If one had to design a respirator that would have the minimal sound distortion, and allow to communicate normally (one-on-one at least), how would one do it?

I wear powered respirators (PAPR) and valved respirators at work, 8-10 hrs a day. This is for biomedical and chemical work I do and, recently, for Covid. The key limitation of all respirators is that they attenuate the sound of human speech. I have tried implanting a microphone and fed the signal to a speaker I wear on the belt. It works, but it's clunky, batteries run low, there is interference etc. In addition, PAPRs blow air on the face, and microphones pick that up.

The rest is my current thought process, I apologize if this is rambly.

Most respirators place multiple layers of material (silicon valves, plastic ducts) in front of the mouth which, from what I understand, attenuate the vibrations and distort the sound. 3M now makes a mask that has a speaker diaphragm which appears to be a Kapton membrane about 40 mm in diameter secured in a plastic frame. Ironically, the diaphragm is placed in front of the user's nose, not the mouth. The sound comes out less distorted compared to a regular respirators, indeed, but it is attenuated.

I am thinking to improve on the 3M design by 3D printing the mask shell, and implant a much bigger membrane (or other material?) in front of the mouth. The transparent membrane would allow one's lips to be read as well. The sketch shows what I am thinking to do -- print my own mask, reproduce 3M bayonets (already have CADs for those), leave a large opening in the front, and use a teflon o-ring gasket to secure the membrane in place (observation-glass style, with a 3D printed steel or Alu plate).

Does this idea have merit? Are there any materials (better than Kapton) that can still hold the positive/negative pressure and transmit sound waves efficiently? Would the o-ring used to secure the membrane attenuate the sound and prevent the membrane from transmitting the sound properly? Are there resonant modes I should consider? Does the membrane shape matter?

enter image description here

  • 1
    $\begingroup$ Build a walkie talkie into it. Already done for divers and space suits… and jet pilot helmets… $\endgroup$
    – Solar Mike
    Oct 3 at 17:12
  • $\begingroup$ Solar Mike, I cannot distribute walkie-talkies to 300 medical students. The fee air-blowing aspect is the limitation (interference from the microphone). I want the solution to be such that only the mask-wearer has to implement it. Some need to wear masks longer than others, due to immune suppression, and cannot expect everyone else to do the same. $\endgroup$ Oct 3 at 18:25
  • $\begingroup$ You failed to mention numbers, like 300, in your question. However, I gave you those comments to help you think, where was my error? The suggestions or the other option? $\endgroup$
    – Solar Mike
    Oct 3 at 18:35
  • $\begingroup$ Mike, I apologize, I did not mean to say that you made an error. I wanted to keep the question to the point, and indeed have failed to explain the setting (many other people involved), and which other solutions I have tried (implanted microphones etc). $\endgroup$ Oct 3 at 19:17
  • $\begingroup$ Design a respirator without covering the mouth but still can be tightly fitted to the face. $\endgroup$
    – r13
    Oct 3 at 21:11

Some hearing aids have a biult in microphone and have adjustable angle of recption and or can transmit sound to un external speaker.

My hearing aids self agjust when i wear my pilot's head sets. while still receiving calls from my phone in my bag.

In your communty many share the same issue, try to find out what they do.

It can even effect your cognitive acuity. you want to fix it.

  • $\begingroup$ In my community they, indeed, face this issue. I am the only one who is seeking solutions (microphones, speakers etc); others quit, or struggle. I haven't found a working solution yet, and I have been researching and testing various masks since the pandemic started. Microphones pick up the noise of the air traveling at 4 m/s when air is blown inside of the PAPR. There are some "nanomembrane"-based face masks, but they are all at the R and D stage (one-off paper in a scientific journal which does not result in a product). I am also looking into nasal respirators. $\endgroup$ Oct 3 at 18:19
  • $\begingroup$ my daughter is a interventional radiologist. she quit working at the hospital for the same reason, she took a big pay cut and now she just reads images from home office. $\endgroup$
    – kamran
    Oct 3 at 18:53
  • $\begingroup$ Kamran -- the NHS (in the UK) has been very slow at distributing PPE, and personnel at non-covid wards wears surgical masks (better than nothing, but insufficient). When I wear my 3M respirator, nurses keep asking me what kind kind of mask this is (almost 2 years into the pandemic). There have been minimal disruption in respirator supplies; I was able to get my PPE off ebay for personal cash as I could not count on the management. $\endgroup$ Oct 3 at 19:24

The vibrating membrane has a fundamental resonant frequency at which it becomes acoustically transparent that has a wavelength approximately equal to the diameter of the membrane. Increasing the diameter of the membrane will lower the cutoff frequency of the membrane so it is closer to the sound spectrum of human speech. In addition, the radiation resistance of the membrane scales as the membrane radius squared, so the bigger membrane will radiate sound more efficiently.

This means that the big membrane will have more bass in its response and will be slightly louder. Whether or not these effects will make the speech more intelligible can only be determined by experiment, and I encourage you to share your experimental results with us here.

  • $\begingroup$ Multiple membranes like a speaker? $\endgroup$
    – DKNguyen
    Oct 4 at 4:34
  • $\begingroup$ @DKNguyen, a membrane with a sound source on one side of it will behave exactly like a loudspeaker. $\endgroup$ Oct 4 at 16:26
  • $\begingroup$ Niels -- I have carried out tests with parts I sourced off ebay. I am going to posts the result shortly. You were exactly right re: the resonance. My own voice created a "buzzing" artefact whereas my wife's voice did not (the membrane is 63 mm in diameter). I solved the issue by making the membrane loose (not stretched tight). $\endgroup$ Nov 10 at 18:42
  • $\begingroup$ @runcyclexcski, most excellent! Now I can pretend to be an engineer again! $\endgroup$ Nov 10 at 19:09

I would like to share the result of an experiment I did on this. Niels's response has stirred me in this direction.

The membrane was a 63 mm diam dyneema fabric (strong and light) loosely stretched over a 63 mm diam aluminum pipe (turbo pipe I got off ebay). The pipe was worked into into a rubber 3M part to provide face seal (the part No is shown in the picture). I found that the fabric had be loose (i.e. not tightly stretched around the pipe), otherwise it created a buzzing artefact with my voice (but not with my wife's voice). At a 2mm distance, my wife indicated that me speaking through the mask was "almost" as good as me speaking w/o it. A regular 3M 7000 series mask was barely clear. Using 50 micron Al foil did not add distortion, but tuned down the amplitude (again, this is subjective).

Overall, I think it worked, but I now need to 3D print a new mask shell, since the frontal area that is normally occupied by the exhale valve is now occupied by the membrane. To make the test quantitative, I am thinking to read a text into a microphone different distances away and have a voice-recognition software "understand" what I am saying. Then score the % of the words it got wrong with the special mask, with a regular mask, and with no mask.

Another direction might be to make the pipe a "funnel shape" with a larger opening and stretch the membrane over that (e.g. 100 mm in diam). If there is a resource which I could use to simulate resonant and cut-off frequencies of different shapes it would be useful. Ideally, I think I want to keep the membrane flat and transparent b.c. one could potentially read the lips through it.

enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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