US007155019B2
`
`US 7,155,019 B2
`(10) Patent No:
`a2) United States Patent
`Hou
`(45) Date of Patent:
`Dec. 26, 2006
`
`
`(54) ADAPTIVE MICROPHONE MATCHINGIN
`MULTI-MICROPHONEDIRECTIONAL
`SYSTEM
`
`(75)
`
`Inventor: Zezhang Hou, Cupertino, CA (US)
`
`(73) Assignee: Apherma Corporation, Sunnyvale, CA
`(US)
`
`(*) Notice:
`
`Subject to any disclaimer, the term ofthis
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 1008 days.
`
`7/1995 Jampolsky
`5,434,924 A
`11/1995 Sasaki et al. oc 381/92
`5,471,538 A *
`12/1995 Lindemannet al.
`5,479,522 A
`6/1996 Killion et al.
`5,524,056 A
`4/1997 Matouketal.
`5,625,684 A
`4/1998 Widrow
`5,737,430 A
`4/1998 Castello Da Costaetal.
`5,740,256 A
`5/1998 Preveset al.
`5,757,933 A
`7/2001 Vernon et al. ......... 324/253
`6,268,725 B1*
`9/2001 Tkeda seceecccssscecsseees 381/71.11
`6,285,768 BL*
`6,654,468 B1* 11/2003 Thompson... 381/92
`
`FOREIGN PATENT DOCUMENTS
`
`(21) Appl. No.: 09/808,694
`
`(22)
`
`Filed:
`
`Mar. 14, 2001
`
`
`
`EP
`569216 Al
`11/1993
`EP
`96 269085
`9/1994
`EP
`0856833 A2
`8/1998
`
`(65) 982971 A2—8/1999Prior Publication Data EP
`
`JP
`63-002500
`1/1988
`US 2002/0034310 Al
`Mar. 21, 2002
`Related U.S. Application Data
`(60) Provisional application No. 60/189,282,filed on Mar.
`14, 2000.
`
`(51)
`
`Int. Cl.
`(2006.01)
`HOAR 3/00
`(52) US. Ch. w..coecoreereeenrnneeie 381/92; 381/313
`(58) Field of Classification Search 0.0.0.0... 381/92,
`oo 381/94.1, 94.9, 313, 71.7, 71.6, 711M
`See application file for complete search history.
`Refe
`Cited
`emerges Oe
`U.S. PATENT DOCUMENTS
`
`56
`(56)
`
`3,836,732 A
`3,975,599 A
`4,131,760 A
`4,245,313 A *
`4,701,953 A
`4,712,244 A
`4,751,738 A
`4,956,867 A
`5.214.709 A
`5,325,436 A
`5,390,254 A
`
`9/1974 Johanson etal.
`8/1976 Johanson
`12/1978 Christensen etal.
` VI981 Coateseee 702/13
`10/1987 White
`12/1987 Zwicker et al.
`6/1988 Widrowetal.
`9/1990 Zurek et al.
`5/1993. Ribic
`6/1994 Soli et al.
`2/1995 Adelman
`
`(Continued)
`OTHER PUBLICATIONS
`Glover, “A review of Cardioid Type Unidirectional Microphones”,
`(Jan. 1940) J.AS.A,, vol. 11, pp. 296-302.
`
`(Continued)
`Primary Examiner—Vivian Chin
`Assistant Examiner—Justin Michalski
`(74) Attorney, Agent, or Firm—Beyer Weaver & Thomas
`LLP
`(57)
`
`ABSTRACT
`
`Improved approaches to matching sensitivities of micro-
`phones in multi-microphonedirectional processing systems.
`These approaches operate to adaptively match microphone
`sensitivities so that directional noise suppression is robust.
`As a result, microphone sensitivities remain matched not
`only over
`time but also while in actual use. These
`approachesare particularly useful for hearing aid applica-
`ti
`:
`hich directi
`1
`noi a rtant
`ions in which
`directional noise suppression is important.
`
`14 Claims, 14 Drawing Sheets
`
`
`
`
`MICA
`
`502
`
`50a)
`
`MIN.
`~] EST.
`504-510)
`UNIT
`
`
`
`
`DELAY
`UNIT
`
`506
`
`|
`
`
`MIN.
`
`EST-
`UNIT
`
`[_
`
`[A
`
`ia 500
`
`512
`SUBTRACTION
`UNIT
`
`
`
`
`
`DIVIDE
`UNIT
`
`
`OUTPUT
`SIGNAL
`
`516
`
`iarytiienl
`
`514
`
`Amazon v. Jawbone
`US. Patent 8,321,213
`
`Amazon Ex. 1008
`
`1
`
`Amazon v. Jawbone
`U.S. Patent 8,321,213
`Amazon Ex. 1008
`
`

