the microphone is a MEMS microphone which operates as a condenser type microphone but not sure if there is any reason that the recorded signal changes depends on how user touch the microphone (or smartphone body).

\$\endgroup\$
3
  • \$\begingroup\$ Hmm its hard to say but if your phone body is somehow metallic or so to say conducting, and the microphone is electret type the outer plate(cylinder) should be getting a induced or static voltage noise, its still impossible to say any thing until both the type of microphone and phone structure is not described but still it doesnot have anything to do with amplification. \$\endgroup\$
    – MaMba
    Commented Oct 29, 2015 at 15:55
  • 6
    \$\begingroup\$ It can be purely mechanical effect. The body can damp/amplify/change/add some harmonics of the vibrations. \$\endgroup\$
    – Eugene Sh.
    Commented Oct 29, 2015 at 15:57
  • \$\begingroup\$ Hi, thanks for your response!! I suspect it is caused by either mechanical effect or electret as mentioned above but I found there is no detailed tutorial to discuss this phenomenon in microphone design (is it not a serious problem to microphone?). I test it on three different type of phone but all get the same result (iPhone 6s, Samsung S5, Samsung Note4) Few of my friends told me it is impossible for mechanical effect because sound decay too fast in the phone body. However, again, I am unable to find any relevant materials talking about this ... :( \$\endgroup\$
    – Yu-Chih Tung
    Commented Oct 29, 2015 at 16:49

2 Answers 2

Reset to default
5
\$\begingroup\$

I've flagged this question to be moved to a more appropriate SE site, but (to expand on what Eugene Sh. said) the answer can be deduced from the advice given to pro mic users:

Vibrations will not only reach the microphone through the mic stand. They can also enter the microphone body through the XLR connector and the cable, particularly if it is quite stiff. The best way to overcome this is to use the lightest and most flexible cable you can, and to clamp or tape it firmly to the stand to stop vibrations being carried into the mic. Location sound recordists for film and television use very light flexible tails between the mic and pole, and then a more robust cable back to the mixer, simply to avoid vibrations coming from this means of entry. Vocalists using a hand-held directional microphone can isolate cable vibration by making a loop of cable and trapping it between the fingers (not touching the microphone body) so that vibration along the cable is stopped when it reaches the hand, the onward loop to the mic being (hopefully) vibration-free. This technique also avoids straining the XLR connector with the weight of cable as the vocalist moves about the stage.

There are couple more things you need to know about MEMS:

  • Their natural frequency response is usually quite far from flat. So there's [mandatory] electronic post-processing. See this comparison of a MEMS mic and a [conventional/large] electret frequency response from a NASA study

enter image description here

  • MEMS often get advertised as omidirectional, but that's only true of the sensor itself. Once it goes into whatever body (especially something as bulky as a phone), they stop being all that omnidirectional:

enter image description here

\$\endgroup\$
6
  • \$\begingroup\$ None of this explains why the amplitude of subsequent sounds went up after the disturbance. It is common for some kind of disturbance to be loud enough to cause automatic gain circuitry to turn the gain down temporarily, but the OP is reporting the opposite affect. \$\endgroup\$
    – Olin Lathrop
    Commented Oct 29, 2015 at 16:47
  • \$\begingroup\$ @Olin Lathrop: It's indeed possible there's more to it than this. MEMS often see heavy post-processing. (I wasn't the one who downvoted you, by the way.) \$\endgroup\$
    – got trolled too much this week
    Commented Oct 29, 2015 at 16:58
  • \$\begingroup\$ @Olin Lathrop:I also feel strange that sometimes the volume goes up after the disturbance. If it is caused by mechanical damping, I suspect it will go low, right? (e.g., making the Diaphragm harder to vibrate). \$\endgroup\$
    – Yu-Chih Tung
    Commented Oct 29, 2015 at 17:02
  • \$\begingroup\$ @ Respawned Fluff: Hi, could you explain more about the post-processing at MEMS? I read some article about the principle of MEMS microphone, but not sure what is the post-processing you are suggesting ^^ \$\endgroup\$
    – Yu-Chih Tung
    Commented Oct 29, 2015 at 17:04
  • \$\begingroup\$ @Yu-ChihTung: I'm honestly not really familiar with the details, but from the frequency response alone [see graph added] you can see that something has to be done at least to remove that 1000% spike at a higher [but still audible] frequencies. \$\endgroup\$
    – got trolled too much this week
    Commented Oct 29, 2015 at 17:08
-1
\$\begingroup\$

There is nothing like this inherent to the microphone. This is apparently something the software in the phone is doing.

\$\endgroup\$
1
  • \$\begingroup\$ Hi, I think it is not caused by software things since I have remove the gain controller in software and my test is perfect when my hands don't touch the phone body. \$\endgroup\$
    – Yu-Chih Tung
    Commented Oct 29, 2015 at 16:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.