United States Patent [19J
`Bartlett et al.
`
`[54] NOISE-CANCELING DIFFERENTIAL
`MICROPHONE ASSEMBLY
`
`[75]
`
`Inventors: Charles S. Bartlett, Clinton, Md.;
`Michael A. Zuniga, Fairfax, Va.
`
`[73] Assignee: AT&T Corp., Murray Hill, N.J.
`
`[21] Appl. No.: 230,.955
`
`[22] Filed:
`
`Apr. 21, 1994
`
`Int. Cl.6
`[51]
`.......................•.............................. H04M 1/00
`[52] U.S. Cl. .......................... 379/387; 379/389; 379/390;
`379/433; 381/71; 381/168; 381/189
`[58] Field of Search ..................................... 379/387, 433,
`379/432, 390, 395, 389; 381/71, 168, 169,
`189
`
`[56]
`
`References Cited
`
`U.S. PATENT DOCUMENTS
`
`4,584,702
`4,773,091
`4,850,016
`5,239,578
`5,260,998
`5,341,420
`
`4/1986 Walker, Jr ................................. 38lnl
`9/1988 Busche et al. . ......................... 379/433
`7 /1989 Groves et al. . ......................... 379/433
`8/1993 Regen et al ............................. 379/387
`9/1993 Takagi ..................................... 379/433
`8/1994 Hollier et al.
`.......................... 379/433
`
`I 1111111111111111 11111 lllll 111111111111111 111111111111111 IIIIII Ill lllll llll
`5,473,684
`Dec. 5, 1995
`
`US005473684A
`[11] Patent Number:
`[45] Date of Patent:
`
`5,381,473
`
`1/1995 Andrea et al. .......................... 379/433
`
`OTHER PUBLICATIONS
`
`"Second-Order Gradient Noise-Cancelling Microphone,"
`A. J. Brouns, CH1610-5/81/0000-0786 1981 IEEE, pp.
`786-789.
`"A Second-Order Gradient Noise Canceling Microphone
`Using a Single Diaphragm," W. A. Beaverson and A. M.
`Wiggins, Journal of the Acoustical Society of America, vol.
`22, No. 5, Sep. 1950, pp. 592-601.
`"Adaptive Noise Cancelling: Principles and Applications,"
`B. Widrow et al., Proc. IEEE 63 (Dec. 1975), pp.
`1692-1716.
`
`Primary Examiner-Jeffery A. Hofsass
`Assistant Examiner-Jacques M. Saint-Surin
`Attorney, Agent, or Firm-Martin I. Finston
`
`[57]
`
`ABSTRACT
`
`Improved microphone performance is achieved by config(cid:173)
`uring a second-order derivative microphone assembly in
`such a way that radially divergent near-field input produces
`a microphone response proportional to a first-order spatial
`derivative of the acoustic pressure field.
`
`20 Claims, 8 Drawing Sheets
`
`r:----, r-----------------,
`130
`I
`I
`I
`50
`I
`I
`._~1 __ 1 - - - - - - - .
`I
`I
`I
`I
`I
`I
`I
`I
`I
`: 32
`:
`:
`I
`I
`I
`......... 1 _ _ 1 _ _ ......i
`FILTER
`I
`I
`I
`1
`I
`I
`I
`60
`I
`L---l~ L--l~----------
`__ _J
`SOD OUTPUT
`
`- 1 -
`
`Amazon v. Jawbone
`U.S. Patent 8,019,091
`Amazon Ex. 1006
`
`

`

`U.S. Patent
`
`Dec. 5, 1995
`
`Sheet 1 of 8
`
`5,473,684
`
`........ - - -~G
`
`FIG. 1
`
`FIG. 2
`
`FIG. 3
`
`- 2 -
`
`

