Turn Your House Into Noise Cancellation Headphones

[glen herrmannsfeldt]

In comp.dsp Jerry Avins <jya@ieee.org> wrote:

(snip, I wrote)

To do it well, though, would take a LOT of speakers and microphones.

Maybe one per square meter of wall and ceiling area. (You get
a quantity discount on the parts.) Calibration would also be
interesting.

Calibration will change as occupants move around and as doors and
windows are repositioned.

You mean opened? Doors and windows normally don't move around in
the walls. Yes, they have to stay closed. (Maybe an airlock instead
of a door.)

With enough DSP, maybe you could get away with fewer microphones,
but still with many speakers, by doing some correlation and delay,
and with the assumption that the sound mostly comes from one,
or a small number of, directions.

When the noise of an airplane comes through walls and windows, which
directions might those be?

If you consider the airplane as a point source, ignoring reflections
off other houses and fences, then you compute the direction to
the source from using correlators, apply the appropriate delay to
each speaker. That depends a lot on the house position relative
to other houses, the direction of the runway, and the direction
the airplanes go.

That depends on the speakers being much closer than the shortest
wavelength of interest, but I believe that jet fighters make mostly
low frequencies. (One reason why noise cancelling headphones work
well on airplanes.)

-- glen

 

[Bret Cahill]

If you have a whole wall of microphones outside, and speakers inside,
each adjusted with the appropriate delay (though the wall) and
amplitude (how much leaks through the wall that needs cancelling)
then, for wavelengths somewhat longer than the speaker spacing it
should pretty much cancel throughout the room.
What about windows in the wall?

I wondered about windows, but didn't write about them until today.
Our living room has a large (about 6 by 10 foot) window, which
would not work well with the suggested system.  An array of smaller
windows, that allowed space for the speakers, might.

Or equivalently, speakers spaced out on a large window. Another
solution, at least as ugly is to use the window glass or plastic
itself as a speaker cone. Lots of coils and magnets would be visible,
however.

Maybe spin it as post modern architecture.

Presumably, one should start with good mechanical (passive) noise
reduction, such as (someone mentioned) offset studs to reduce the
coupling through the walls.  With enough such reduction, windows,
even triple pane, might be the big leak source.  

The dome skylights that make loud plonks when it rains have been re
done to be quieter but that's an easier problem.

One might also
want an air exchange system that didn't allow noise through.
One should be able to test ideas with a smaller system, big enough
for one person to sit inside, and surrounded by materials commonly
used for walls.  That should allow testing for a modest cost,
before going for the full-sized system (house).  Among others,
to test the frequency dependence and processing needed for each.

First do the much easier 2-D version, water waves generated on the rim
of a pond. Children like to splash a lot so kids would make low cost
wave generators.

Obviously a circular wall would block all the water waves from getting
inside the wall but that's just the 2-D equivalent of the trivial
bunker solution.

Instead of a wall you must have an enclosure of under water surfaces
that react in concert to eliminate all waves in a small area somewhere
in the box. The rectangle of panels is broken one one or more sides
with holes to act as windows.

If that can be done then it could lead to doing it in 3-D.

If the source can be considered a point source some distance away,

Navy jets are at about 50 meters altitude by the time they're over a
civilian neighborhood. That radius of curvature presents a pretty
flat front over the length of a room.

such that it is a plane wave as it arrives, a smaller number of
microphones and appropriate programmable delays might be able
to drive a large speaker array.

It might be cost effective to just eliminate most of the noise with
far fewer speakers.

To me, the Bessel array idea isn't so much use for noise reduction,
but is needed if one also wants to use the speaker array for an
audio system.  

 

[Robert Adams]

On Sep 16, 6:13 pm, Bret Cahill <Bret_E_Cah...@yahoo.com> wrote:

If you have a whole wall of microphones outside, and speakers inside,
each adjusted with the appropriate delay (though the wall) and
amplitude (how much leaks through the wall that needs cancelling)
then, for wavelengths somewhat longer than the speaker spacing it
should pretty much cancel throughout the room.
What about windows in the wall?

I wondered about windows, but didn't write about them until today.
Our living room has a large (about 6 by 10 foot) window, which
would not work well with the suggested system.  An array of smaller
windows, that allowed space for the speakers, might.

