Noise Cancellation Frequency Inverter

Yevmeister · 10-16-2006, 11:50 AM

Dear members i am starting working on active noise cancellation for automotive, and household uses

i know there are existing topics on this subject, but they just have suggestions

One of the poeple i know has Acura RL with similar set up, i believe it's the only car on the market available on the market with such feature

questions i need to be answered

1) are there are stand alone devices avaliable on the market that produce negative frequencies on whatever audio input (there are some noise cancelling headphoens on the market, that i could disassemble to get the chip, but first i wanted to know what that device is called for experiements

2) I am fairly new to microphones, and sensitive microphones, i know what you pay is what you get, but are there good sensitive microphones out there with under $50 USD price range

so that's it so far, i lot of people said it's complicated to do, i don't really see it as complicated issue, 50-60% effectiveness for audio deadening is success to me

Yev

evandude · 10-16-2006, 12:17 PM

Noise cancellation relies on generating a signal that is 180 degrees out of phase with the noise, and having it arrive at your ear at the same time, giving rise to destructive interference.
Noise cancelling headphones work so well because the microphone and the speaker are both very close to your ear (relative to the wavelengths of sound), so the microphone can pick up nearly all of the external sound that would have ever reached your ear.

In a car, you have a ton of reflections, and the noise is coming from all around. If you place a microphone somewhere other than the person's ear, there's no way for it to know exactly what sound is actually at their ear, and thus it can't know exactly what signal to produce to cancel it. Maybe if you had some sort of detailed audio profile of the car and could generate some sort of custom filter to predict it... but that's way out of the range of a hobbyist project. And, since you would be generating the cancellation signal from a speaker that was not next to the person's ear, you'd have the same issue there, plus you'd have even more issues with the reflection of that signal. I just don't expect it would be possible to make a system that wasn't heavily engineered for the specific car that it was in (making a drop-in solution impossible), and even that would be way too difficult for a hobbyist. Even if you got it tailored for your car, you'd probably be looking at quite a bit of complicated Digital Signal Processing to actually implement the system... at the very least, I would expect it would be WAY beyond just hacking up some noise-cancelling headphones. I can see how it's possible for this kind of system to be put in a luxury car, where the engineering is done and the system is implemented in a bunch of identical cars... but not in a general case.

One thing you may not realize is that while 50-60% effectiveness sounds like it should be easy, if your "noise cancelling" system isn't providing sound with just the right phase, not only will it not reduce noise like you'd expect, it can just as easily make the noise WORSE, if it gives rise to constructive interference instead.

I guess a good first proof-of-concept experiment, if you really feel you want to attempt it, would be to buy a set of cheap noise-cancelling headphones, tear them apart, and hook a small amplifier and speaker up to where the headphone speaker was, and listen to how well they cancel noise when you place them further away from your ear... try placing them several feet away, like where your car speakers would be...

Dragonknell911 · 10-16-2006, 01:00 PM

What exactly were you trying to gain by this?

Yevmeister · 10-16-2006, 01:24 PM

i am just trying to help my frind out eho lives in noisy neighborhood, and has trouble sleeping, i am thinking of 2 parabolic dishes with microphones places near his window> signal processed my inverted, and played on a speaker position next to the window to cancel some amount of noise

Yevmeister · 10-16-2006, 01:30 PM

Quote: Originally Posted by evandude

Noise cancellation relies on generating a signal that is 180 degrees out of phase with the noise, and having it arrive at your ear at the same time, giving rise to destructive interference.
Noise cancelling headphones work so well because the microphone and the speaker are both very close to your ear (relative to the wavelengths of sound), so the microphone can pick up nearly all of the external sound that would have ever reached your ear.

In a car, you have a ton of reflections, and the noise is coming from all around. If you place a microphone somewhere other than the person's ear, there's no way for it to know exactly what sound is actually at their ear, and thus it can't know exactly what signal to produce to cancel it. Maybe if you had some sort of detailed audio profile of the car and could generate some sort of custom filter to predict it... but that's way out of the range of a hobbyist project. And, since you would be generating the cancellation signal from a speaker that was not next to the person's ear, you'd have the same issue there, plus you'd have even more issues with the reflection of that signal. I just don't expect it would be possible to make a system that wasn't heavily engineered for the specific car that it was in (making a drop-in solution impossible), and even that would be way too difficult for a hobbyist. Even if you got it tailored for your car, you'd probably be looking at quite a bit of complicated Digital Signal Processing to actually implement the system... at the very least, I would expect it would be WAY beyond just hacking up some noise-cancelling headphones. I can see how it's possible for this kind of system to be put in a luxury car, where the engineering is done and the system is implemented in a bunch of identical cars... but not in a general case.

One thing you may not realize is that while 50-60% effectiveness sounds like it should be easy, if your "noise cancelling" system isn't providing sound with just the right phase, not only will it not reduce noise like you'd expect, it can just as easily make the noise WORSE, if it gives rise to constructive interference instead.