`

`U.S. Patent
`
`Dec. 5, 1995
`
`Sheet 2 of 8
`
`5,473,684
`
`FIG. 4
`
`y
`
`30
`
`d/2
`
`X
`
`96
`
`r+ -----
`--
`-
`-
`-
`e
`--
`--- ------r- --
`e
`
`98
`
`a
`
`FIG. 5
`
`0 . - - - - - - - - - - - - - - - - - - - - ,
`
`•.J..'!...'_,:. ~-. ' ·. ' . ' ·.
`' . ' ·.
`' . ' ·.
`' . ' ..
`', ··.
`' . ' •,
`', · ..
`' ..
`', •,.
`'
`•,
`', ·•.
`', • ..
`..
`..
`'
`..
`'.... . ..
`'
`'
`..
`..
`', ...
`'
`..
`. ..
`·• .
`..
`....
`............. -
`~ -
`-... .....
`.............. _ .....
`
`'
`
`..........
`
`.....
`
`....
`
`-5
`
`-10
`
`-15
`
`-20
`
`-25'------'----'----l------'---....,__ _ __,
`5
`6
`0
`2
`4
`3
`
`- 3 -
`
`

`

`U.S. Patent
`
`FIG. 6
`
`Dec. 5, 1995
`
`Sheet 3 of 8
`
`5,473,684
`
`60
`
`-----/120
`ADAPTIVE
`ALGORITHM
`
`ERROR
`SIGNAL
`
`130
`
`SILENCE
`DETECTOR
`
`FIG.
`
`7
`
`-5
`
`-10
`
`-cc
`"'O -L..J
`
`c:::l => ._
`·z
`c.::>
`<
`2
`
`-15-.__........___....__.1..-.J...-l.-l....L..----L..----L----l--l......L.....J........J.....L....L.J...J.J
`
`104
`
`300
`
`103
`FREQUENCY (Hz)
`
`- 4 -
`
`

`

`U.S. Patent
`
`Dec. 5, 1995
`
`Sheet 4 of 8
`
`5,473,684
`
`....-----------------------. "'b
`
`-
`
`--(cid:173)N ::c -
`
`0
`c--.,
`
`0
`
`c:,
`
`c:o
`
`~
`E;::
`
`0
`I
`
`-
`(BP) 30nllN~\fN
`
`C)
`c-,.i
`I
`
`,.,.,
`
`0
`I
`
`c--.,
`0
`
`0
`~
`I
`
`- 5 -
`
`

`

`U.S. Patent
`
`Dec. 5, 1995
`
`Sheet 5 of 8
`
`5,473,684
`
`....,
`cc::
`c::c
`2
`< ::c
`.. u
`- - - - - - - - - - - - - - - - - - - - - - - - ~ ~u
`25
`::c u
`Lw
`:z:
`<
`
`-N ::c ->(cid:173)u
`
`.,, :z
`C)Lw
`-::>
`C,
`'"""
`cc::
`'"'-
`
`0 -I
`
`0
`C',J
`I
`
`0 ....,
`
`I
`
`0 .. I
`
`0
`U")
`I
`(BP) 3Qnl1N~vn
`
`0
`
`"° I
`
`0 ,._
`I
`
`0 co
`I
`
`0 en
`I
`
`- 6 -
`
`

`

`.a;;...
`00
`"' ~
`tu
`""-l
`"' .a;;...
`Ul
`
`00
`
`s,
`="
`~
`00 ;
`
`lll
`1,0
`1,0
`1-4
`...,lll
`ti, r
`~
`
`t't) = ~
`
`~
`~
`~
`rJ1 •
`~ •
`
`REVERBERANT ROOM
`
`FREQUENCY (Hz)
`
`103
`
`104
`-40'-----'-----'--'--L......I..-..L.--'--1._..l.,__ __ _t__-.L.1'--4.L..L..-L.-L-.L....LI
`
`102
`
`::::IE
`...:
`
`c::l
`1...1
`
`:::> -ts -20
`
`20,--------=-=-~=.:....:.:.==---.:.::.:::.:::.-:~~-~
`
`NEAR Fl£LD SENSITIVITY or SOD MICROPHONE VERSUS DISTANCE
`
`1 cm
`
`~ 0
`>
`
`-a
`
`, I
`
`104
`
`I
`I
`
`I
`
`I
`
`I
`
`2.5Cffl .-~ I
`
`103
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`102
`
`__-
`
`-40
`
`,../~ ~ .,,,,-
`
`1 cm
`
`-20
`
`<.!)
`
`a>
`
`N[AR rt[LD SENSITIVITY or roD MICROPHONE VERSUS DISTANCE
`
`10
`
`FIG.
`
`~ ~
`;
`:::::>
`"'0 -1...1
`c::l
`-ol
`20,
`
`- 7 -
`
`