`

`US 7,155,019 B2
`Page 2
`
`FOREIGN PATENT DOCUMENTS
`
`JP
`WO
`
`11-220796
`WO 99/03091
`
`8/1999
`1/1999
`
`OTHER PUBLICATIONS
`
`Killion et al., “The case of the missing dots: AI and SNR loss”,
`(May 1998) The Hearing Journal, vol. 51, No. 5, pp. 1-6.
`Gravel et al., “Children’s Speech Recognition in Noise Using
`Omni-Directional and Dual-Microphone Hearing Aid Technology”,
`Ear & Hearing, Feb. 1999, pp. 1-11.
`Ricketts et al., “Comparison of Performance across Three Direc-
`tional Hearing Aids”, J Am Acad Audiol vol. 10, 1999, pp. 180-189.
`Buerkli-Halevy, “The directional microphone advantage”, Hearing
`Instruments, vol. 38, No. 8, 1987.
`Preves, “Directional Microphone Use in ITE Hearing Instruments”,
`The Hearing Rev., Jul. 1997, pp. 21-27.
`Greenberg et al., “Evaluation of an adaptive beamforming method
`for hearing aids”, J. Acoust. Soc. Am. 91 (3), Mar. 1992, pp.
`1662-1676.
`Agnew, “How multi-microphone arrays can improve directional-
`ity”, The Hearing Journal, vol. 50, No. 8, Aug. 1997, pp. 34-46.
`Roberts et al., “Measurement and. Intelligibility Optimization of
`Directional Microphones for Use in Hearing Aid Devices”, 103"4
`Convention of Audio Engineering Society (AES), Sep. 1997, 13 pp.
`Desloge et al., “Microphone-Array Hearing Aids with Binaural
`Output—Part I: Fixed-Processing Systems”, IEEE Trans. on Speech
`& Audio Proc., vol. 5, No. 6, Nov. 1997, pp. 529-542.
`Welker et al., “Microphone-Array Hearing Aids with Binaural
`Output—Part
`II: A Two-Microphone Adaptive System’,
`IEEE
`Trans. On Speech & Audio Proc., vol. 5, No. 6, Nov. 1997, pp.
`543-551.
`Greenberg, “Modified LMS Algorithms for Speech Processing with
`an Adaptive Noise Canceller”, IEEE Trans. On Speech & Audio
`Proc., vol. 6, No. 4, Jul. 1998, pp. 338-351.
`Van Tasell, “New DSP instrument designed to maximize binaural
`benefits”,
`, The Hearing Journal, vol. 51, No. 4, Apr. 1998, pp.
`40-49.
`
`Edwards etal., “New digital processor for hearing loss compensa-
`tion is based on auditory system”, The Hearing Journal, vol. 51, No.
`8, Aug. 1998.
`Stadler et al., “On the potential of fixed arrays for hearing aids”, J.
`Acoust. Soc. Am. 94 (3), Pt. 1, Sep. 1993, pp. 1332-1342.
`Zurek et al., “Prospects and Limitations of Microphone-Array
`Hearing Aids”, World Scientific, Aug. 1995, pp. 233-244.
`Killion et al., “Real-world performance of an ITE directional
`microphone”, The Hearing Journal., vol. 51, No. 4, Apr. 1998, pp.
`1-6.
`
`‘I Can Hear What People Say, But I Can’t
`Killion, “SNR Loss:
`Understand Them’”, The Hearing Rev., vol. 4, No. 12, Dec. 1997.
`Edwardset al., “Signal-processing algorithms for a new software-
`based, digital hearing device”, The Hearing Journal, vol. 51, No. 9,
`Sep. 1998.
`Killion, “The SIN report: circuits haven’t solved the hearing-in-
`noise problem”, The Hearing Journal., vol. 50, No. 10, Oct. 1997,
`pp. 28-31.
`Cadalli et al., “Wideband Maximum Likelihood Direction Finding
`and Signal Parameter Estimation by Using the Tree-Structured EM
`Algorithm’, IEEE Trans. On Signal Proc., vol. 47, No. 1, Jan. 1999,
`pp. 201-206.
`Souza et al., “Quantifying the Contribution of Audibility to Rec-
`ognition of compression-Amplified Speech”, Ear & Hearing, vol.
`20, No. 1, Feb. 1999, pp. 12-20.
`microZOOM™, Marketing pamphlet, Phonak, 2 pgs.
`Senso: The Single Most Advanced Hearing Instrument Ever Devel-
`oped, Marketing pamphlet, Widex, 2 pgs.
`“HJReport”, The Hearing Journal, vol. 50, No. 8, Aug. 1997, pp.
`7-8.
`
`International Search Report, EPO/ISA, re PCT/US01/08256, dated
`Aug. 2, 2002.
`
`* cited by examiner
`
`2
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 1 of 14
`
`US 7,155,019 B2
`
`
`
`
`SUBTRACTION
`
`
`
`FIG. 1
`
`OUTPUT
`
`400x
`
`
`+| SUBTRACTION
`
`
`
`
`
` FIXED
`
`
`FILTER
`
`1+0.5z"'
`1+ 22°
`
`OUTPUT
`
`3
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 2 of 14
`
`US 7,155,019 B2
`
`
`180
`
`FIG. 2(b)
`
`4
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 3 of 14
`
`US 7,155,019 B2
`
`5
`
`

