Active noise reduction, anything new on the marked

[Tom]

In the 70s- 80s I have been reading a lot about noise reduction by simply
producing the same spectrum shifted by 180 degrees.



I have not heart anything about this since then. I wonder if there is
anything or anybody working on this. Or this idea simply did not work. I
guess it would require very fast computer to calculate the same spectrum
shifted by 180 degrees, so when the waves meet would, the output would be
wave with amplitude equal or close to 0.

 

[Pooh Bear]

In the 70s- 80s I have been reading a lot about noise reduction by simply
producing the same spectrum shifted by 180 degrees.

I have not heart anything about this since then. I wonder if there is
anything or anybody working on this. Or this idea simply did not work. I
guess it would require very fast computer to calculate the same spectrum
shifted by 180 degrees, so when the waves meet would, the output would be
wave with amplitude equal or close to 0.

Noise is random. The concept can't work.

Graham

 

[Tim Williams]

In the 70s- 80s I have been reading

Ah, been smoking enough crack and marijuana that you still think it's the
20th century? Well, explains itself well enough. (c; (c;

a lot about noise reduction by simply
producing the same spectrum shifted by 180 degrees.

Ya, works great (I haven't tried it personally).

I guess it would require very fast computer to calculate the same
spectrum shifted by 180 degrees ...

Well, I suppose an FFT, phase shift and composition back to time domain
could work, but the usual way is to INVERT the signal with a single
transistor. ;-O!

Well, okay...20 transistors...(inverting op-amp, mic gain, output, double
for stereo.)

Note that, noise cancellation works great for localized noise, but anything
slightly too close or too far from the cancelling source isn't going to have
a perfect null. If you set up a noise cancelling device in a room (as if),
the chances are it'll actually double (or more?) the noise in other areas!

In a room, you also have acoustics against you, since the cancelling
amplifier has to detect the noise in the first place, requiring a
microphone. It takes a certain time for the sound to pass through the air,
to the mic, to the amplifier and back to the speaker, and the chances are,
some frequency is going to be amplified repeatedly until some element (mic
or amplifier) is at its maximum signal level. This is otherwise known as
FEEDBACK...

As a result, noise cancellation is most effective for very local
cancellation, like headphones. Tack mics on the outside of a pair of
'phones, add an inverting amplifier, adjust gain and you can have reasonable
cancellation inside your ears.

Tim

 

[John Perry]

Tom wrote:



Noise is random. The concept can't work.

If you change "180 degrees" to "inverted", it does work, and is in
fairly common use right now.

John Perry

 

[Pooh Bear]

If you change "180 degrees" to "inverted", it does work, and is in
fairly common use right now.

For headphones yes.

Actually I had something else in mind when I replied but nm.

Graham

 

[Richard Henry]

If you change "180 degrees" to "inverted", it does work, and is in
fairly common use right now.

And change "spectrum" to "waveform".

 

[Genome]

In the 70s- 80s I have been reading a lot about noise reduction by simply
producing the same spectrum shifted by 180 degrees.



I have not heart anything about this since then. I wonder if there is
anything or anybody working on this. Or this idea simply did not work. I
guess it would require very fast computer to calculate the same spectrum
shifted by 180 degrees, so when the waves meet would, the output would be
wave with amplitude equal or close to 0.

If you buy a helicopter then you can have one.

DNA

 

[Genome]

If you buy a helicopter then you can have one.

DNA

Oh, and if you join the right agency.... they come in black.

DNA

 

[Phil Hobbs]

Oh, and if you join the right agency.... they come in black.

DNA

Piker. _My_ black helicopters all have megaphone exhausts.

There are methods that can do this under certain rather restrictive
assumptions, mainly that you can adequately mike all the noise sources,
that you have lots of appropriately-placed speakers, that you want it to
work only in a very restricted volume, and that nothing is changing
rapidly in the acoustic configuration, e.g. large scale turbulence. It
would take a gigantic adaptive system, and if you had enough computing
power, you might be able to update it fast enough to take account of
seasonal temperature variations. ;-)

A similar sort of thing is done in noise abatement systems for
smokestacks, IIRC--it helps a lot that it's the low-frequency rumble
that they're trying to get rid of.

Cheers,

Phil Hobbs

 

[Genome]

Piker. _My_ black helicopters all have megaphone exhausts.

There are methods that can do this under certain rather restrictive
assumptions, mainly that you can adequately mike all the noise sources,
that you have lots of appropriately-placed speakers, that you want it to
work only in a very restricted volume, and that nothing is changing
rapidly in the acoustic configuration, e.g. large scale turbulence. It
would take a gigantic adaptive system, and if you had enough computing
power, you might be able to update it fast enough to take account of
seasonal temperature variations. ;-)

A similar sort of thing is done in noise abatement systems for
smokestacks, IIRC--it helps a lot that it's the low-frequency rumble
that they're trying to get rid of.

Cheers,

Phil Hobbs

I meant the one that comes in the headsets for the drivers and
passengers....

Burp

DNA

 

[Joseph2k]

Noise is random. The concept can't work.

Graham

Not so, it is just computationally intractable. You can separate the
various components of a signal by using autocorrolation. You must apply it
recursively with auto adaptation until all the signal separations converge.
If you are working on a four person band with two singers it takes about 1
pentium GHz hour per minute of audio. For a twenty piece "orchestra"
figure on fourty times the time. It is NP hard after all. There should be
one dominant noise source and one or more lesser ones. I have a recording
that has three nearly equal power noise sources (about 40 dB below the
signal). I figure if i can ever get it noise reduced that i would like a
copy of the three noise sources without the main signal for a test signal
to add to recordings to test noise reduction systems.