United States Patent [19]
`United States Patent [191
`GraUlpe et al.
`Graupe et a1. '
`
`[76]
`[76]
`
`[54] METHOD AND MEANS FOR ADAPTIVELY
`METHOD AND MEANS FOR ADAPTIVELY
`[54]
`lF1LTERING NEAR-STATIONARY NOISE
`FILTERING NEAR-STATIONARY NOISE
`FROM AN INFORMATION BEARING
`FROM AN INFORMATION BEARING
`SIGNAL
`SIGNAL
`Inventors: Daniel Graupe, 2821 Redwing,.Fort
`Inventors: Daniel Graupe, 2821 Redwing, .Fort
`Collins, Colo. 80521; G. Donald
`Collins, Colo. 80521; G. Donald
`Causey, 3504 Dunlop, Chevy Chase,
`Causey, 3504 Dunlop, Chevy Chase,
`Md. 20015
`Md. 20015
`Appl. No.:
`866,938
`[21] Appl. No.: 866,938
`[21]
`Jan. 4, 1978
`Jan. 4, 1978
`[22] Filed:
`[22]
`Filed:
`
`Related US. Application Data
`Related U.S. Application Data
`[63] Continuation-in-part of Ser. No. 707,569, Jul. 21, 1976,
`Continuation-impart of Ser. No. 707,569, Jul. 21, 1976,
`[63]
`abandoned, which is a continuation-in-part of Ser. No.
`abandoned, which is a continuation-in-part of Ser. No.
`683,234, May 4, 1976, Pat. No. 4,025,721.
`683,234, May 4, 1976, Pat. No. 4,025,721.
`Int. Cl.2 ............................................. H04R 27/00
`[51]
`[51]
`llnt. Cl.2 ........................................... .. H04R 27/00
`[52] U.S. CI ................................... 179/1 P; 179/1 SC
`[52]
`US. Cl. ............. ..
`.. 179/1 P; 179/1 SC
`[58] Field of Search ................................ 179/1 P, 1 D
`[58]
`Field of Search
`............. .. 179/1 P, l D
`References Cited
`[56]
`References Cited
`[56]
`U.S. PATENT DOCUMENTS
`U.S. PATENT DOCUMENTS
`4/1974 Sachs .................................... 179/1 P
`3,803,357
`3,803,357
`4/1974 Sachs .................................. .. 179/1 P
`5/1977 Graupe et al. ....................... 179/1 P
`4,025,721
`4,025,721
`5/1977 Graupe et al. ..................... .. 179/1 P
`Primary Examiner—Kathleen H. Claffy
`Primary Examiner-Kathleen H. Claffy
`Assistant Examiner-E. S. Kemeny
`Assistant Examiner-E. S. Kemeny
`Attorney, Agent, or Firm-Sandler & Greenblum
`Attorney, Agent, or Firm—Sandler & Greenblum
`[57]
`ABSTRACT
`ABSTRACT
`[57]
`An input signal containing information such as speech
`An input signal containing information such as speech
`or music as well as near-stationary noise is applied in
`or music as well as near-stationary noise is applied in
`
`[11]
`[11]
`[45]
`[45]
`
`4,185,168
`4,185,168 '
`Jan. 22, 1980
`Jan. 22, 1980
`
`parallel to a noise-analysis circuit and a noise-reduction·
`parallel to a noise-analysis circuit and a noise-reduction -
`circuit, each of which comprises a plurality of bandpass
`circuit, each of which comprises a plurality of bandpass
`?lters covering the range of frequencies associated with
`filters covering the range of frequencies associated with
`the information. The absolute value, or a function
`the information. The absolute value, or a function
`thereof, of the output of each bandpass ?lter in the
`thereof, of the output of each bandpass filter in the
`noise-analysis circuit is produced and smoothed. The
`. noise-analysis circuit is produced and smoothed. The
`presence of near-stationary noise in the input signal is
`presence of near-stationary noise in the input signal is
`determined by examining the nature of the smoothed
`determined by examining the nature of the smoothed
`‘signal in each band assuming noise has a frequency
`signal in each band assuming noise has a frequency
`spectrum which does not vary with time or varies only
`spectrum which does not vary with time or varies only
`within a narrow range over a predetermined period of
`within a narrow range over a predetermined period of
`time with respect to the spectral parameters of the infor
`time with respect to the spectral parameters of the infor(cid:173)
`mation signal. If noise is detected, the noise-analysis
`mation signal. If noise is detected, the noise-analysis
`circuit identi?es spectral parameters of the information
`circuit identifies spectral parameters of the information
`and/or noise in each band using the smoothed signal
`and/or noise in each band using the smoothed signal
`therein.