`

`US 7,155,019 B2
`
`ree.4 [Yiu
`
`(e
`
`eOl
`
`oooe8
`
`o00L
`
`ooo9
`
`ooo0s
`
`ooor
`
`ooo
`
`oo0z
`
`OOOL
`
`
`
`(ZH)AsueEenbesd
`
`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 4 of 14
`
`
`
`Perryrererrerrrr
`
`SHONAT:sf
`
`6
`
`

`

`U.S. Patent
`
`ireriratesitilTirTD
`
`PreearrrrrryeeWrcrrrrerri
`
`Dec. 26, 2006
`
`Sheet 5 of 14
`
`US 7,155,019 B2
`
`
`
`g009oo0zgo009ooosooorQo00eo00coouto
` {UOaye:Mesua
`
`owes:coal
`
`
`
`(q)e‘OI4
`
`
`
`(RH)Avuenboisy
`
`
`
`7
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 6 of 14
`
`US 7,155,019 B2
`
`
`
`EerrterwereWT
`
`EAiiatilLilcies
`
`(
`2)
`
`€ ‘
`
`Old
`
`
`
`(2H)Avuenbeay
`
`
`
`ooo8 ooo.
`
`o00o
`
`o00n
`
`900r
`
`soon
`
`o00z
`
`OOOL
`
`wea
`
`auSidTHeORO,
`
`“Payend-
`
`
`
`8
`
`
`
`

`

`U.S. Patent
`
`
`
`perrerrereweerrrrerrerary
`
` pZz
`
`Dec. 26, 2006
`
`Sheet 7 of 14
`
`US 7,155,019 B2
`
`(P
`
`)e“Ol
`
`
`
`(3H)Abenbeay
`
`onor
`
`o00eEougeoooL
`
`TTT
`
`IAA.
`
`rors
`itz4
`
`9
`
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 8 of 14
`
`US 7,155,019 B2
`
`f 500
`
`
`
`UNIT
`
`OUTPUT
`SIGNAL
`
` SUBTRACTION
`
`
`DIVIDE
`UNIT
`
`
`
`
`MULTIPLY
`UNIT
`
`506
`
`514
`
`FIG. 5
`
`10
`
`10
`
`

`

`NOILOVYLENS
`
`009
`
`YVAN ScO09
`
`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 9 of 14
`
`US 7,155,019 B2
`
`AldILINW
`
`LINN
`
`029
`
`9Sls
`
`909
`
`NOISYSANOD
`
`LINN
`
`
`
`YVANITOLSO1LINN
`
`
`
`NOILOVYLENSLINN
`
`NOISYSANOD
`
`SOTOLYVANIT
`
`NOISHSAANOO
`
`LINA
`
`11
`
`SOTOL
`
`11
`
`
`
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 10 of 14
`
`US 7,155,019 B2
`
`002_7
`
`Old
`
`OZ
`
`ZOld
`
`AdLIN
`
`‘HO
`
`LOVYLENS
`
`“LH
`
`
`
`LNNOWYLaSss0
`
`12
`
`12
`
`
`
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 11 of 14
`
`US 7,155,019 B2
`
`per|og
`
`008_-A
`
`OlZ
`
`8Old
`
`Old
`
`AldLINW
`
`LOVYLENS
`
`‘LH
`
`13
`
`
`
`
`
`AINNOWY14S440
`
`13
`
`
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 12 of 14
`
`US 7,155,019 B2
`
`BAND
`SPLIT
`FILTERS
`
`2902
`
`(nc)
`cp
`
`MINIMUM
`
`ESTIMATE2
`
`OUTPUT
`
`SIGNAL
`
`MINIMUM
`ESTIMATE3|||DIVIDE 3
`
`510-1
`
`— MINIMUM ESTIMATE 14
`
`504
`
`904
`
`506
`
`BAND
`SPLIT
`FILTERS
`
`MULTIPLY
`3
`
`FIG. 9
`
`14
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 13 of 14
`
`US 7,155,019 B2
`
`
`
`MINIMUM
`C)
`
`
`
`ESTIMA
`OUTPUT
`
`SIGNAL
`
`MINIMUM
`ESTIMATE 1
`
`TE 2
`
`
`
`MINIMUM
`ESTIMATE 3
`
`_f
`
`902
`
`BAND
`SPLIT
`FILTERS
`
`(nc)
`
`502
`
`504
`
`BAND
`
`SPLIT
`FILTERS
`
`3
`
`FIG. 10
`
`15
`
`15
`
`

