 |
Description  |
|
|
BACKGROUND OF THE INVENTION
This invention relates generally to sound dampening techniques and more
particularly to methods and apparatus for active noise cancellation.
It is often desirable to reduce the ambient noise level in a particular
environment. This is particularly true when the noise is loud and
unpleasant, such as the noise produced by machinery. In fact, loud noise
can be more than annoying; at certain sustained levels it can cause pain
and permanent injury.
Generally speaking, prolonged exposure to noise levels below about 70
decibels (dB) is perfectly sustainable to most people. When the noise
level is within the range of about 70 to 90 decibels, most people will
begin to experience irritation and stress. Sustained exposure to noise in
the range of 90 to 120 dB can cause permanent hearing loss, and exposure
to noise much in excess of 120 dB can reach the threshold of pain for most
people.
The classical approach to noise reduction is to block the compression wave
generated by the sound source with a sound absorbing substance. This type
of noise reduction is known as passive noise reduction because it does not
require an external energy source to accomplish its task. Examples of
passive noise reduction include standard automobile mufflers, enclosures
for noisy machinery and acoustical ceiling tile. Passive noise reduction
tends to be more effective for high frequency noise than for low frequency
noise.
Another approach to noise reduction is active sound reduction, which refers
to any electro-acoustical method in which an undesired sound wave is
canceled by a second sound wave that has the same amplitude but is
180.degree. out of phase. As shown in FIG. 1a, an undesired tone can be
canceled by generating a second tone of the same amplitude and frequency,
and adjusting its phase so that the peaks of one tone coincide with the
valleys of the other. FIG. 1b illustrates the cancellation of wideband
noise, such as that generated by an automobile, by an appropriately
generated anti-noise. In practice, active noise reduction is most often
used to attenuate low frequency noise and vibration and, therefore, tends
to be complementary with passive noise reduction techniques. It is well
known that active and passive noise reduction methods can be used together
to attenuate a variety of wideband noise sources.
Active noise reduction research dates back at least as far as the 1930's.
In the early days of the research success was limited by the available
technology which essentially consisted of vacuum-tube-based analog
circuitry. Signal processing errors caused by the inherent instability of
the analog circuitry made it difficult to produce the correct anti-noise,
thereby greatly limiting the effectiveness of the noise reduction. The
development of semiconductor-based digital signal processing in the late
1960's provided new tools for analyzing sound waves and allowed sufficient
control over the anti-noise signal to achieve moderate levels of noise
reduction. Virtually all commercially available active noise reduction
equipment is now based upon digital signal processing technology.
Prior art commercial applications of active noise reduction are
concentrated in the areas of headsets and in the quieting of noise in
heating, ventilation and air conditioning (HVAC) ducts. For example,
headsets which utilize the principles of active noise reduction are
manufactured by Bose Company of Framingham, Mass. Devices for quieting
HVAC ducts are made by Digisonix/Nelson Industries of Stoughton, Wis.
The commercial products mentioned above have a number of characteristics in
common. Firstly, all of the commercially available products cancel noise
within an enclosure, chamber or waveguide. In the case of headsets, the
chamber is defined as the volume of air enclosed by the earpieces of the
headsets and the ears of the persons wearing the headsets. In HVAC
applications the noise to be reduced propagates inside of an enclosed
duct. Secondly, all commercially available active noise reduction products
are single-channel devices which operate on sound waves traveling along a
single path. Commercially available products are not, therefore, well
adapted to provide effective noise reduction in environments which support
complex multiple wavefronts, such as within large enclosures or in open
spaces.
There are a great number of patent disclosures describing active noise
cancellation systems. Examples of patents describing active noise
cancellation methodologies for HVAC ducts include: U.S. Pat. Nos.
4,122,303; 4,171,465; 4,473,906; 4,480,333; 4,596,033; 4,665,549;
4,669,122; 4,677,677; 4,783,817; 4,815,139; and 4,837,834. Some of these
patents, such as U.S. Pat. Nos. 4,473,906 and 4,665,549, disclose the use
of multiple input microphones to detect the noise to be canceled. Others
of these patents, such as U.S. Pat. Nos. 4,171,465 and 4,669,122 disclose
multiple speakers used to cancel noise in a duct. U.S. Pat. No. 4,815,139
discloses both the use of multiple input microphones to sense noise and
multiple speakers to cancel noise in a duct. Other examples of active
noise cancellation patents include U.S. Pat. No. 4,637,048 which teaches
the cancellation of noise from an automobile tail pipe, and U.S. Pat. Nos.
4,562,589, 4,689,821 and 4,715,559 which teach the cancellation noise in
the fuselage or cockpit of aircraft.