Or equivalently, speakers spaced out on a large window.  Another
solution, at least as ugly is to use the window glass or plastic
itself as a speaker cone.  Lots of coils and magnets would be visible,
however.

Maybe spin it as post modern architecture.

Presumably, one should start with good mechanical (passive) noise
reduction, such as (someone mentioned) offset studs to reduce the
coupling through the walls.  With enough such reduction, windows,
even triple pane, might be the big leak source.  

The dome skylights that make loud plonks when it rains have been re
done to be quieter but that's an easier problem.

One might also
want an air exchange system that didn't allow noise through.
One should be able to test ideas with a smaller system, big enough
for one person to sit inside, and surrounded by materials commonly
used for walls.  That should allow testing for a modest cost,
before going for the full-sized system (house).  Among others,
to test the frequency dependence and processing needed for each.

First do the much easier 2-D version, water waves generated on the rim
of a pond.  Children like to splash a lot so kids would make low cost
wave generators.

Obviously a circular wall would block all the water waves from getting
inside the wall but that's just the 2-D equivalent of the trivial
bunker solution.

Instead of a wall you must have an enclosure of under water surfaces
that react in concert to eliminate all waves in a small area somewhere
in the box.  The rectangle of panels is broken one one or more sides
with holes to act as windows.

If that can be done then it could lead to doing it in 3-D.

If the source can be considered a point source some distance away,

Navy jets are at about 50 meters altitude by the time they're over a
civilian neighborhood.  That radius of curvature presents a pretty
flat front over the length of a room.

such that it is a plane wave as it arrives, a smaller number of
microphones and appropriate programmable delays might be able
to drive a large speaker array.

It might be cost effective to just eliminate most of the noise with
far fewer speakers.



To me, the Bessel array idea isn't so much use for noise reduction,
but is needed if one also wants to use the speaker array for an
audio system.  - Hide quoted text -

- Show quoted text -

One issue I have run into in the past when experiementing with
algorithms of this type is the notion of "coherence". Based on my
experience I would guess that if you measured the coherence between
the jet noise at one point in the room relative to another microphone
placed near your ear, the coherence would be nearly zero. This means
there is NO FILTER that can pick up the noise far away and then make
an anti-noise at another location that is closer to your ear. I have
always been a bit confused about WHY this is this case; you would
think that if air is linear and time-invariant there should be some
transfer function between two points in an acoustic space. Maybe
someone here can enlighten us on this subject.

Bob

 

[Marvin the Martian]

On Fri, 16 Sep 2011 15:37:43 -0400, default wrote:

On Fri, 16 Sep 2011 10:46:45 -0500, Marvin the Martian
marvin@ontomars.org> wrote:


IF you think so, give it a go. :-D The problem comes is that you end up
negating the sound very near the speakers, but since the speakers are
emitting sound from a point just behind each speaker in the wall, but
the noise is coming from a point source - the distant jet engines.

What I was thinking is that for it to work you'd pretty much have to be
surrounded by a big cylindrical speaker - or there'd have to be some
complex electronics to check the position of the ears in the room and
adjust the speakers accordingly.

I anxiously await your paper and experimental results.

 

[Marvin the Martian]

On Fri, 16 Sep 2011 21:07:22 +0000, glen herrmannsfeldt wrote:

Different posts have made different suggestions, but mine have been
pretty much an array of microphone/processor/speaker devices, each to
cancel from one spot on the wall. If spaced somewhat closer than the
wavelength, the direction of the noise doesn't matter.

That will work close to the wall, but since sound has direction, and
propagates from a source point, and the source point of the noise is far
away and the source point of the speakers is just behind the speakers,
you will not be able to cancel the whole room.

There is the complication of large windows, with no place to put one, so
one might have to avoid those. There is also the problem of
calibration, but it should be possible to work that out.

-- glen

You need Bessel array between the listener and the noise source. With a
properly controlled Bessel array, you can make it appear like your noise
canceling sound is at the same source as the noise, within a limited
area.

It is sort of like a sonic hologram, as someone put it.