`

`U.S. Patent
`
`Dec. 26, 2006
`
`Sheet 14 of 14
`
`US 7,155,019 B2
`
`OUTPUT
`SIGNAL
`
`MULTIPLY
`
`
`
`
`
`
`| 906
`
`4 6
`
`20-2
`
`904
`
`>
`DELAY a
`
`606
`
`BAND
`SPLIT
`FILTERS
`
`s
`U
`M
`
`MULTIPLY
`2
`620-3
`
`MULTIPLY
`3
`
`FIG. 11
`
`16
`
`16
`
`

`

`US 7,155,019 B2
`
`1
`ADAPTIVE MICROPHONE MATCHING IN
`MULTI-MICROPHONE DIRECTIONAL
`SYSTEM
`
`CROSS-REFERENCE TO RELATED
`APPLICATIONS
`
`This application claims the benefit of U.S. Provisional
`Application No. 60/189,282,
`filed Mar. 14, 2000, and
`entitled “METHODS FOR ADAPTIVE MICROPHONE
`MATCHING IN MULTI-MICROPHONE DIRECTIONAL
`
`SYSTEM”, the contents of which is hereby incorporated by
`reference. This applicationis also related to U.S. application
`Ser. No. 09/788,271,
`filed Feb. 16, 2001, and entitled
`“NULL ADAPTATION IN MULTI-MICROPHONE
`
`DIRECTIONAL SYSTEM”, the contents of which is hereby
`incorporated by reference.
`
`BACKGROUND OF THE INVENTION
`
`1. Field of the Invention
`
`The present invention relates to multi-microphone sound
`pick-up systems and, more particularly, to matching micro-
`phonesensitivity in multi-microphone sound pick-up sys-
`tems.
`
`2. Description of the Related Art
`Suppressing interfering noise is still a major challenge for
`most communication devices involving a sound pick up
`system such as a microphone or a multi-microphonearray.
`The multi-microphonearray can selectively enhance sounds
`coming from certain directions while suppressing interfer-
`ence coming from other directions.
`FIG. 1 showsa typical direction processing system in a
`two-microphone hearing aid. The two microphones pick-up
`sounds and convert them into electronic or digital signals.
`The output signal form the second microphone is delayed
`and subtracted from the output signal of the first micro-
`phone. Theresult is a signal with interference from certain
`directions being suppressed.
`In other words,
`the output
`signal is dependent on which directions the input signals
`come from. Therefore, the system is directional. The physi-
`cal distance between the two microphones andthe delay are
`two variables that control the characteristics of the direc-
`tionality. For hearing aid applications, the physical distance
`is limited by the physical dimension of the hearing aid. The
`delay can be set in a delta-sigma analog-to-digital converter
`(A/D)or by use of an all-pass filter.
`The sensitivity of the microphones of the sound pick up
`system must be matched in order to achieve good direction-
`ality. When the sensitivities of the microphones are not
`properly matched,
`then the directionality is substantially
`degraded and thus the ability to suppress interference com-
`ing from a particular direction is poor. FIGS. 2(a), 2(6), 2(c)
`and 2(d) illustrate representative polar patterns for micro-
`phonesensitivity discrepancies of 0, 1, 2, and 3 dB, respec-
`tively. Note that the representative polar pattern shown in
`FIG. 2(a) is the desired polar pattern which offers maxi-
`mized directionality. The representative polar patterns
`shown in FIGS. 2(5)-2(d) are distorted polar patterns that
`respectively illustrate directionality becoming progressively
`worse as the sensitivity discrepancy increases respectively
`from 1, 2 and 3 dB. FIGS. 3(a), 3(6), 3(c) and 3(d) illustrate
`representative spectrum response for microphonesensitivity
`discrepancies of 0, 1, 2, and 3 dB,
`respectively, with
`reference to a 1 kHz pure tone in white noise. Note that the
`Signal-to-Noise Ratio of the spectrum shown in FIGS.
`3(a)-3(d) is 14, 11, 9 and 7 dB, respectively. Accordingly, a