`therein.
`In the preferred embodiment of the invention, the band(cid:173)
`In the preferred embodiment of the invention, the band
`pass filters of the noise-reduction circuit have gain ele(cid:173)
`pass ?lters of the noise-reduction circuit have gain ele
`ments that are adjusted in accordance with the identi(cid:173)
`ments that are adjusted in accordance with the identi
`fied parameters to minimize, under some continuous
`?ed parameters to minimize, under some continuous
`minimization criterion, the effect of the noise in the
`minimization criterion, the effect of the noise in the
`input signal thus enhancing intelligibility of the informa
`input signal thus enhancing intelligibility ofthe informa(cid:173)
`tion therein. Minimization can be such that the gain-to(cid:173)
`tion therein. Minimization can be such that the gain-to
`parameter relationships are similar to those in Weiner or
`parameter relationships are similar to those in Weiner or
`Kalman ?ltering theory with a-priori knowledge of the
`Kalman filtering theory with a-priori knowledge of the
`noise, or of the noise and information, except that in this
`noise, or of the noise and information, except that in this
`case, a-priori knowledge of the noise is acquired via
`case, a-priori knowledge of the noise is acquired via
`identification and is not preassumed.
`identi?cation and is not preassumed.
`
`27 Claims, 8 Drawing Figures
`27 Claims, 8 Drawing Figures
`
`SUMMIN6
`t---1r--[li;~U--------------I CIRClIl7'
`
`zoo
`
`/04
`
`8/I)/DI
`
`BANDZ
`
`/OZ
`
`CLOCK.
`
`fog
`
`110
`
`. ___ .~'-170
`
`RTL345-1_1028-0001
`
`

`
`00
`0'\
`~
`...
`U\
`00
`--~
`~
`
`4,185,168
`
`w
`~
`o
`.....
`g.
`en go
`
`Sheet 1 of3
`
`." a (D a
`~ . en .
`
`00 o
`1.0
`.....
`JV
`tv
`~
`
`US. Patent Jan. 22, 1980
`
`.~.-170
`
`--
`
`/07
`
`N QM
`
`/Ot'o
`
`7HRESJ.lO'-D~
`
`Fi(5. j
`
`(SPEECH/MUSIC, €7C.)
`INFO.eM~TION
`
`.CLOCK.
`
`I. I
`
`I
`
`I
`
`..,
`
`! ...
`
`S/6N'7ILS
`MIN COA/T,eOL
`
`,/'
`
`.,./
`
`/04
`
`200
`
`Cl12CuI7
`SUMMINt5
`
`/02
`
`BtlI'ID2 ~
`
`8{;,vD I
`
`/Fo
`
`RTL345-1_1028-0002
`
`

`
`00
`0'\
`~
`-.
`U\
`00
`~
`~ -.
`
`(t) -N
`en g
`
`w
`g,
`
`~
`
`00 o
`\0
`I-"
`,.N
`N
`
`""C a ('I) a
`
`t:I.l
`~ o
`
`o
`
`fn (Hz)
`
`fm
`
`f2
`
`fl
`
`1
`o
`
`FREQ.
`
`IBANN 0
`
`B~N 01 B~NDI
`
`CD a.