`

`U.S. Patent
`U.S. Patent
`
`Dec. 5, 1995
`Dec. 5, 1995
`
`Sheet 7 of 8
`Sheet 7 of 8
`
`5,473,684
`5,473,684
`
`11
`
`FIG.
`
`
`
`- 8 -
`
`

`

`U.S. Patent
`
`Dec. 5, 1995
`
`Sheet 8 of 8
`
`5,473,684
`
`DIRECTION or
`/
`RECEPTION
`✓ EXISTING TYPE K
`OUTLINE
`
`\r165
`
`0
`
`FIG.
`
`12
`
`SECOND ORDER
`MICROPHONE
`
`MICROPHONE
`ELECTRONICS
`
`FIG.
`
`13
`
`- 9 -
`
`

`

`5,473,684
`
`1
`NOISE-CANCELING DIFFERENTIAL
`MICROPHONE ASSEMBLY
`
`BACKGROUND OF THE INVENTION
`
`2
`lips. Specifically, the voice response is expressible as the
`sum of three terms: a frequency-independent term inversely
`proportional to R3, a term proportional to the angular
`frequency co and inversely proportional to R2
`, and a term
`5 proportional to ro2 and inversely proportional to R. That is,
`with increasing distance from the lips, prior art SOD micro(cid:173)
`phones very soon exhibit an undesirable, ro2 component of
`the near-field voice response. This effect tends to reduce the
`net transmitted voice power, and to make the voice sound
`10 deficient in low frequencies.
`
`SUMMARY OF THE INVENTION
`
`Telephone handsets usually are fitted with omnidirec(cid:173)
`tional microphones, which offer little discrimination against
`background noise. As a consequence, noise may be trans(cid:173)
`mitted together with the speaker's voice, and interfere with
`the far end party's ability to understand what is being said.
`Noise cancelling microphones have been proposed as a
`possible solution to this problem. These microphones, some(cid:173)
`times referred to as pressure gradient or first-order differ(cid:173)
`ential (FOD) microphones, have a vibratable diaphragm
`which is acted upon by the difference in sound pressure 15
`between the front and back sides of the diaphragm. An
`electrical signal is thus produced which is proportional to the
`gradient in the sound pressure field at the microphone. At the
`telephone mouthpiece, the acoustic field due to ambient
`noise will generally have a smaller pressure gradient than 20
`the acoustic field due to the speaker's voice. As a conse(cid:173)
`quence, the voice will be preferentially sensed and trans(cid:173)
`mitted relative to the ambient noise.
`U.S. Pat. Nos. 4,584,702, 4,773,091 and 4,850,016 25
`describe designs for incorporating a pressure gradient micro-
`phone into a telephone handset. Although useful, these
`designs fail to take into account all of the acoustic spatial
`information that might be used to enhance the speaker's
`voice relative to ambient noise. Moreover, the frequency 30
`response of pressure gradient microphones generally causes
`a change in the frequency content of the transmitted voice
`that becomes more noticeable as the distance from the
`speaker's mouth is increased. This tendency is readily offset
`by electronic frequency shaping. However, the use of elec-
`tronic frequency shaping tends to partially counteract the
`ability of the microphone to reject noise. Thus, if frequency
`shaping is to be used, it is desirable to have a microphone
`with improved noise-rejection characteristics so that some
`marginal loss of performance can be tolerated.
`In order to achieve still better noise rejection, practitioners
`in the microphonic art have proposed the use of second order
`differential (SOD)· microphones, which measure a spatial
`second derivative of the acoustic pressure field. A ratio can
`be taken of two such second derivatives, the numerator
`corresponding to the speaker's voice (near the lips), and the
`denominator corresponding to the ambient noise field. Gen(cid:173)
`erally, this ratio will be significantly greater than the analo(cid:173)
`gous ratio of first derivatives (such as would characterize the
`performance of a FOD microphone). Consequently, a SOD
`microphone is expected to exhibit much greater sensitivity
`to a speaker's voice relative to ambient noise than a FOD
`microphone.
