I have a noisy car. The engine is not a real issue, the wind noise increases quickly with speed, even at allowed highway speed.

I have some around-ear headphones with good sound insulation but I don't want to wear them as-is because I would lose awareness of the surroundings.

My guess is that I need to keep the sound in an audio band for speech, plus one for horns, and one for emergency sirens from ambulance and police. Wind noise is broadband and if these bands are well selected, it could be reduces significantly.

I thought about making a circuit with:

  • electret capsule,
  • multi-band equalizer to keep only the useful audio bands,
  • audio mixer to sum the various audio bands with the line-level output of my car radio.

Everything powered by a regulated 9-11V supply from the car battery.

The electret circuit is easy and compact (for example Do electret condenser microphones require phantom power? ).

The signal mixer is also simple to implement (for example fig. 4 here).

The multi-band equalizer is not, since it requires a passband and opamp for each band.

With that in mind, how can I optimally reduce the number of audio bands with useful audio signals to maintain awareness while driving (in this case the mixer would have the n audio bands plus the external audio input), or what can I use to replace the discrete equalizer with some IC?

Maybe going digital is easier? however I only know Arduino and that one has a poor ADC and no DAC.

  • \$\begingroup\$ Very off topic: I added sound proofing, a copy of Dynamat, to pretty much all sheet metal in my car, doors, floor and some quarter panels and it was an imediate 10 dB reduction of engine and road noise. Highly recommended and cheap but labour intensive. \$\endgroup\$
    – winny
    Commented Aug 19, 2016 at 10:33
  • \$\begingroup\$ @winny I added bitumen+aluminium strips in many metal surfaces of doors and lateral panels, replaced the wool+bitumen layer below the carpet and added some sound insulation rubber (14 kg/m^2) to the floor, skipped wheel arches and engine room wall due to lack of time and difficulty in accessing them. The improvement is there, but in no way this affects wind noise from the upper part of the vehicle. I could place the bitumen+aluminium foil strips below the roof, but it wouldn't change much: I have thin glass that is also flat, meaning noise gets through easily. I won't put strips on windscreen. \$\endgroup\$
    – FarO
    Commented Aug 19, 2016 at 11:54
  • \$\begingroup\$ I see you are no stranger to car modifications :-) Yes, wind noise is difficult. Old BMW 7-series came with double windows for this very purpose. Very quiet but very heavy. Luxuary was a real integrate part of it but the econometer stopped at 30 liter/100 km and was bottomed out most of the time even under normal driving with a V12 and all that glass :-) \$\endgroup\$
    – winny
    Commented Aug 19, 2016 at 12:02

2 Answers 2


Probably easiest to make a simple filter that covers only the speech band (300Hz to 3kHz.)

That should cover the horns and sirens as well.

So, you need the following:

  1. Input amplifier to bring the electret microphone signal up to something workable.
  2. A high pass filter with cutoff around 300Hz
  3. A low pass filter with a curoff around 3kHz
  4. An output amplifier to drive the headphone speakers.

No need for mixers or anything. Just string the stages one after the other using AC coupling.

You probably ought to start by looking into the individual parts listed above (how to make a high pass filter with an op-amp, how to make a low pass filter, how to amplify a microphone signal, what to use to drive a speaker) then rough out a circuit diagram, then ask here when you have trouble with specific sections.

Noise reduction is far from simple. A DSP based system is probably far from your abilities right now.

  • \$\begingroup\$ The mixer is to sum the resulting signal to music I can play at the same time from the car radio. \$\endgroup\$
    – FarO
    Commented Aug 19, 2016 at 9:45
  • \$\begingroup\$ Ah. That's not so much a problem. Easy to solve once you have the rest of the circuit. Concentrate just on getting the filtered microphone audio to the headphones. Mixing in an additional signal is pretty trivial, then. \$\endgroup\$
    – JRE
    Commented Aug 19, 2016 at 9:47
  • \$\begingroup\$ Concerning the DSP, I was simply thinking about a minimal R-Pi with mic plus line inputs, equalizer, audio output, not to a real DSP card. \$\endgroup\$
    – FarO
    Commented Aug 19, 2016 at 9:47
  • \$\begingroup\$ If you go the Pi route, I might suggest using PureData to experiment with. There's also a noise reduction filter for it that you can find on the PureData forums. \$\endgroup\$
    – JRE
    Commented Aug 19, 2016 at 9:48

I don't think you need the complexity of a multiband equaliser to do what you want to do.

Consider the telephony bandwidth, 300Hz to 3.4kHz. It encompasses intelligible speech, horns and sirens, and cuts out a huge chunk of high frequency noise.

It is relatively straightforward to include the filter components around the amplifiers that you will need to change the level from microphone to headphones.

Even if that isn't the final system you end up with, it would be worth prototyping something as simple as that to see where it gets you.


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