`etO
`z~
`OW
`~a::
`
`SPEECH SPECTRU M
`
`FIG. 2
`
`(f) ~
`(f)~
`CD
`
`-
`
`21
`
`/If
`
`26
`
`FILTER
`NOTCH
`
`BYPASSED
`
`20
`
`25
`
`23
`
`22
`
`FILTERED
`
`BYPASSED
`
`FI ...., . ....,.
`
`18-1
`
`"I
`
`NF2
`
`IlFILT.ERED
`
`BYPASS _
`
`•
`
`'7.
`.J
`; JNF1 I NFl
`
`/ 13) r-=-~:f;;;.---f
`
`Ass. V4WE.
`StvJoOTI-IE£> ~14N THRESHOLD 1-
`
`CIRCUIT
`
`DET.
`
`"
`
`LOG I C
`
`: /14
`
`-
`
`-
`
`-
`
`. ~14-2
`
`~S. V-9LcJ€.
`iptvroOnfSD
`
`DET.
`
`BPF 2
`
`13-2"
`
`I
`4
`L
`
`I DET.
`I SMOOTHc~114-1
`---
`
`-
`
`CLOCK
`
`118S. Vl'flU(£.
`
`BPF I
`
`13-1
`
`'"
`
`-
`
`~
`
`i
`
`~ BPF N
`,
`
`I
`
`i
`
`.. /-13-N
`
`OUTPUT
`FILTERED
`
`"f
`I 1+~17
`
`;-
`i
`"l
`
`I
`
`NF N
`
`FILTERED
`
`'"
`N F 2
`BY PASS
`
`'15
`
`-
`
`~~
`~
`1
`
`et (f) ~
`1-1
`a:: 0 c:t
`et a:: 9:-
`::i: et ~
`~ (!) ~
`W w~
`a:: -oJ-~
`Q:
`l3
`\.
`
`·~I a.
`
`15-N
`
`:
`
`-
`
`/15-1 FIG.4
`
`15-2
`
`_ CIRCUI_~
`"
`-[
`
`THRESHOLD
`
`LOGIC
`
`CIRC~IT 1
`i
`.,
`l
`
`THRESHOLD
`
`LOGIC
`
`_
`
`RTL345-1_1028-0003
`
`

`
`u.s. Patent
`
`Jan. 22, 1980
`
`Sheet 3 of 3
`
`4,185,168
`
`CLOCK..
`
`I
`I
`I
`I
`ITHRESHOLD
`I
`.
`I
`
`/g
`
`ZO ZZ Z4 TIME
`t
`
`FiG.~
`~--...I.-""",,- 16B
`
`NOISe
`
`NO
`NOIS£
`~
`OUTPUrs TO BLOC/:: J7
`
`RTL345-1_1028-0004
`
`

`
`El
`
`4,185,168
`
`:2
`2
`a function thereof (e.g., the RMS value) of the signal in
`a function thereof (e.g., the RMS value) of the signal in
`each band is obtained and smoothed, preferably by
`each band is obtained and smoothed, preferably by
`computing a running average of the absolute value in
`computing a running average of the absolute value in
`the band. In one technique, noise is assumed to be pres-
`the band. In one technique, noise is assumed to be pres
`S ent in the input signal if a threshold is exceeded by all or .
`ent in the input signal if a threshold is exceeded by all or _
`CROSS-REFERENCE TO RELATED
`CROSS-REFERENCE TO RELATED
`by a predetermined percentage of successive minima of
`by a predetermined percentage of successive minima of
`APPLICATION
`APPLICATION
`the smoothed signal in a band. Alternatively, the
`the smoothed signal in a band. Alternatively, the
`smoothed signal in each band can be sampled to enable
`This application is a continuation-in-part of copend
`smoothed signal in each band can be sampled to enable
`This application is a continuation-in-part of cop end(cid:173)
`the inverse signal-to-noise ratio (SNR -1) to be com-
`in.C application Ser. No. 707,569, ?led July 21, 1976,
`the inverse signal~to-noise ratio (SNR—1) to be com
`ing application Ser. No. 707,569, filed July 21, 1976,
`now abandoned, which application is hereby incorpo
`lD puted, the presence of noise being assumed if an
`puted, the presence of noise being assumed if an
`now abandoned, which application is hereby incorpo(cid:173)
`. SNR - 1 in a band exceeds a threshold. The SNR -1 in a
`rated by reference. The last mentioned application is,
`rated by reference. The last mentioned application is,
`' SNR-1 in a band exceeds a threshold. The SNR—1 in a
`itself, a continuation-in-part of application Ser. No.