 

[Marvin the Martian]

On Fri, 16 Sep 2011 21:19:49 +0000, glen herrmannsfeldt wrote:

In comp.dsp Jerry Avins <jya@ieee.org> wrote:

(snip)
"Bessel array" is a pretty loose term:

AES E-Library
Effective Performance of Bessel Arrays

The Bessel array is a configuration of five, seven, or nine identical
loudspeakers in an equal-spaced line array that provides the same
overall polar pattern as a single loudspeaker of the array. The results
of a computer simulation are described, which uses point sources to
determine the effective operating frequency range, working distance,
efficiency, power handling, maximum acoustic output,
efficiency-bandwidth product, and power-bandwidth product of the array.

(snip)
That confuses me. At RF, I would call it a broadside array, and its
directional characteristics would be highly frequency dependent when
the spacing is an appreciable fraction of wavelength.

I hadn't heard of this before, either. But note that the phase and
amplitude are not the same. My thought, though without going through
the math, is that they select the phase and amplitude to reduce, as much
as possible with the number of free parameters, the frequency and
directional dependence.

If you take an array of equal spaced, equal amplitude, and in phase
sources, there is a sharp directional dependence. But also note that,
in an actual speaker array, they are not point sources, but circular
sources ideally sized equal to the source spacing. That complicates the
math somewhat.

Consider an array of an odd number of sources, each opposite phase from
its neighbor. Far field, and being odd, all but one will cancel. Now
consider (I didn't do the math) the off axis frequency dependence. It is
obvious that some will add and some subtract, but one has to go through
the math to see which.

As I understand it, the Bessel array optimizes in the far field. Near
field might be very far off.

-- glen

Not all Bessel arrays are linear. They can also be a 2D array. As I said,
they pretty much have to be computer controlled in order to be used for
noise canceling.

 

[Bret Cahill]

If you have a whole wall of microphones outside, and speakers inside> >,
each adjusted with the appropriate delay (though the wall) and
amplitude (how much leaks through the wall that needs cancelling)
then, for wavelengths somewhat longer than the speaker spacing it
should pretty much cancel throughout the room.
What about windows in the wall?

I wondered about windows, but didn't write about them until today.
Our living room has a large (about 6 by 10 foot) window, which
would not work well with the suggested system. =A0An array of smaller
windows, that allowed space for the speakers, might.

Or equivalently, speakers spaced out on a large window. =A0Another
solution, at least as ugly is to use the window glass or plastic
itself as a speaker cone. =A0Lots of coils and magnets would be visible,
however.

Maybe spin it as post modern architecture.

Presumably, one should start with good mechanical (passive) noise
reduction, such as (someone mentioned) offset studs to reduce the
coupling through the walls. =A0With enough such reduction, windows,
even triple pane, might be the big leak source. =A0

The dome skylights that make loud plonks when it rains have been re
done to be quieter but that's an easier problem.

One might also
want an air exchange system that didn't allow noise through.
One should be able to test ideas with a smaller system, big enough
for one person to sit inside, and surrounded by materials commonly
used for walls. =A0That should allow testing for a modest cost,
before going for the full-sized system (house). =A0Among others,
to test the frequency dependence and processing needed for each.

First do the much easier 2-D version, water waves generated on the rim
of a pond. =A0Children like to splash a lot so kids would make low cost
wave generators.

Obviously a circular wall would block all the water waves from getting
inside the wall but that's just the 2-D equivalent of the trivial
bunker solution.

Instead of a wall you must have an enclosure of under water surfaces
that react in concert to eliminate all waves in a small area somewhere
in the box. =A0The rectangle of panels is broken one one or more sides
with holes to act as windows.

If that can be done then it could lead to doing it in 3-D.

If the source can be considered a point source some distance away,

Navy jets are at about 50 meters altitude by the time they're over a
civilian neighborhood. =A0That radius of curvature presents a pretty
flat front over the length of a room.

such that it is a plane wave as it arrives, a smaller number of
microphones and appropriate programmable delays might be able
to drive a large speaker array.

It might be cost effective to just eliminate most of the noise with
far fewer speakers.

To me, the Bessel array idea isn't so much use for noise reduction,
but is needed if one also wants to use the speaker array for an
audio system. =A0- Hide quoted text -

- Show quoted text -

One issue I have run into in the past when experiementing with
algorithms of this type is the notion of "coherence". Based on my
experience I would guess that if you measured the coherence between
the jet noise at one point in the room relative to another microphone
placed near your ear, the coherence would be nearly zero. This means
there is NO FILTER that can pick up the noise far away and then make
an anti-noise at another location that is closer to your ear. I have
always been a bit confused about WHY this is this case; you would
think that if air is linear and time-invariant there should be some
transfer function between two points in an acoustic space. Maybe
someone here can enlighten us on this subject.