`SOD microphone designs have been described, for
`example, in A. J. Brouns, "Second-Order Gradient Noise- 55
`Cancelling Microphone," IEEE International Conference on
`Acoustics, Speech, and Signal Processing CH1610-5/81
`(May 1981) 786-789, and in W. A. Beaverson and A. M.
`Wiggins, "A Second-Order Gradient Noise Canceling
`Microphone Using a Single Diaphragm," J. Acoust. Soc.
`Am. 22 (1950) 592-601. In general, these designs are
`configured to measure a second order derivative of the
`acoustic field near the speaker's lips, but they do not
`optimally exploit the spherical wave nature of the speaker's
`voice field to maximize sensitivity to the speaker's voice. As 65
`a consequence, the voice response of prior art SOD micro(cid:173)
`phones is very sensitive to the distance R from the speaker's
`
`35
`
`40
`
`We have invented an improved SOD microphone that
`overcomes the deficiencies of the prior art by responding to
`the speaker's voice like a FOD microphone (i.e., with
`respect to the dependence of voice response on distance
`from the speaker's lips), but responding to the far field noise
`like a SOD microphone. As a consequence, the inventive
`microphone is significantly less sensitive than prior art SOD
`microphones to positioning with respect to the speaker's
`lips, and it gives better voice quality than prior art SOD
`microphones while maintaining comparable far-field noise
`discrimination.
`A further advantage of our microphone is that it can be
`provided in a compact design that is readily incorporated in
`the mouthpiece of a telephone handset with little modifica(cid:173)
`tion to the existing structure. This incorporation can be
`achieved in such a way that only the acoustic field exterior
`to the mouthpiece is sensed to any significant degree, and
`diffraction and wind-noise effects are no greater than with
`conventional microphones.
`In a broad sense, the invention involves apparatus com(cid:173)
`prising a transducer for converting acoustic signals, emitted
`by a source, to electrical output signals in the presence of
`acoustic noise; and further comprising a platform for main(cid:173)
`taining the transducer at a substantially constant distance
`from the source along an axis to be referred to as the major
`axis. The transducer is adapted to respond to a second spatial
`derivative of the acoustic pressure field. Accordingly, it is
`referred to herein as a "second-order differential micro-
`phone," or "SOD."
`·
`The operation of the invention is conveniently described
`45 with reference to a minor axis perpendicular to the major
`axis. The transducer includes means for sensing the pressure
`field at respective first and second locations that are sepa(cid:173)
`rated at least by a displacement along the minor axis. The
`first sensing means produce a first output signal proportional
`to the first spatial derivative of the pressure field, along the
`minor axis, at the first location. Similarly, the second sensing
`means produce a second output signal proportional to the
`first spatial derivative of the pressure field, along the minor
`axis, at the second location.
`The transducer further comprises means for combining
`the first and second output signals into a net output signal
`which represents the difference between the first and second
`output signals. The first and second locations are chosen in
`such a way that radially divergent acoustic signals emitted
`60 by the source will contribute first and second output signals
`that mutually reinforce in the net output signal, and are
`proportional to the first spatial derivative of the pressure
`field along the minor axis.
`In one embodiment, the transducer comprises an array of
`two FOD microphones, separated by a distance d along the
`minor axis. The FOD microphones are situated within the
`platform in such a way that in use, they will be on the same
`
`50
`
`- 10 -
`
`