`sampling interval is computed by taking the ratio of the
`sampling interval is computed by taking the ratio of the
`itself, a continuation-in-part of application Ser. No.
`683,234, ?led May 4, 1976, now US. Pat. No. 4,025,721,
`latest stored lowest minimum of the smoothed signal in
`683,234, filed May 4, 1976, now U.S. Pat. No. 4,025,721,
`latest stored lowest minimum of the smoothed signal in
`which patent is hereby incorporated by reference.
`a band to the difference between the sampled value of
`which patent is hereby incorporated by reference.
`a band to the difference between the sampled value of
`15 the smoothed signal and such minimum.
`the smoothed signal and such minimum.
`If the noise-analysis circuit detects the presence of
`If the noise-analysis circuit detects the presence of
`noise in accordance with the above tests, spectral pa(cid:173)
`noise in accordance with the above tests, spectral pa
`rameters of the noise, or of the noise and information
`rameters of the noise, or of the noise and information
`are then identified by the circuit and used to control the
`are then identi?ed by the circuit and used to control the
`operation of a noise-reduction circuit to which the input
`operation of a noise-reduction circuit to which the input
`signal is also applied. The last mentioned circuit com-
`signal is also applied. The last mentioned circuit corn
`prises a plurality of bandpass filters having a one-to-one
`prises a plurality of bandpass ?lters having a one-to-one
`correspondence with the bandpass filters of the noise(cid:173)
`correspondence with the bandpass ?lters of the noise
`analysis circuit. Each of the bandpass filters of the
`analysis circuit. Each of the bandpass ?lters of the
`noise-reduction circuit is responsive to a parameter
`noise-reduction circuit is responsive to a parameter
`identified from the smoothed signal in the correspond-
`identi?ed from the smoothed signal in the correspond
`ing band of the noise-analysis circuit for minimizing the
`ing band of the noise-analysis circuit for minimizing the
`effect of noise in the input signal on the output of the
`effect of noise in the input signal on the output of the
`noise-reduction circuit, minimization occuring under
`noise-reduction circuit, minimization occuring under
`some continuous minimization criterion.
`some continuous minimization criterion.
`The noise-analysis circuit can identify either noise
`The noise-analysis circuit can identify either noise
`parameters, or noise and information parameters in a
`parameters, or noise and information parameters in a
`band by utilizing the smoothed signal in the band. A
`band by utilizing the smoothed signal in the band. A
`parameter of the noise alone in a band can be in the form
`parameter of the noise alone in a band can be in the form
`of a predetermined function of the lowest one of a pre(cid:173)
`of a predetermined function of the lowest one of a pre
`determined number of successive minima of the
`determined number of successive minima of the
`smoothed signal in the band. Alternatively, the noise
`smoothed signal in the band. Alternatively, the noise
`parameter can te in the form of a predetermined func(cid:173)
`parameter can be in the form of a predetermined func
`tion of a more recent minimum of a succession of min(cid:173)
`tion of a more recent minimum of a succession of min
`ima if such recent minimum does not exceed the previ-
`ima if such recent minimum does not exceed the previ
`ous lower minimum by more than a given percentage
`ous lower minimum by more than a given percentage
`(e.g.; 25%).
`(e.g., 25%).
`_
`On the other hand, a parameter of the noise and infor-
`On the other hand, a parameter of the noise and infor
`45 mation in a band can be in the form of a predetermined
`mation in a band can be in the form of a predetermined
`45
`function of the signal-to-noise ratio (SNR) in the band.
`function of the signal-to-noise ratio (SNR) in the band.