These patents share the same limiting characteristcs as the above-mentioned
commercial products: they all operate on noise within enclosed spaces such
as ducts or airplane fuselages, and they all disclose signal-channel
cancellation devices. Even the patents which disclose multiple input
microphones and/or multiple output speakers are single-channel devices in
that the signals obtained from the multiple input microphones and the
signals delivered to the multiple speakers are processed within a
single-channel processing device. In consequence, prior art active noise
cancellation devices are not well adapted to the creation of large quiet
zones in open spaces or in large enclosed spaces.
SUMMARY OF THE INVENTION
The present invention includes a method and an apparatus for the active
reduction of complex noise and other compression waves in essentially
unrestricted environments. This is accomplished by a combination of
multi-channel noise reduction techniques coupled with novel signal
processing methods.
The apparatus of the present invention includes a number of microphones
placed within a medium, a multi-channel signal processor, and at least one
speaker or the equivalent placed within the medium to produce
complementary waves that have the same amplitude but opposite phase as the
compression waves to be reduced. For many applications, a number of
speakers are used to produce waves at a variety of locations within the
medium that combine to produce complementary waves which at least
partially cancel the undesired compression waves over a large region of
space known as the "quiet zone."
The signal processor includes a number of forward filters, each of which
has an input coupled to one of the microphones and an output coupled to
one of the speakers. Preferably, each of the speakers is coupled to each
of the microphones by at least one unique forward filter such that the
signal processor is a multi-channel processor having a number of channels
equal to the product of the number of microphones and the number of
speakers.
The apparatus also includes a number of neutralization filters where the
input of each of the neutralization filters is coupled to one of the
inputs to the speakers and where the outputs of the feedback filters are
combined with the input signals from the microphones. The purpose of the
neutralization filters is to compensate for the acoustic feedback that
inevitably occurs whenever speakers and microphones are in close
proximity. Preferably, each of the outputs to the speakers is filtered and
combined with each of the input signals from the microphones so that the
number of neutralization filters equals the number of forward filters.
The method of the present invention includes developing a number of
compression signals from compression waves detected at a number of
locations within a medium, processing the compression signals to develop
at least one complementary signal, and producing at least one
complementary compression wave from the complementary signal. Again, it is
preferable to develop a number of complementary signals and complementary
compression waves in the medium for more effective cancellation of the
compression waves.
An advantage of providing multi-channel noise reduction is that it is far
more effective than the single-channel methods of the prior art at
attenuating noise in unbounded environments or in large enclosed spaces.
This and other advantages of the present invention will become clear to
those skilled in the art upon study of the detailed description of the
invention and of the several figures of the drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1a is a graph illustrating the concept of canceling an undesired first
tone with a second tone which is 180.degree. out of phase with the first
tone, as it is known in the prior art.
FIG. 1b is a graph illustrating the concept of canceling wideband noise
with a 180.degree. out-of-phase anti-noise, as it is known in the prior
art.
FIG. 2a is a pictorial, in-situ representation of an apparatus in
accordance with the present invention.
FIG. 2b is a block diagram of the apparatus and its environment as it is
pictorially illustrated in FIG. 2a.
FIG. 3 is a schematic of a preferred embodiment for a signal processor of
FIGS. 2a and 2b.
FIGS. 4a and 4b are graphs illustrating the noise level at a location
within a desired quiet zone with the apparatus turned OFF and the
apparatus turned ON, respectively.
FIG. 5 is a graph of the signal level versus frequency of the noise with
the apparatus turned OFF and the apparatus turned ON.
FIGS. 6a and 6b are three-dimensional depictions which include the
graphical information of FIGS. 4a and 5 in FIG. 6a and FIGS. 4b and 5 in
FIG. 6b.
FIG. 7 illustrates the use of the apparatus of the present invention to
provide omni-directional noise control in an unbounded medium.
FIG. 8 illustrates the use of the present apparatus to provide directional
noise control in a portion of an unbounded medium.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
FIGS. 1a and 1b illustrate the concept of active noise cancellation as was
discussed in the background section. As used herein, "noise" means any
undesired compression wave produced in any medium, be it solid, liquid, or
gaseous, and in any frequency range, including the sonic, subsonic and
supersonic ranges.