`

`5,473,684
`
`3
`side of the source and approximately equidistant from it. The
`membrane of each FOD microphone is substantially per(cid:173)
`pendicular to the minor axis, so that each microphone is
`individually sensitive to the first spatial derivative of the
`pressure field along the minor axis.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`5
`
`FIG. 1 is a block diagram of a microphone assembly and
`associated electronics according to the invention in one lO
`embodiment.
`FIG. 2 is a partially cut-away view of a conventional first
`order differential microphone useful as a component of the
`inventive microphone assembly, in one embodiment.
`FIG. 3 is a perspective view of a second order differential
`microphone assembly according to the invention, in one
`embodiment.
`FIG. 4 is a schematic diagram showing a preferred
`geometric arrangement of the second order differential
`microphone assembly relative to a near field source such as
`a user's mouth.
`FIG. 5 is a theoretical plot showing how the output of a
`SOD assembly depends on the separation between the two
`FOD microphones. The vertical axis represents the magni- 25
`tude of the SOD output, relative to the output of a conven(cid:173)
`tionally oriented FOD microphone.
`FIG. 6 is a block diagram of the invention in one
`embodiment, including a component for adaptively equal(cid:173)
`izing the two FOD microphones.
`FIG. 7 is a graph of the frequency response of a typical
`voice-shaping filter useful for matching the near-field output
`of the inventive second-order differential microphone to that
`of an omnidirectional microphone.
`FIG. 8 is a graph that compares the near-field sensitivities
`of SOD, FOD, and omnidirectional microphones. Plotted is
`the frequency response of a microphone of each type.
`FIG. 9 is a graph that compares the far-field sensitivities
`of the microphones of FIG. 8.
`FIG. 10 is a graph that compares the FOD and SOD
`microphones with respect to the effect of source-to-micro(cid:173)
`phone separation on near-field sensitivity. Plotted for a
`microphone of each type is the frequency response at two
`different separations from the source.
`FIG. 11 is a perspective view of a cellular telephone
`handset incorporating a SOD microphone assembly accord(cid:173)
`ing to the invention in one embodiment.
`FIG. 12 is a perspective view of a telephone handset
`incorporating a SOD microphone assembly according to the
`invention in one embodiment.
`FIG. 13 is a perspective view of a headset including a
`boom that incorporates a SOD microphone assembly
`according to the invention in one embodiment.
`
`30
`
`35
`
`4
`The purpose of element 40 is to assure, within practical
`limits, that the microphones respond to identical acoustic
`input with identical sensitivity and identical phase response.
`Generally, a variable-gain amplifier will suffice of the
`respective microphone responses differ by a gain offset that
`is independent of frequency in the range of interest. An
`active filter may be desirable if the imbalance in gain or
`phase-response is frequency-dependent.
`In some cases, changing environmental conditions such as
`temperature and humidity may affect the respective micro(cid:173)
`phone responses unequally. In such cases, it may be advan(cid:173)
`tageous for element 40 to perform its balancing function by
`adaptive filtering, as described in greater detail below.
`It should be noted in this regard that for microphones 30
`15 and 32 we currently prefer to use mass-produced, commer(cid:173)
`cially available, electret microphones. These microphones
`are readily available in large lots that are well matched in
`phase response and that have a fixed gain offset, independent
`of frequency (within the range of interest) that is typically
`20 within ±4 dB.
`The output of differencing amplifier 50 is optionally
`passed through voice-shaping electronic filter 60 to make the
`frequency response of the microphone assembly match a
`desired characteristic, such as the frequency response of an
`omnidirectional microphone.
`Turning to FIG. 2, exemplary FOD electret microphone
`34 includes diaphragm 35 enclosed between front cover 37
`and back cover 38. Both sides of the diaphragm sense the
`acoustic pressure field. This is facilitated by air holes 36 in
`the front and back covers. Typically, a felt cover layer 39
`overlies front cover 37.
`As is well-known in the art, this bilateral sensing results
`in an output proportional to the acoustic velocity field
`perpendicular to the face of the diaphragm.
`Turning to FIG. 3, each of microphones 30 and 32 of
`microphone assembly 10 is mounted within a respective one
`of cartridges 80, 82, each of which defines a partial enclo(cid:173)
`sure having ports 70 for admitting the acoustic pressure field
`so that it can be sensed on both sides of the microphone
`40 diaphragm. Each cartridge is made from a material of
`sufficient rigidity to substantially isolate the enclosed vol(cid:173)
`ume sensed by the microphone from the surrounding mouth(cid:173)
`piece cavity of the handset (or other platform), and from
`vibrations transmitted through the platform. By way of
`45 example, we have successfully used cartridges that, with the
`ports sealed, will reduce the detected sound level by about
`20 dB or more.
`We have built working prototypes in which the cartridges
`are made from brass, 0.008 inch (0.20 mm) thick. However,
`for a mass-produced embodiment of the invention, we
`believe it would be preferable, for economic reasons, to use
`a polymeric material such as hard plastic or hard rubber.
`It is desirable to subdivide the interior of the cartridge into
`55 a pair of mutually acoustically isolated chambers, one in
`front of the microphone, and one behind it. This is achieved,
`for example, by surrounding the microphone with a mount(cid:173)
`ing ring 90 that fits tightly against the microphone body and
`also seals tightly against the inner wall of the cartridge. Such
`60 a mounting ring is exemplarily made from rubber.
`In addition, a layer 110 of an appropriate foam material,
`such as an open-cell, 65-pore, polyurethane foam, is usefully
`interposed between the acoustic signal source and each of
`the microphones for reducing pickup of acoustic turbulence
`when a human operator speaks into the microphone assem(cid:173)
`bly. Such a foam layer is advantageously placed over each
`port 70. Alternatively, foam bodies can be placed, e.g.,
`
`50
`
`DETAILED DESCRIPTION
`
`Turning to FIG. 1, the invention in a preferred embodi(cid:173)
`ment includes a microphone assembly 10 and electronics
`package 20. Within the electronics package, the outputs of
`first-order differential (FOD) microphones 30, 32 are sub(cid:173)
`tracted, one from the other, by differencing amplifier 50.
`Before this subtraction, the output of, e.g., FOD microphone
`32 is advantageously passed through element 40 to equalize 65
`the outputs of the two FOD microphones. This element may
`be, for example, a variable-gain amplifier or a filter.
`
`- 11 -
`
`