`The SNR in a band is the difference between samples of
`The SNR in a band is the difference between samples of
`the smoothed signal in the band and the most recently
`the smoothed signal in the band and the most recently
`accepted noise-related minimum of a time-sequence of
`accepted noise-related minimum of a time-sequence of
`minima of the smoothed signal. The function's referred
`minima of the smoothed signal. The function’s referred
`to above may be continuous functions of the variables to
`to above may be continuous functions of the variables to
`yield either spectral parameters or their time domain
`yield either spectral parameters or their time domain
`equivalents.
`equivalents.
`In the preferred form of the invention, the bandpass
`In the preferred form of the invention, the bandpass
`55 filters of the noise-reduction circuit have gain elements
`?lters of the noise-reduction circuit have gain elements
`55
`controlled by the parameters identified by the noise(cid:173)
`controlled by the parameters identi?ed by the noise
`analysis circuit. When noise parameters alone are identi(cid:173)
`analysis circuit. When noise parameters alone are identi
`fied, the gain-to-parameter relationship necessary to
`?ed, the gain-to-pararneter relationship necessary to
`minimize noise may be similar to or approximate a Wie-
`minimize noise may be similar to or approximate a Wie
`60 ner filter with a~priori knowledge of noise as described
`ner ?lter with a-priori knowledge of noise as described
`60
`in Chapter 8.6 of "Foundations of the Theory of Learn(cid:173)
`in Chapter 8.6 of “Foundations of the Theory of Learn
`ing Systems" by Y. Tsypkin, Academic Press, New
`ing Systems” by Y. Tsypkin, Academic Press, New
`York (1973). In such case, a-priori knowledge of the
`York (1973). in such case, a-priori knowledge of the
`noise is obtained via the identification process referred
`noise is obtained via the identi?cation process referred
`65 above and is not preassumed.
`above and is not preassumed.
`65
`When both noise and information parmL ~ers are
`When both noise and information pararr. ters are
`identified, the gain adjustment of a bandp~~" filter will
`identi?ed, the gain adjustment of a bandpass filter will
`be according to the signal-to-noise ratio in the bandpass
`be according to the signal-to-noise ratio in the bandpass
`
`BACKGRGUND OF THE INVENTION
`BACX.GROUND OF THE INVENTION
`The invention relates to a method of and means for
`The invention relates to a method of and means for
`?ltering near-stationary, relatively long duration noise
`filtering near-stationary, relatively long duration noise
`from an input signal containing information such as
`from an input signal containing information such as
`speech or music.
`20
`speech or music.
`20
`For the purpose of the present application, the term
`For the purpose of the present application, the term
`"hear-stationary, relatively long duration noise”, is
`"near-stationary, relatively long duration noise", is
`hereinafter referred to as noise of the type described,
`hereinafter referred to as noise of the type described,
`and refers to noise whose frequency spectrum does not
`and refers to noise whose frequency spectrum does not
`vary with time o:r varies only within a narrow range 25
`vary with time or varies only within a narrow range
`over a predetermined period of time particularly with
`over a predetermined period of time particularly with
`respect to the spectral parameters of the information.
`respect to the spectral parameters of the information.
`Vehicular and machinery noises are examples of noise
`Vehicular and machinery noises are examples of noise
`or the type described. The minimum duration of noise of
`of the type described. The minimum duration of noise of
`type described is thus considerably longer than
`the type described is thus considerably longer than 30
`periods over which the spectrum of normal speech
`periods over which the spectrum of normal speech
`remains nearly fixed, such’ periods being of the order of
`remains nearly fixed, such. periods being of the order of
`a second. In addition, the invention is described below
`a second. In addition, the invention is described below
`as being applied to ?ltering noise of the type described
`as being applied to filtering noise of the type described
`from speech, but the invention is also applicable to
`from speech, but the invention is also applicable to 35
`other information bearing signals.
`other information bearing signals.