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`good match of sensitivity between microphones is very
`important to good directionality.
`Conventionally, manufacturers manually match the
`microphonefor their multi-microphone directional process-
`ing systems. While manual matching of the microphones
`provides for improved directionality,
`the operational or
`manufacturing costs are substantial. Besides cost-effective-
`ness, manual matching has other problems that compromise
`manual matching. One problem is that microphone sensi-
`tivity tends to drift over time. Hence, once matched micro-
`phones can become mismatchedovertime. Another problem
`is that the sensitivity difference can depend on how the
`multi-microphone directional processing systems is used.
`For example, in hearing aid applications, a microphonepair
`that is perfectly matched as determined by measurements at
`manufacture may become mismatched whenthe hearing aid
`is put on a patient. This can occur because at manufacture
`the microphones are measured in a field where sound
`pressure level is the same everywhere(free field), while in
`real life situation (in situ) sound pressure may notdistribute
`uniformly at microphone locations. Hence, when such pres-
`sure differences result, the microphones are in effect mis-
`matched. In another word, because the microphones are
`matched in free field, not
`in situ,
`the microphones can
`actually be mismatched when used in real
`life, which
`degrades directionality.
`Some manufacturers have used a fixed filter in their
`designs of multi-microphone directional processing sys-
`tems. FIG. 4 illustrates a conventional
`two-microphone
`directional processing system 400 havinga first microphone
`402, a second microphone 404, a delay 406, a fixed filter
`408, and a subtraction unit 410. The fixed filter 408 can
`serve to compensate for a mismatch in microphone sensi-
`tivity. Thefixedfilter approach is more cost-effective that the
`manual matching. However, the other problems (e.g., drift
`over time and in-situ mismatch) of manual matchingarestill
`present with the fixed filter approach.
`Thus, there is a need for improved approaches to match
`sensitivities of microphones in multi-microphonedirectional
`processing systems.
`
`SUMMARY OF THE INVENTION
`
`the invention relates to improved
`Broadly speaking,
`approaches to matching sensitivities of microphones in
`multi-microphone directional processing systems. These
`approaches operate to adaptively match microphonesensi-
`tivities so that directional noise suppression is robust. As a
`result, microphone sensitivities remain matched not only
`over time but also while in actual use. These approaches are
`particularly useful for hearing aid applications in which
`directional noise suppression is important.
`The invention can be implemented in numerous ways
`including as a method, system, apparatus, device, and com-
`puter readable medium. Several embodiments of the inven-
`tion are discussed below.
`
`As an adaptive directional sound processing system, one
`embodimentof the invention includesat least: at least first
`and second microphones spacedapart by a distance, the first
`microphonesproducing a first electronic sound signal and
`the second microphoneproducing a second electronic sound
`signal; means for processing the second electronic sound
`signal to adaptively produce a compensation scaling amount
`that compensates for sensitivity differences betweenthefirst
`and second microphones; a scaling circuit operatively con-
`nected to the means for scaling and the second microphone,
`the scaling circuit operates to scale the second electronic
`
`17
`
`17
`
`