`

`5,473,684
`
`6
`a mutually reinforcing combination of the respective micro(cid:173)
`phone responses to these velocity-field components. That is,
`the magnitudes of the velocity-field components directed
`along the minor axis will add constructively in the combined
`5 microphone output.
`The magnitude of the output from the inventive SOD
`microphone assembly can be compared with the magnitude
`of the output from a single FOD microphone situated at
`point 94, with its diaphragm oriented perpendicular to the
`10 . major axis. Theoretically, the SOD:FOD ratio of these
`magnitudes (when the major and minor axes are mutually
`perpendicular) is given by the expression:
`
`Ratio= (
`
`112
`
`)
`
`5
`directly within the ports, or within the cartridge.
`In currently preferred embodiments, each cartridge 80, 82
`is conformed as a right circular cylinder having a lengthwise
`slot along the top (i.e., along the side intended to face the
`speaker or other source of acoustic signals) that extends
`approximately the full length of the cartridge. The micro(cid:173)
`phone and its mounting are placed within the cartridge in
`such a way as to form a partition that subdivides the interior
`of the cartridge into two approximately equal parts 112, 114.
`The microphone, together with its mounting, also subdivides
`the lengthwise slot, and in that way defines each port 70 as
`a subdivided part of the slot.
`Electrical leads 116 from each microphone are readily
`directed out of each cartridge through a small hole 118
`formed, e.g., in the side of the cartridge opposite the ports. 15
`Within each of these hobs, the space surrounding the leads
`is desirably filled with an airtight seal in order to reduce
`acoustical leakage.
`By way of example, we have successfully used cartridges
`of 0.008-inch brass that are shaped as cylinders 0.55 inch 20
`long and 0.38 inch in inner diameter. The corresponding
`microphones were electric microphones 0.38 inch in diam(cid:173)
`eter, and having a front-to-back length of 0.2 inch.
`In at least some embodiments, it is advantageous to place
`the two cartridges end-to-end in contact or at least in close
`juxtaposition. This arrangement, together with symmetric
`sizes and placement of the ports in each cartridge, helps to
`ensure that within each cartridge, the two sides of the
`corresponding diaphragm will sample the acoustic field
`equivalently. As a result, microphone assembly 10 can
`exhibit second-order differential microphone behavior simi(cid:173)
`lar to that which would result if FOD microphones 30, 32
`.were operated in a free-field environment.
`It should be noted that because microphone assembly 10
`can be acoustically isolated from the platform upon which it
`is mounted, it is easily adapted to a variety of application
`environments.
`It is well known that for many, typical microphonic
`applications, a speaker's voice behaves to a significant
`extent as though emanating from a theoretical point source.
`Accordingly, the geometric arrangement of FIG. 4 is the
`currently preferred arrangement when microphone assembly
`10 is mounted on a communication platform for operation
`by a human speaker.
`As shown in FIG. 4, FOD microphones 30, 32 are
`separated by a distance d along line segment 92, which will
`be referred to as the "minor axis." The midpoint 94 of the
`minor axis is a distance a from theoretical point source 96
`along axis 98, which Will be referred to as the "major axis."
`The FOD microphones preferably lie at equal radii r+, r(cid:173)
`from source 96. It is clear that if the microphones are
`equidistant from the source, the major and minor axes are
