I ran across an ad for noise reducing headphones the other day that sparked an idea in my head: I’ve got a Honda Civic that gets obnoxiously loud once it hits about 65 MPH. I’ve also got a decent background in physics, and a bunch of spare audio equipment lying around 
Here’s what I was thinking: put a microphone in the trunk of a my car, where it would still pick up road noise but be shielded from the wind and any possible feedback, feed it into my laptop, reverse the polarity in realtime (if this is possible), then play it through the car stereo system. If my physics is right, shouldn’t this cancel the roadnoise reasonably well? Has anyone ever experimented with this before, or have any ideas?
I have a feeling that using a laptop and your stereo system isn’t going to cut it. To be effective, the entire sequence of gathering the data, reversing the phase, and broadcasting the altered sound must be done very quickly. I doubt most home computers are designed to handle such real-time tasks. You’d likely need as system designed from the hardware level up specificly for this task to be able to do this efficiently enough to work.
Also, I wonder if the acoustic properties of the trunk are different enough from the passenger area to make such a setup feasible. In other words, will the trunk even get the same road noise that you’re hearing?
It’s a really great idea. Unfortunately a lot of car manufacturers have already beaten you to it. You’re going to need more than 1 microphone and more speakers than you got now. Typically these systems have 8 or 10 speakers at a minimum. You’re forgetting that sounds come from 3 dimensions. Your single microphone idea only works if they are coming from a single direction.
I could have sworn one of the luxry car manufacturers like Cadillac had already put a system like this into production, but my google skills are failing me. It’s possible that the article I read was only about a prototype system.
I think this is the key point. The sound in the trunk will be vastly different then what your ears hear up front (unless of curse you are planning to reduce road noise and screams coming from the trunk. In which case, I don’t want to know about it. ;))
There are two different models of noise reduction you could try, each with it’s own problems.
-
The headphone model puts the mike(s) right on your head, which gets the most accurate sample of the noise that needs to be canceled. The headphones themselves have two important functions; to generate the same noise 180 degrees out of phase (a really simple process that doesn’t require a computer, AFAIK, merely switching the wires in the mike cable leading to the ampliofier*), and to isolate the mike from the amplifier-created negative sound. It would be very difficult to re-create this model without using headphones, because the farther away the mike is from your ears, the less accurate the sample it gets of the offending noise, and the less effective it is.
-
The other model is to get at the noise at it’s sources. At each identified source of noise, you put a tiny speaker that generates a corresponding anti-noise. If it’s a regular mechanical noise, say the AC fan, you don’y even need a mike. In theory, you can alanyse the sound on an oscilloscope, and build an amp which generates the appropriate anti-noise, and place the speaker as near as possible to the fan. Repeat for each noisy mechanical element. There will be other sources of environmental noise that will, however, need sampling by a mike; the effect of various wind directions/velocities and pavement types on the frame and suspension, for example.
I would suggest the headphone model is far easier and cheaper.
*My personal experience with anti-noise involved recording interviews in noisy outdoor environments. The point was not to reduce the noise to my ears, but to reduce it on the recording of the interview, so a loudspeaker was not involved. How it worked was this: You put one mike in front of your subject, and a second mike maybe 10 feet away with the subject well out of the pattern. The second mike used a cable in which the negative and positive wires were reversed. The two mikes were mixed into the same channel. The result was two mikes feeding in the ambient noise out of phase to each other and largely cancelling each other out, with the subject voice being relatively unaffected. It worked pretty well, and there was no sophisticated signal processing necessary.
I think what you’re talking about are called “noise cancellation headphones” and they’re made by many manufacturers. They have been made for some time for airplane pilots, and also have the communications radio built in (you can add another signal to the cancellation signal so you’ve got something to listen to).
IIRC Bose makes them for general purpose use and has been advertising them on television recently.
BTW you can’t use signals from the noise sources themselves because that way you don’t know the effects of all the echos from other objects in the environment.
That’s partly true, and the result is that the noise cancellation is not perfect. But, the closer the two sources are to each other, the more similar their sound reflections will be, and the more effecient the cancellation will be. I have seen media stories about plans to build noise cancellation into industrial and home products like air handlers based my second model.
or, instead of wasting money on something like this, you could pinpoint the problem of WHY your car is letting in more road noise than it should…
maybe you need new tires? maybe its vibration from a bent wheel…
if it bothers you that much, go buy some acoustic deadening material, avaliable at any car audio place…
but i still say something is wrong with your car…