`

`US 7,155,019 B2
`
`3
`sound signal in accordance with the compensation scaling
`amount; and a subtraction circuit operatively connected to
`the scaling circuit and the first microphone, the subtraction
`circuit producing an output difference signal by subtracting
`the scaled second electronic sound signal from the first
`electronic sound signal.
`As an adaptive directional sound processing system,
`another embodiment of the invention includes at least: at
`
`least first and second microphones spaced apart by a pre-
`determined distance, the first microphones producinga first
`electronic sound signal and the second microphone produc-
`ing a second electronic sound signal; a first minimum
`estimate circuit operatively coupledto the first microphone,
`the first minimum estimate circuit producesa first minimum
`estimate for the first electronic sound signal from the first
`microphone; a second minimum estimate circuit operatively
`coupled to the second microphone, the second minimum
`estimate circuit produces a second minimum estimate for the
`second electronic sound signal from the second microphone;
`a divide circuit operatively connectedto the first and second
`minimum estimate circuits, the divide circuit operates to
`produce a scaling signal from the first and second minimum
`estimates; a multiply circuit operatively connected to the
`divide circuit and the second microphone,
`the multiply
`circuit operates to multiply the second electronic sound
`signal by the scaling signal to produce a scaled second
`electronic sound signal; and a subtraction circuit operatively
`connected to the multiply circuit and the first microphone,
`the subtraction circuit producing an output difference signal
`by subtracting the scaled second electronic sound signal
`from the first electronic sound signal.
`As a hearing aid device having an adaptive directional
`sound processing, one embodimentofthe invention includes
`at least: at least first and second microphones spaced apart
`by a distance, the first microphones producing a first elec-
`tronic sound signal and the second microphoneproducing a
`second electronic sound signal; sensitivity difference detec-
`tion circuitry operatively connected to the first and second
`microphones, the sensitivity difference detection circuitry
`adaptively produces a compensation scaling amountcorre-
`sponding to sensitivity differences between the first and
`second microphones; a scaling circuit operatively connected
`to the sensitivity difference detection circuitry and the
`second microphone, the scaling circuit operates to scale the
`second electronic sound signal in accordance with the com-
`pensation scaling amount; and a subtraction circuit opera-
`tively connected to the scaling circuit and the first micro-
`phone,the subtraction circuit producing an output difference
`signal by subtracting the scaled second electronic sound
`signal from the first electronic sound signal.
`As a method for adaptively measuring and compensating
`for acoustical differences between sound signals picked up
`by microphones, one embodimentof the invention includes
`at least the acts of: receiving first and second electronic
`sound signals from first and second microphones, respec-
`tively; determining a compensation scaling amount
`that
`compensates for acoustic differences with respectto thefirst
`and second microphones; scaling the second electronic
`sound signal in accordance with the compensation scaling
`amount; and producing a differential electronic sound signal
`by subtracting the scaled second electronic sound signal
`from the first electronic sound signal.
`Other aspects and advantages of the invention will
`become apparent from the following detailed description
`taken in conjunction with the accompanying drawings which
`illustrate, by way of example, the principlesof the invention.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`4
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`The invention will be readily understood by the following
`detailed description in conjunction with the accompanying
`drawings, wherein like reference numerals designate like
`structural elements, and in which:
`FIG. 1 shows a typical direction processing system in a
`two-microphonehearing aid;
`FIGS. 2(a@)-2(d) illustrate representative polar patterns for
`various microphonesensitivity discrepancies;
`FIGS. 3(a)-3(d) illustrate representative Signal-to-Noise
`Ratio spectrums respectively correspondingto the represen-
`tative polar patterns shown in FIGS. 2(a)-2(d);
`FIG.4 illustrates a conventional two-microphone direc-
`tional processing system;
`FIG. 5 is a block diagram of a two-microphonedirectional
`processing system according to one embodiment of the
`invention;
`FIG. 6 is a block diagram of a two-microphonedirectional
`processing system according to another embodimentof the
`invention;
`FIG. 7 is a block diagram of a minimum estimate unit
`according to one embodiment of the invention;
`FIG. 8 is a block diagram of a minimum estimate unit
`according to another embodiment of the invention;
`FIG. 9 is a block diagram of a multi-microphone direc-
`tional processing system that operates to perform multi-band
`adaptive compensation for microphone mismatch;
`FIG. 10 is a block diagram of a multi-microphonedirec-
`tional processing system according to one embodiment of
`the invention; and
`FIG. 11 is a block diagram of a multi-microphone direc-
`tional processing system according to another embodiment
`of the invention.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`The invention relates to improved approaches to matching
`sensitivities of microphones in multi-microphonedirectional
`processing systems. These approaches operate to adaptively
`match microphone sensitivities so that directional noise
`suppression is robust. As a result, microphonesensitivities
`remain matched not only over time but also while in actual
`use. These approachesare particularly useful for hearing aid
`applications in which directional noise suppression is impor-
`tant.
`
`According to one aspect, the invention operates to adap-
`tively measure a sensitivity difference between microphones
`in a multi-microphone directional processing system, and
`then compensate (or correct) an electronic sound signal from
`one or more of the microphones. Asa result of the adaptive
`processing, the microphones“effectively” become matched
`and remain matched over time and while in use.
`
`the invention enables multi-microphone
`Consequently,
`directional processing systems to achieve superior direction-
`ality and consistent Signal-to-Noise Ratio (SNR) across all
`conditions. The invention is described below with respect to
`embodiments particularly well suited for use with hearing
`aid applications. However, it should be recognized that the
`invention is not limited to hearing aid applications, but is
`applicable to other sound pick-up systems.
`Embodimentsofthis aspect of the invention are discussed
`below with reference to FIGS. 5-11. However, those skilled
`in the art will readily appreciate that the detailed description
`
`18
`
`18
`
`