`U.S. Pat. No. 4,025,721 discloses a method of and
`US. Pat. No. 4,025,721 discloses a method of and
`means for adaptively ?ltering noise of the type de
`illeans for adaptively filtering noise of the type de(cid:173)
`scribed from speech by recognizing the existence of
`scribed from speech by recognizing the existence of
`such noise, identifying its parameters, adjusting the
`such noise, identifying its parameters, adjusting the 40
`parameters of a filter so as to filter such noise, and ap(cid:173)
`parameters of a ?lter so as to ?lter such noise, and ap
`plying an input signal containing such noise and speech
`plying an input signal containing such noise and speech
`to the ?lter. Termination of the noise of the type de
`to the filter. Termination of the noise of the type de(cid:173)
`scribed is also recognized in order to cause the input
`scribed is also recognized in order to cause the input
`signal to then bypass the ?lter.
`signal to then bypass the filter.
`Basic to the disclosure of the above identi?ed patent
`Basic to the disclosure of the above identified patent
`is the necessity for recognition, in an input signal, of
`is the necessity for recognition, in an input signal, of
`pauses between speech intervals. When a pause contain
`pauses between speech intervals. ·When a pause contain(cid:173)
`in0 noise of the type described is recognized, the param
`ing noise of the type described is recognized, the param(cid:173)
`eters of the noise within the pause are identi?ed and
`eters of the noise within the pause are identified and 50
`used to
`just the parameters of the ?lter. Subsequently,
`used to adjust the parameters of the filter. Subsequently,
`the input signal is applied to the ?lter which is effective
`the input signal is applied to the filter which is effective
`to ?lter the noise from subsequent speech intervals.
`to filter the noise from subsequent speech intervals.
`
`i‘d'lJANS FER ADAPTIVELY
`METHUD
`METHOD AND MEANS FOR ADAiPJrIVELY
`NEAR-STATIGNARY NQZSE FROfs/ll
`FiLTERING NEAR-§'fATlIONARY NOISE FROM
`Alb-1 ZNFGRMATIUN BEARING SIGNAL
`AN INFORMA'rION BEARING SIGNAL
`
`SUMMARY OF THE INVENTION
`SUlvlMARY OF THE INVEI\lTION
`in the present invention, neither recognition of pauses
`In the present invention, neither recognition of pauses
`between speech intervals, nor recognition of noise ter
`between speech intervals, nor recognition of noise ter(cid:173)
`mination, is required; and the hardware for carrying out
`mination, is required; and the hardware for carrying out
`the present invention is extremely simple and fast.
`the present invention is extremely simple and fast.
`Briefly, identi?cation is performed in the frequency
`Briefly, identification is performed in the frequency
`domain alone allowing a spectral ?lter to be used to
`domain alone allowing a spectral filter to be used to
`achieve the desired ?ltering of the noise of the type
`achieve the desired filtering of the noise of the type
`described.
`described.
`An input signal containing noise of the type described
`An input signal containing noise of the type described
`and information, such as speech or music, is applied in
`and information, such as speech or music, is applied in
`parallel to a noise-analysis circuit that includes a plural
`parallel to a noise-analysis circuit that includes a plural(cid:173)
`ity of bandpass ?lters covering the range of frequencies
`ity of bandpass filters covering the range of frequencies
`associated with the information. The absolute value, or
`associated with the information. The absolute value, or
`
`RTL345-1_1028-0005
`
`

`
`15
`15
`
`20
`
`4,185,168
`4,185,168
`4
`4
`3
`3
`but it is sufficient for the bandpass of the ?lters to ap
`but it is sufficient for the bandpass of the filters to ap(cid:173)
`to approximate a Kalman filter as described in page 35,
`to approximate a Kalman ?lter as described in page 35,
`proximate the ideal situation illustrated in FIG. 2.
`fjroximate the ideal situation illustrated in FIG. 2.
`Transactions ASME, Journal of Basic Engineering
`Transactions ASME, Journal of Basic Engineering
`- A detector 140 in a band downstream of a bandpass
`A detector 140 in a band downstream of a bandpass
`(1960). In such a case, filtering will be facilitated even if
`(1960). In such a case, ?ltering will be facilitated even if
`?lter 130 converts the signal passed thereby into the
`filter 130 converts the signal passed thereby into the
`the noise is white over the whole range of frequencies
`the noise is white over the whole range of frequencies
`5 absolute value of such signal, or a function thereof, as
`absolute value of such signal, or a function thereof, as
`of interest.