Bob

In EM propagation for comm the lack of coherence is due to
reflections, and I think it'd be about as hard to get rid of
reflections with audio as it is with EM propagation.  Unless there's
only a point source and a receiver with nothing reflective or
dispersive in between, then there's going to be memory in the channel
that'll degrade the coherence.  

One frequency one source in a 2-D ripple tank looks like it would
scare many away:

http://www.falstad.com/ripple/

So here's another invention: A float instrumented with an
accelerometer. If you have the pool to yourself you try to get across
the pool in the shortest time making minimal waves. A hydrophone
provides prompt feedback.

A recent study indicated strength cross training helps runners and
cyclists but not swimmers. The theory is the stroke is so highly
technical upper body strength isn't all that important for swimming.

Let the other guy beat himself to death. You don't move one single
drop of water that doesn't need to be moved.


Bret Cahill

Eric Jacobsenhttp://www.ericjacobsen.orghttp://www.dsprelated.com/blogs-1//Eric_Jacobsen.php- Hide quoted text -

- Show quoted text -

 

[Mark]

 

One frequency one source in a 2-D ripple tank looks like it would
scare many away:

http://www.falstad.com/ripple/


this is awesome..

thanks!

Mark

 

[Bret Cahill]

One frequency one source in a 2-D ripple tank looks like it would
scare many away:

http://www.falstad.com/ripple/

this is awesome..

thanks!

If you figure out a way to get a good sized flat patch in the middle
let us know.


Bret Cahill

 

On Fri, 16 Sep 2011 21:12:14 -0500, Marvin the Martian
<marvin@ontomars.org> wrote:

On Fri, 16 Sep 2011 15:37:43 -0400, default wrote:

On Fri, 16 Sep 2011 10:46:45 -0500, Marvin the Martian
marvin@ontomars.org> wrote:


IF you think so, give it a go. :-D The problem comes is that you end up
negating the sound very near the speakers, but since the speakers are
emitting sound from a point just behind each speaker in the wall, but
the noise is coming from a point source - the distant jet engines.

What I was thinking is that for it to work you'd pretty much have to be
surrounded by a big cylindrical speaker - or there'd have to be some
complex electronics to check the position of the ears in the room and
adjust the speakers accordingly.

I anxiously await your paper and experimental results.

I'd go for the soundproofed house idea first. Cheaper, doable with
current technology, and added insulation benefits.

Using today's technology to solve this problem is likely to be cost
ineffective. Cheaper to just wear noise canceling ear protection.
(for the privilege of living close to a runway?)

 

[Bob Masta]

On Wed, 14 Sep 2011 16:04:09 -0700 (PDT), Bret Cahill
<Bret_E_Cahill@yahoo.com> wrote:

There is some otherwise nice real estate at the end of Navy runways,
neighborhoods in San Diego and Virginia Beach where pilots practice
taking off an aircraft carrier with after burners wide open.

A lot of residents would be willing to pay $5,000 or more to be able
to talk to other people in a room at the flip of a switch. If it
draws a lot of power or when there is little outdoor noise it might be
desirable to turn it off or have it automatically turn off after 5
minutes and then back on as soon as the noise exceeds a threshold.

Noise cancellation should be cheaper than redoing the walls and in
some ways it might be an easier problem than headphones where the
distances involved are only a cm.

A few noisy zones might be tolerable as long as the locations of quiet
zones in a room could be moved and adjusted.

Maybe I missed it, but in all the proposals that have been
bandied about here, nobody seems to have suggested "simple"
servo systems: Dealing for the moment only with noise that
gets into the house via transmission through the walls, we
can envision walls covered with active panels. Each uses a
servo system to hold a stationary position. If the interior
of the wall doesn't move, then no sound passes through it.

Of course, to handle high frequencies would require a *lot*
of small servo panels.

OK, I didn't say it would be *easy*. But maybe easier than
dealing with microphone arrays, speaker arrays, head
tracking, phase cancellation, etc, etc, etc.