`

`US 7,155,019 B2
`
`5
`given herein with respect to these figures is for explanatory
`purposes as the invention extends beyond these limited
`embodiments.
`
`As noted above, microphone matching is important for
`multi-microphone directional systems. Different and undes-
`ired responses will result when the sensitivities of the
`microphones are not matched. The acoustic delay between
`the microphones further complicates matching problems.
`For example, even if the microphonesare perfectly matched,
`the instantaneous response of the microphones can be dif-
`ferent because of the delay and/or fluctuation in the acoustic
`signals. Therefore,
`it
`is not enough to simply use the
`difference of the responses to correct the problem. More
`complex processing is necessary to eliminate the effects of
`acoustic delay between the microphones and/or the fluctua-
`tion in the acoustic signals.
`According to one aspect of the invention, responses from
`each microphone are processed such that
`the resulting
`processed signals are not sensitive to the acoustic delay
`between the microphones and the fluctuation of acoustic
`conditions. A difference between the processed signals from
`the microphone channels can then be usedto scale at least
`one microphone’s response so as to compensate or correct
`for sensitivity differences between the microphones.
`FIG. 5 is a block diagram of a two-microphonedirectional
`processing system 500 according to one embodimentof the
`invention. The two-microphonedirectional processing sys-
`tem 500 includes a first microphone 502 and a second
`microphone 504. The first microphone 502 producesa first
`electronic sound signal and the second microphone 504
`produces a secondelectronic sound signal. A delay unit 506
`delays the second electronic sound signal. The two-micro-
`phonedirectional processing system 500 also includesa first
`minimum estimate unit 508, a second minimum estimate
`unit 510 and a divide unit 512. The first minimum estimate
`unit 508 estimates the minimum for thefirst electronic sound
`
`signal. The second minimum estimate unit 510 estimates the
`minimum of the second electronic sound signal. Typically,
`these minimums are measured overa time constant duration,
`such that the minimum is a relatively long-term minimum.
`The divide unit 512 produces a quotient by dividing the first
`minimum estimate by the second minimum estimate. The
`quotient represents a scaling amount that is sent to a mul-
`tiplication unit 514. The second electronic sound signal is
`then multiplied with the scaling amount to produce a com-
`pensated sound signal. The compensated soundsignal is thus
`compensated (or corrected) for the relative difference in
`sensitivity between the mismatchedfirst and second micro-
`phones 502 and 504. A subtraction unit 516 then subtracts
`the compensated electronic sound signal
`from the first
`electronic sound signal to produce an output signal. At this
`point, the output signal has been processed by the two-
`microphone directional processing system 500 to have
`robust directionality despite a mismatch between thefirst
`and second microphones 502 and 504.
`The two-microphone directional processing system 500
`uses a single-band adaptive compensation scheme to com-
`pensate for sensitivity differences between the microphones.
`In this embodiment, minimum estimates and division cal-
`culations are performed. The minimum estimates can, for
`example, be performed by minimum estimate units shown in
`more detail below with respect to FIGS. 7 and 8. It should
`also be noted that the delay unit 506 can be positioned within
`the two-microphone directional processing system 500 any-
`where in the channel associated with the second electronic
`
`sound signal prior to the subtraction unit 516. Still further,
`
`6
`it should be noted that a multiple-band adaptive compensa-
`tion scheme could alternatively be utilized.
`Moreover, although the two-microphonedirectional pro-
`cessing system 500 uses minimum estimates of the elec-
`tronic sound signals produced by the first and second
`microphones 502 and 504, other signal characteristics can
`alternatively be used. For example, Root-Mean-Square
`(RMS)average of the electronic sound signals produced by
`the microphones could be used. With such an approach, the
`RMSaverage could be measured over a time constant
`duration. The time constant can be set such that the average
`is relatively long-term so as to avoid impact of signal
`fluctuations. The time constant with an RMSapproach is
`likely to be longer than the time constant for the minimum
`approach.
`The two-microphone directional processing system 500
`operates to scale the intensity of an electronic soundsignal
`from one or more of the microphones. With respect to the
`two-microphonedirectional processing system 500, the pro-
`cessing (including the scaling) is performed in a linear
`domain. However, the scaling or other processing can also
`be performed in a logarithm (or dB) domain.
`FIG. 6 is a block diagram of a two-microphonedirectional
`processing system 600 according to another embodiment of
`the invention. The two-microphone directional processing
`system 600 includes a first microphone 602 and a second
`microphone 604. The first microphone 602 producesa first
`electronic sound signal and the second microphone 604
`produces a secondelectronic sound signal. A delay unit 606
`delays the second electronic sound signal. The two-micro-
`phonedirectional processing system 600 also includesa first
`minimum estimate unit 608 and a second minimum estimate
`unit 610. The first minimum estimate unit 608 estimates the
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`minimum for the first electronic sound signal. The second
`minimum estimate unit 610 estimates the minimum of the
`
`second electronic sound signal. Typically, these minimums
`are measured over a time constant duration, such that the
`minimum is a relatively long-term minimum.
`The two-microphone directional processing system 600
`also includes a first linear-to-log conversion unit 612, a
`second linear-to-log conversion unit 614, a subtraction unit
`616, and a log-to-linear conversion unit 618. The first
`minimum estimate is converted from the linear domain to
`
`the logarithm domain by the first linear-to-log conversion
`unit 612, and the second minimum estimate is converted
`from the linear domain to the logarithm domain by the
`second linear-to-log conversion unit 614. The subtraction
`unit 616 then subtracts the second minimum estimate from
`the first minimum estimate to produce a difference amount.
`The log-to-linear conversion unit 614 then converts the
`difference amount to the linear domain.
`The converted difference amount produced bythe log-to-
`linear conversion unit 614 represents a scaling amount that
`is sent to a multiplication unit 620. The second electronic
`sound signal is then multiplied with the scaling amount to
`produce a compensated sound signal. The compensated
`sound signal
`is thus compensated (or corrected) for the
`relative difference in sensitivity between the mismatched
`first and second microphones 602 and 604. A subtraction
`unit 622 then subtracts the compensated electronic sound
`signal from the first electronic sound signal to produce an
`output signal. The output signal has been processed by the
`two-microphonedirectional processing system 500 to have
`robust directionality despite a physical mismatch between
`the first and second microphones 602 and 604.
`It should be noted that the two-microphone directional
`processing system 600 is generally similar to the two-
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`19
`
`19
`
`