In FIG. 2a, an apparatus 10 in accordance with the present invention is
used to reduce undesired compression waves 12 in a medium 14 produced by a
noise source 16. The apparatus 10 includes a number of input microphones
such as microphones 18a and 18b, a signal processor 20, and a number of
speakers such as speakers 22a, 22b, 22c, and 22d. As used herein, the term
"speaker" means any electro-acoustical transducer, such as a loudspeaker,
a piezoelectric transducer, etc. An error microphone 24 can be used to
detect the effectiveness of the apparatus 10 in reducing the undesired
compression waves in a quiet zone 26 of the medium 14. The error
microphone 24 can be moved to a number of positions 24' to sample the
effectiveness of the apparatus 10 at various angular positions relative to
the noise source 16. Alternatively, a number of error microphones can be
used to simultaneously sample the noise field in the quiet zone.
FIG. 2b illustrates the system of FIG. 2a in a block diagram form. The
fluid medium 14, in this example, is air, and acoustic paths through the
medium 14 are indicated by arrows drawn in a heavy line. Electrical paths
within the apparatus 10 between the input microphones 18a-b, signal
processor 20, and speakers 22a-d are indicated with arrows drawn in a
finer line.
The noise source 16 develops noise wavefronts which travel along a number
of paths such as the acoustic paths 28 and 30. The wavefront along
acoustic path 28 combines with acoustic feedback from speakers 22a-d along
an acoustic path 32 and impinge upon input microphones 18a-b along an
acoustic path 34. The input microphones 18a-b serve as transducers to
convert the compression waves on acoustic path 34 to electrical signals
("compression signals") on lines 36a and 36b. The signal processor 20
processes the electrical signals on lines 36a-b to produce electrical
signals ("complementary signals") on lines 38a, 38b, 38c and 38d. The
speakers 22a-d produce complementary compression waves in medium 14, part
of which are fed back along acoustic path 32 and part of which travel
along an acoustic path 40. The compression waves on acoustic paths 30 and
40 are combined in the fluid medium 14 and travel on an acoustic path 42
to impinge upon error microphone 24.
Referring now to FIG. 3, the signal processor 20 includes a pair of input
summers 42a and 42b, eight forward filters F, four output summers 44a,
44b, 44c, and 44d, and eight neutralization filters N. The two-digit
subscripts of the forward filters F are determined by inputs and outputs
they couple together. For example, forward filter F.sub.11 couples input 1
to output 1 and forward filter F.sub.23 couples input 2 to output 3. In
other words, the first digit of the subscript of the forward filters F
indicates the input number it is attached to and the second digit of the
subscript of the forward filters indicates which output it is coupled to.
In a similar fashion, the eight neutralization filters have two-digit
subscripts where the first digit indicates which input it is coupled to
and the second digit indicates which output it is coupled to. The signal
processor 20 further includes a pair of input buffers 43a and 43b coupling
lines 36a and 36b to summers 42a and 42b, respectively, and four output
buffers 45a, 45b, 45c, and 45d coupling the outputs of summers 44a-44d to
lines 38a-38d, respectively.
In the forward path of signal processor 20, the inputs 1 and 2 are
processed within summers 42a and 42b, respectively, and the output of
summers 42a and 42b are each applied to the inputs of four forward filters
F. The output of summer 42a on a line 46a is applied to the inputs of
forward filters F.sub.11, F.sub.12, F.sub.13, and F.sub.14. Similarly, the
output of summer 42b on a line 46b is applied to the inputs of the forward
filters F.sub.21, F.sub.22, F.sub.23, and F.sub.24. The outputs of the
forward filters F are applied to the inputs of summers 44a-d in the
following fashion: the outputs of filters F.sub.11, and F.sub.21 are
applied to summer 44a, the outputs of filters F.sub.12 and F.sub.22 are
applied to summer 44b, the outputs of filter F.sub.13 and F.sub.23 are
applied to summer 44c, and the outputs of filters F.sub.14 and F.sub.24
are coupled to the inputs of the summer 44d. The outputs of the summers
44a-d are coupled to the lines 38a-38d by the output buffers 45a-d,
respectively.
In a reverse or feedback path, the output signals 1-4 are fed back through
neutralization filters N to the summers 42a and 42b. More specifically,
neutralization filters N.sub.11, N.sub.12, N.sub.13, and N.sub.14 feed
back the signals from outputs 1-4 to the summer 42a and neutralization
filters N.sub.21, N.sub.22, N.sub.23, and N.sub.24, feed back the signals
from outputs 1-4 to the summer 42b.
The filters F and N can be made from discrete components such as inductors,
capacitors and resistors. Preferably, however, the filters F and N are
digital filters and part of a digital signal processing apparatus 20. The
best mode currently known for practicing this invention utilizes a
mini-computer, such as a VAX 3600 mini-computer from Digital Equipment
Corporation, and digital signal processing (DSP) boards which plug into
bus slots provided in the mini-computer. A typical DSP board uses
commercially available DSP integrated circuits such as I.C. part DSP-32 of
AT&T, Inc. or I.C. part number 56000 of Motorola, Inc. The architecture of
a suitable DSP board is described in a paper entitled "A Real-Time,
Multichannel System with Parallel Digital Signal Processors" by William A.