`perpendicular, and each of radii r+ and r- makes an angle 0
`with the major axis. Significantly, the angle 0 is less than
`90°; i.e., both microphones lie on the same side of the
`source.
`The membrane of each microphone is oriented substan(cid:173)
`tially perpendicular to the minor axis (i.e., perpendicular to
`the y-axis in the representation of FIG. 4). As a result, each
`membrane is sensitive (within practical limits) only to that
`component of the acoustic velocity field directed along the
`minor axis.
`Because the acoustic signal is radially divergent, this
`velocity-field component has one sign at r+, and the opposite
`sign at r-. Consequently, the difference between the electri(cid:173)
`cal outputs of the respective FOD microphones will include
`
`25
`
`35
`
`where z=d/(2a), k=m/c, co to is the angular frequency and c
`is the speed of sound in air.
`This function is plotted in FIG. 5 versus the normalized
`distance d/(2a), for frequencies of 500 Hz, 1000 Hz, and
`1500 Hz. It is evident that to an excellent approximation
`(provided the distance a, in units of one wavelength at the
`given frequency, is much smaller than ½it), the SOD output
`is maximal (relative to the hypothetical FOD output), with(cid:173)
`out regard to frequency, when d=l.4a. This design formula
`may be used to optimize the geometric configuration of the
`microphone assembly for a particular communications
`device based upon the expected distance between the assem-
`30 bly and the user's lips.
`It is evident from FIG. 5 that when microphone assembly
`10 is optimally configured as described, its sensitivity in the
`near field is nearly as large as a FOD microphone placed at
`the midpoint of the minor axis. However, the far-field noise
`rejection of the SOD will be substantially enhanced relative
`to the FOD microphone because the second spatial deriva(cid:173)
`tive of the acoustic pressure field is smaller than its first
`spatial derivative in typical diffuse sound environments.
`Thus, a significant improvement in microphone perfor-
`40 mance is achieved by configuring a microphone assembly in
`such a way that when them is radially divergent near-field
`input, the corresponding component of the second-order
`differential output is proportional to the first-order spatial
`derivative. The configuration of FIG. 4 is a currently pre-
`45 ferred configuration. However, the invention as we envisage
`it is meant to encompass other microphone configurations
`that also embody this conceptual approach to improving
`microphone performance. For example, we believe that as
`an alternative to the two-diaphragm implementation of FIG,
`50 4, it is possible to achieve qualitatively similar improve(cid:173)
`ments in microphone performance through the use of a
`single-diaphragm configuration. The use of single dia(cid:173)
`phragms to achieve first- and second-order differential
`operation is described generally in the journal article by
`55 Beaverson and Wiggins, cited above.
`As noted, it is advantageous to include a balancing
`element such as a variable gain amplifier to equalize the
`outputs of microphones 30, 32. The appropriate setting of,
`e.g., a variable gain may be determined during production.
`60 Alternatively, an adaptive algorithm can be used to balance
`the microphones adaptively, each time they are used. An
`appropriate algorithm for this purpose is described, for
`example, in B. Widrow et at., "Adaptive Noise Cancelling:
`Principles and Applications," Proc. IEEE 63 (December
`65 1975) 1692-1716.
`Turning to FIG. 6, adaptive algorithm 120 is implemented
`by, for example, a digital signal processor whose output is
`
`- 12 -
`
`