I guess a good first proof-of-concept experiment, if you really feel you want to attempt it, would be to buy a set of cheap noise-cancelling headphones, tear them apart, and hook a small amplifier and speaker up to where the headphone speaker was, and listen to how well they cancel noise when you place them further away from your ear... try placing them several feet away, like where your car speakers would be...

all you said is pretty much thue, but the bozos at BOSE (i hate them so much) actually made it work out pretty neatly in the Acura RL

metal transfer audio pretty well, and the way their system is set it is 2 sensitive microphones attached between headlined, and and a metal body, one of them is aboive driver's seats, and one in the rear passenger area.

they pick up the signals, and play them along while person is listening to music, and have the stereo mute, i swear i could hardly hear a passing car with windows open.

having it implement in car should not be hard, cause there are softwares on the market that would do such function, simultaniously with the media player

Yev

jasonsjwou · 10-16-2006, 01:31 PM

Acoustic Noise Cancellation... Honda does it, Nissan does it, Some business jet does it too. Certainly it wouldn't hurt too much to give it a try. Just beaware that it is a lot more than just record-invert-output.... there are tons of Scienctific journals (like Journal of Acoustics and Vibration, IEEE etc) that has studies concerning this. I've read few of these papers.

You basically have to do a spatial spectrum analysis. It only works in very limited bandwidth, and very very low frequency (almost like a rumble). I am afraid you'll just waste your money if you tear one of those noise cancelling headphone to rewire it.

Anyway

A good microphone - Behringer ECM8000. This is a measurement microphone that is used in many RTA equipment. I use it with my RTA/EQ. It's not the est, but you can buy it for $50. You'd need a phantom power supply though. That'll run you another $40. Check out www.musiciansfriend.com. I think you might be able to remove the top and have rest of the microphone hidden somewhere else.

Yevmeister · 10-16-2006, 01:36 PM

my third question

what are the softwares available that capture, and invert waves for that purpose

Yev

Yevmeister · 10-16-2006, 01:42 PM

i have also found this write up
http://www.headwize.com/projects/noise_prj.htm

the problem is that i am not circuitry inclined

Yev

evandude · 10-16-2006, 01:56 PM

Quote: Originally Posted by Yevmeister

what are the softwares available that capture, and invert waves for that purpose

I think there's quite a bit more to it than that. if that's all there was to it, then in theory you would be able to just wire a microphone to an amplifier to a speaker and just flip the speaker wires so it was running out-of-phase with the ambient sound. if you do that, I'm sure you'll see some amount of interference, but it's certainly not going to be very good cancellation in any situation other than perhaps headphones.

And as stated by someone who clearly knows more about the details of it:

Quote: Originally Posted by jasonsjwou

You basically have to do a spatial spectrum analysis. It only works in very limited bandwidth, and very very low frequency (almost like a rumble).

Dragonknell911 · 10-16-2006, 04:25 PM

If all your trying to do is help your friend sleep better at night in a noisy enviornment, there is always white/ambient controlled noise

Yevmeister · 10-16-2006, 05:31 PM

Quote: Originally Posted by Dragonknell911

If all your trying to do is help your friend sleep better at night in a noisy enviornment, there is always white/ambient controlled noise

no it's not only that, it noisy during the day as well, i recommended him white noise generator, but he doesn;t like it as much

Yev

stuthemong · 10-30-2006, 08:55 AM

You are really going to have no joy here chap, bin the idea and move on IMO. Bit of info for you....

Headphones can cancel noise using analog circitry, due to the mic + speakers proximity to ear, and most importantly their insulation from eachother (mic only hears noise, not noice+speaker as it is external to speaker can). You will have zero joy trying to replicate this.

Noise cancellation in-room is possible, though as mentioned really only ever works with low freq harmonic sounds. The reason for this is that you program a DSP chip, and you are limited by the delay through the chip (delay in cycles of noise to cancel). You typically have a good few ms delay - if you can get the cancel signal out <10ms you have done very well (remembering the ADC - DSP (Z-transform and long filter multiplication) - DAC - playback.

If it was possible to get an infinitely fast unit you could cancel any sounds, but its impossible really, so all the units can do is cancel sound that doesnt change much from cycle to cycle - back to the 'drone' type noises. Trying to cancel something like a bottle smashing is impossible with a system like this, as the noise is very random.

The algorithm best suited to this application would be the filtered xLMS - do a paper search on F-XLMS and you should see the recursive formula you can use (after system characterisation).

Youll do much better with passive techniques, better windows, sound insulation etc - will be cheaper and more effective IMO

DavC · 10-30-2006, 10:45 AM

spend the money on some double/tripple glazed windows, and some wall/roof insulation.

oh, and a pair of ear plugs.

durwood · 10-30-2006, 10:50 AM

With music it can be done but the music/sound has to be delayed..almost put on hold until the system has to have time to analyze the sound before you apply the noise cancelation. If the noise is a constant frequency/sound (like in a jet) then it is rather simple. Otherwise you need super fast processing to do what you are talking about (adaptive noise cancelation) in real-time and even then its near impossible. Just my 2 cents.