Once we've dealt with wall transmission, we can deal with
windows. Suppose the window glass is gripped at the edges
(out of sight) by drivers that can apply bending forces.
They could also sense bending to derive the feedback signal,
but we'd probably need to supplement that with something
like an IR reflection scheme to pick up more central
motions.

Lastly, sound entry via ductwork is probably the easiest to
deal with. (As I recall, ducts were the first things to
have ANC applied successfully, decades before headphones.)

Of course, to power all these servos will require your own
electrical generation plant... noisy, but now we know how to
deal with that!

Best regards,



Bob Masta

DAQARTA v6.02
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Scope, Spectrum, Spectrogram, Sound Level Meter
Frequency Counter, FREE Signal Generator
Pitch Track, Pitch-to-MIDI
Science with your sound card!

 

[Marvin the Martian]

On Sat, 17 Sep 2011 08:15:15 -0400, default wrote:

On Fri, 16 Sep 2011 21:12:14 -0500, Marvin the Martian
marvin@ontomars.org> wrote:

On Fri, 16 Sep 2011 15:37:43 -0400, default wrote:

On Fri, 16 Sep 2011 10:46:45 -0500, Marvin the Martian
marvin@ontomars.org> wrote:


IF you think so, give it a go. :-D The problem comes is that you end
up negating the sound very near the speakers, but since the speakers
are emitting sound from a point just behind each speaker in the wall,
but the noise is coming from a point source - the distant jet engines.

What I was thinking is that for it to work you'd pretty much have to
be surrounded by a big cylindrical speaker - or there'd have to be
some complex electronics to check the position of the ears in the room
and adjust the speakers accordingly.

I anxiously await your paper and experimental results.

I'd go for the soundproofed house idea first. Cheaper, doable with
current technology, and added insulation benefits.

Using today's technology to solve this problem is likely to be cost
ineffective. Cheaper to just wear noise canceling ear protection. (for
the privilege of living close to a runway?)

The problem with noise canceling ear protection is...

Oh, what the hell. I said that in the first post.

 

[Sam Wormley]

On 9/17/11 8:00 AM, Marvin the Martian wrote:

The problem with noise canceling ear protection is...

... is that you've never owned a pair that work, and
perhaps don't understand the physic involved. Mine
work beautifully.

 

[jmfbahciv]

krw@att.bizzzzzzzzzzzz wrote:

On 16 Sep 2011 13:16:58 GMT, jmfbahciv <See.above@aol.com> wrote:

John Larkin wrote:
On Fri, 16 Sep 2011 01:47:11 +0000 (UTC), glen herrmannsfeldt
gah@ugcs.caltech.edu> wrote:

In comp.dsp John Larkin <jjlarkin@highnotlandthistechnologypart.com
wrote:

(snip, someone wrote)
There is some otherwise nice real estate at the end of Navy runways,
neighborhoods in San Diego and Virginia Beach where pilots practice
taking off an aircraft carrier with after burners wide open.

A lot of residents would be willing to pay $5,000 or more to be able
to talk to other people in a room at the flip of a switch.

(snip)
Won't work, probably physically impossible. Buy earplugs.

I probably would have said that before hearing (or not) noise
canceling headphones. In this case, if you put the microphones
outside and the speakers inside, and have a reasonable noise
reduction due to insulation, there should be very little feedback
from the speaker to the microphone. I believe that isn't so true
for the headphones, though they still seem to work.

To do it well, though, would take a LOT of speakers and microphones.

Maybe one per square meter of wall and ceiling area. (You get
a quantity discount on the parts.) Calibration would also be
interesting.

With enough DSP, maybe you could get away with fewer microphones,
but still with many speakers, by doing some correlation and delay,
and with the assumption that the sound mostly comes from one,
or a small number of, directions.

-- glen

Well, let me know when you get it working.

Just making the refrigerators shut up would be helpful.

Buy a decent refrigerator. OTOH, the ice falling is loud. ;-)

I have, several times. I think it's the idle ice machinery which
gets loose and rattles but I never checked out my hypothsis.

Hi BAH!

'ey, krw! Miss you in certain newsgroups.

/BAH

 

[jmfbahciv]

glen herrmannsfeldt wrote:

In comp.dsp CWatters <colin.watters@noturnersoakspam.plus.com> wrote:
On 15/09/2011 16:45, Bret Cahill wrote:

The system could monitor the location of the head of each person in
the room.