`

`5,473,684
`
`10
`
`25
`
`7
`used, in effect, to adjust variable gain or adaptive filter 125
`in order to minimize the (subtractively) combined outputs of
`microphones 30, 32.
`It is desirable to adapt only when them is no speech
`present; i.e., when the human speaker (or other acoustic 5
`signal source) is quiescent. Accordingly, a component 130 is
`advantageously provided for detecting when the signal
`source is quiescent. Such a component is also exemplarily
`implemented by a digital signal processor. The output of
`component 130 sets the adaptation parameter to a non-zero
`value (i.e., permits adaptation to take place) when the output
`of, e.g., microphone 32 is less than a predetermined thresh(cid:173)
`old, which is taken as an indication that the microphone is
`receiving only far field noise. By way of example, we have
`found that the speaker's sound level at the microphones is
`typically about 10-15 dB higher than that due to the ambient 15
`noise level. Thus, a threshold can (at least initially) be set
`with reference to this quantity.
`The near-field frequency response of SOD microphone
`assembly 10 is nearly the same as that of a FOD microphone,
`even without the use of a voice-shaping filter. Relative to an 20
`omnidirectional microphone, however,
`the near-field
`response of the SOD assembly increases with frequency.
`This frequency dependence is readily compensated using
`voice-shaping filter 60. A typical curve of frequency
`response for such a filter is shown in FIG. 7.
`The sensitivity plots of FIGS. 8 and 9 represent the
`near-field (FIG. 8) and far-field (FIG. 9) responses of the
`respective microphones relative to a calibrated Brue! and
`Kjaer microphone placed 1 cm from the lips of a Brue! and
`Kjaer head and torso simulator. The simulator was the 30
`acoustic source in each case.
`FIG. 8 shows that for every frequency in the range
`250-3500 Hz, the near-field response of an exemplary SOD
`microphone assembly fell within 5 dB of the FOD micro(cid:173)
`phone included for comparison, and over most of that range, 35
`the response was within 2 dB. These results indicate that in
`the near field, the SOD assembly behaved substantially like
`a FOD microphone.
`FIG. 9 shows that over a frequency range of 100-1000
`Hz, the SOD assembly was less sensitive to far-field stimu- 40
`lation than the FOD microphone by a margin, over most of
`that range, of 10 dB or more. FIG. 9 also shows that on the
`average, the sensitivity of the SOD assembly to low-fre(cid:173)
`quency noises fell off more rapidly (with decreasing fre(cid:173)
`quency) than that of the FOD microphone. This relatively 45
`steep, o>2, far-field frequency dependence is generally
`deskable because it helps to reject noise from low-frequency
`sources such as crowds and vehicular traffic.
`FIG. 10 shows the effect on near-field sensitivity of
`shifting the source-to-microphone distance from 1 cm to 2.5 50
`cm. It is evident from the figure that the relative change for
`the SOD assembly was very similar to the relative change
`for the FOD microphone. This supports our observation that
`in the near field, the acoustic characteristics of the SOD
`assembly are similar to those of a FOD microphone.
`The inventive microphone assembly is readily incorpo(cid:173)
`rated in communication platforms of various kinds. By way
`of example, FIG. 11 illustrates the incorporation of the
`microphone assembly into a cellular handset 140. The
`microphone assembly is readily incorporated either in the 60
`base portion 145 of a handset, or, alternatively, in a hinged,
`flip portion 150 if such a portion is present. Typically, the
`cartridge 155 will occupy a recess within the body of the
`handset, and ports 160 will be exposed to the operator's
`voice. As discussed above, foam layers or bodies (not 65
`shown) will typically be included in order to inhibit the
`pickup of turbulence.
`
`55
`
`8
`Another exemplary platform is a telephone handset, as
`depicted in FIG. 12. The broken line 165 shows the outer
`contour of a standard, Type K handset. The solid contour of
`the figure represents a modified shape that permits. the
`microphone assembly to be brought closer to the speaker's
`lips. Visible in the figure are cartridge 170, ports 175, and
`circuit board 180 for the microphone electronics package.
`FIG. 13 shows yet another exemplary platform, namely
`boom 185, which will typically be incorporated in a helmet
`or operator's headset.
`We claim:
`1. Apparatus comprising a transducer for converting
`acoustic signals, emitted by a source, to electrical output
`signals in the presence of acoustic noise; and further com(cid:173)
`prising a platform for maintaining the transducer-at a sub(cid:173)
`stantially constant distance from the source; wherein the
`transducer is adapted to respond to a second-order spatial
`derivative of the pressure field associated with at least some
`acoustic fields, CHARACTERIZED IN THAT the trans(cid:173)
`ducer comprises:
`a) two first-order differential microphones separated by a
`distance d, the microphones so situated within the
`platform that in use, they are on the same side of the
`source and approximately equidistant therefrom, and
`each microphone including a membrane having a sub(cid:173)
`stantially perpendicular orientation relative to a straight
`line drawn between the two microphones; and
`b) differencing means for receiving an electrical output
`signal from each of the two microphones, and for
`producing, in response thereto, an electrical difference
`signal proportional to the difference between the
`respective microphone output signals.
`2. Apparatus in accordance with claim 1, further com(cid:173)
`prising means for balancing the sensitivities of the two
`micro