`

`US 7,155,019 B2
`
`7
`microphonedirectional processing system 500 illustrated in
`FIG. 5. Both use similar circuitry to produce a single-band
`adaptive compensation scheme for a multi-microphone
`directional processing system. However, the divide unit 512
`shown in FIG. 5 is replaced by the linear-to-log conversion
`units 612 and 614, the subtraction unit 616 and the log-to-
`linear conversion unit 618 shown in FIG. 6. Mathematically,
`the divide unit 512 is equivalent to the combination of the
`linear-to-log conversion units 612 and 614, the subtraction
`unit 616 and the log-to-linear conversion unit 618. However,
`with certain approximations, the design shown in FIG. 6
`maybeable to perform a “divide” operation more efficiently.
`Also the delay unit 606 in FIG. 6 can be positioned any-
`where in the channel associated with the second electronic
`
`sound signal prior to the subtraction unit 622.
`FIG.7 is a block diagram of a minimum estimate unit 700
`according to one embodiment of the invention. The mini-
`mum estimate unit 700 is, for example, suitable for use as
`the minimum estimate units discussed above with respect to
`FIGS. 5 and 6. The minimum estimate unit 700 receives an
`input signal (e.g., electronic sound signal) that is to haveits
`minimum estimated. The input signal
`is supplied to an
`absolute value circuit 702 that determines the absolute value
`
`of the input signal. An add circuit 704 adds the absolute
`value of the input signal together with an offset amount 706
`and thus produces an offset absolute value signal. The
`addition of the offset amount, which is typically a small
`positive value, such as 0.000000000001,
`is used to avoid
`overflow in division or logarithm calculations performed in
`subsequent circuitry in the multi-microphone directional
`processing systems. The offset absolute value signal from
`the add circuit 704 is supplied to a subtract circuit 708. The
`subtract circuit 708 subtracts a previous output 710 from the
`offset absolute value signal to produce a difference signal
`712. The difference signal 712 is supplied to a multiply
`circuit 714. In addition, the difference signal 712 is supplied
`to a switch circuit 716. The switch circuit 716 selects one of
`two constants that are supplied to the multiply circuit 714.
`A first of the constants is referred to as alphaB and is
`supplied to the multiply circuit 714 when the difference
`signal 712 is greater than or equal to ze