That's not the main problem.

The difficulty is getting the antinoise signal to the persons head
without it being audible at another persons head.

If you get the noise low at the walls, it should (on average) stay
low throughout the room. That can take a lot of microphones and
speakers, as the frequency increases.

In fact it's worse than that. Even when there is only one person, the
anti noise needed to suppress noise in one ear is a source of noise that
has to be suppressed in the other ear.

At some point, it is an N variables and N unknowns problem.
For N people in 3D, you have 3N variables, so you need at least 3N
speakers to select the phase for. (Well, maybe one less.) More
will get you larger, depending on wavelength, regions around the head.

Unless you figure out how to make sound waves incredibly directional
(for all frequencies) I can't see how it can be done.

If you have a whole wall of microphones outside, and speakers inside,
each adjusted with the appropriate delay (though the wall) and
amplitude (how much leaks through the wall that needs cancelling)
then, for wavelengths somewhat longer than the speaker spacing it
should pretty much cancel throughout the room.

But the other speaking person is part of that noise. If the OP
is a guy, can't he just use his selective hearing software? )))

/BAH

 

[Marvin the Martian]

On Sat, 17 Sep 2011 09:23:54 -0500, Sam Wormley wrote:

On 9/17/11 8:00 AM, Marvin the Martian wrote:
The problem with noise canceling ear protection is...

... is that you've never owned a pair that work, and perhaps don't
understand the physic involved. Mine work beautifully.

Talking out of your ass is what you do best, worm.

 

[Marvin the Martian]

On Sat, 17 Sep 2011 07:51:55 -0700, Bret Cahill wrote:

The problem with noise canceling ear protection is...

   ... is that you've never owned a pair that work, and perhaps don't
   understand the physic involved. Mine work beautifully.

With other people in the room, however, it's anti social in the same way
as lap swimming. They are right next to you yet you cannot talk.

And speaking of swimming, your ears don't seem to have to be very far
under water to almost eliminate noise. A couple of inches is enough.
If the active noise reduction is too difficult maybe water jacketing the
room would be the most cost effective solution. You could still have
sunlight coming in through the windows.

Water transmits sound VERY WELL. Whales and dolphins know this, and you
can too!

Bret Cahill

 

[Bret Cahill]

The problem with noise canceling ear protection is...

   ... is that you've never owned a pair that work, and
   perhaps don't understand the physic involved. Mine
   work beautifully.

With other people in the room, however, it's anti social in the same
way as lap swimming. They are right next to you yet you cannot talk.

And speaking of swimming, your ears don't seem to have to be very far
under water to almost eliminate noise. A couple of inches is
enough. If the active noise reduction is too difficult maybe water
jacketing the room would be the most cost effective solution. You
could still have sunlight coming in through the windows.


Bret Cahill

 

[Sam Wormley]

On 9/17/11 9:47 AM, Marvin the Martian wrote:

On Sat, 17 Sep 2011 09:23:54 -0500, Sam Wormley wrote:

On 9/17/11 8:00 AM, Marvin the Martian wrote:
The problem with noise canceling ear protection is...

... is that you've never owned a pair that work, and perhaps don't
understand the physic involved. Mine work beautifully.

Talking out of your ass is what you do best, worm.

Marvin, you appear to be just plain ignorant about so many things.
And so stubborn about educating yourself about the relevant science.

 

[Sam Wormley]

On 9/17/11 10:11 AM, Marvin the Martian wrote:

On Sat, 17 Sep 2011 07:51:55 -0700, Bret Cahill wrote:

The problem with noise canceling ear protection is...

... is that you've never owned a pair that work, and perhaps don't
understand the physic involved. Mine work beautifully.

With other people in the room, however, it's anti social in the same way
as lap swimming. They are right next to you yet you cannot talk.

And speaking of swimming, your ears don't seem to have to be very far
under water to almost eliminate noise. A couple of inches is enough.
If the active noise reduction is too difficult maybe water jacketing the
room would be the most cost effective solution. You could still have
sunlight coming in through the windows.

Water transmits sound VERY WELL. Whales and dolphins know this, and you
can too!


Bret Cahill

Perhaps, Marvin, you are ignorant about the acoustic impedance
mismatch at the air-water interface. :-o