Weeks and Brian L. Curless, published in the Proceedings of the 1990
International Conference on Acoustics, Speech, and Signal Processing
(ICASSP 90), Albuquerque, N. Mex., Apr. 3-6, 1990. Alternatively, a less
powerful system uses a personal computer such as a Macintosh II personal
computer available from Apple Computers, Inc. of Cupertino, Calif.
equipped with commercially available DSP boards from such vendors as
Spectral Innovations, Inc. of Santa Clara, Calif.
In a digital signal processing system 20, the input buffers 43a-b include
analog-to-digital (A/D) converters which convert the analog signals
produced by the input transducers on lines 36a-b into digital inputs 1 and
2, respectively. As is well known to those skilled in the art, the input
buffers can also include pre-amplifiers, anti-aliasing (low-pass) filters,
etc. Lines 46a and 46b couple the digital sum calculated by the digital
summers 42a and 42b to the digital forward filters F. The outputs of the
digital forward filters are coupled to the inputs of the digital summers
44a-d to produce digital outputs 1-4. The output buffers 45a-d include
digital-to-analog (D/A) converters to convert the digital outputs 1-4 to
the analog signals on line 38a-38d to drive the output transducers. As is
also well known to those skilled in the art, the output buffers can
include reconstruction filters, power amplifiers, etc. The digital outputs
on output 1-4 are fed-back through digital neutralization filters N to
produce digital inputs for digital summers 42a and 42b.
The method of computing the "weights" of the forward filters F and
neutralization filters N will be described with reference to FIG. 2a. The
error microphone 24 produces an error signal E having an amplitude which
is directly related to the amount of uncancelled noise at that location.
The object, therefore, is to minimize the amplitude of the error signal E
by adjusting the weights of the forward filters F and neutralization
filters N so as to produce the most effective anti-noise. The filter
weights can be adjusted by a variety of methods well known to those
skilled in the art, such as the Wiener least-squares minimization method
as taught in Optimum Signal Processing, An Introduction, by S. J.
Orfanidis, Macmillan Publishing Company, 1988, or the Widrow-Hoff
algorithm as taught in Adaptive Signal Processing, by B. Widrow and S.
Stearns, Prentice-Hall, Inc., 1985.
Once the noise at error microphone 24 has been minimized, the microphone
can be moved to a variety of locations 24' to detect the effectiveness of
sound cancellation at those locations. The filters can then have their
weights further adjusted to, for example, minimize the average
simultaneous noise power at all of the tested locations.
It should be noted that the apparatus 10 will work in a number of
environments and mediums. For example, the apparatus 10 can be used to
reduce compression waves within a liquid medium for such purposes as
underwater noise cancellation to aid in the sonic exploration of the
oceans. As another example, the apparatus 10 can be used to selectively
cancel seismic waves propagating through the earth's crust so that other
compression wave activity in the earth's crust can be monitored more
sensitively. Of course, the input and output transducers of the apparatus
10 are chosen to be suitable for the environment that they will be
subjected to. For example, in a liquid medium where both the input and
output transducers are immersed in a liquid the transducer should be
waterproof and relatively inert to that liquid. Of course, if one of the
transducers, such as the output transducer, is outside of the liquid
medium, this would not be a concern. In a solid medium the input
transducer might be a vibration sensor such as piezoelectric crystal or
magnetic coil detector while the output transducers might be
vibration-creating elements such as electrical, pneumatic, or hydraulic
rams or solenoids.
In FIGS. 4a and 4b, plots of the amplitude versus time function of the
error signal E are shown. In FIG. 4a the apparatus 10 is turned OFF and
the error signal E represents the arbitrary noise to be canceled. In FIG.
4b the apparatus 10 is turned ON and the error signal E indicates that the
undesired noise is quickly and substantially reduced. Under typical
conditions, the apparatus 10 of the present invention has reduced the
noise level by as much as 30 dB in a small fraction of a second.
FIG. 5 illustrates the frequency-dependent behavior of the noise reduction
method of the present invention. In this graph, the amplitudes of the
spectral components of error signal E are taken at a particular point in
space. The frequency-dependent error signal E developed when the apparatus
10 is OFF is shown with a solid line and is the spectrum of the waveform
shown in FIG. 4a. The frequency-dependent error signal E developed when
the apparatus 10 is ON is shown with a broken line and represents the
spectrum of the waveform shown in FIG. 4b. At each frequency, the
difference betwee | | |