`of interest.
`for example, the RMS value of the signal. The absolute
`for example, the RMS value of the signal. The absolute
`The bandpass filters used in the present invention
`The bandpass ?lters used in the present invention
`value signal is passed through an averaging circuit 150
`value signal is passed through an averaging circuit 150
`may be digital filters, or discrete analog filters employ(cid:173)
`may be digital ?lters, or discrete analog ?lters employ
`that smooths the signal, preferably by producing a run
`that smooths the signal, preferably by producing a run(cid:173)
`ing microelectronic tapped analog delays, or conven(cid:173)
`ing microelectronic tapped analog delays, or conven
`ning average of the absolute value. The resultant
`ning average of the absolute value. The resultant
`tional analog filters. The combination of analog and
`tional analog ?lters. The combination of analog and
`smoothed signal is in analog form, and is converted to
`10 smoothed signal is in analog form, and is converted to
`logic hardware can also be utilized.
`logic hardware can also be utilized.
`digital form by analog-to-analog convertor 160.
`digital form by analog-to-analog convertor 160.
`The output of each band is then applied to micro
`The output of each band is then applied to micro-
`processor 170 of circuit 102. The microprocessor can be
`processor 170 of circuit 102. The microprocessor can be
`programmed to process the applied data for the purpose
`programmed to process the applied data for the purpose
`of detecting the presence of noise in the input signal,
`of detecting the presence of noise in the input signal,
`identifying the noise or noise and information parame-
`identifying the noise or noise and information parame
`ters if noise is detected, and adjusting the gain of each of
`ters if noise is detected, and adjusting the gain of each of '
`the bandpass ?lters 180 of circuit 104 in a way that
`the bandpass filters 180 of circuit 104 in a way that
`reduces the noise in the output of this circuit. Alterna
`reduces the noise in the output of this circuit. Alterna(cid:173)
`tively, the outputs of bandpass ?lters 130 could be ap
`tively, the outputs of bandpass filters 130 could be ap(cid:173)
`plied directly to the microprocessors 170 through an
`plied directly to the microprocessors 170 through an
`ADC unit. In such case, the microprocessor would be
`ADC unit. In such case, the microprocessor would be
`programmed to compute the absolute value, smooth the
`programmed to compute the absolute value, smooth the
`same, and then examine the data to determine whether
`same, and then examine the data to determine whether
`noise is present in the input signal before identifying the
`noise is present in the input signal before identifying the
`parameters, etc.
`parameters, etc.
`The presence of noise in the input signal is assumed
`The presence of noise in the input signal is assumed
`when the smoothed signal in any band is such that a
`when the smoothed signal in any band is such that a
`threshold is exceeded by all or by a predetermined
`threshold is exceeded by all or by a predetermined
`percentage of successive minima of the smoothed sig
`percentage of successive minima of the smoothed sig(cid:173)
`nal. For example, the criterion for deciding if noise is
`nal. For example, the criterion for deciding if noise is
`present may be whether two out of three successive
`present may be whether two out of three successive
`minima exceed a threshold in a given band. This test for
`minima exceed a threshold in a given band. This test for
`noise is easily carried out by detecting minima in the
`noise is easily carried out by detecting minima in the
`smoothed signal in a band, storing three successive
`smoothed signal in a band, storing three successive
`minima and comparing the value of each with a thresh
`minima and comparing the value of each with a thresh-
`old. If less than two exceed the threshold, the next
`old. If less than two exceed the threshold, the next
`minimum produced replaces the ?rst minimum stored in
`minimum produced replaces the first minimum stored in
`this group, and a new threshold test is made. Reference
`this group, and a new threshold test is made. Reference
`can be made to FIG. 3A which represents, in idealized
`can be made to FIG. 3A which represents, in idealized
`form, the output in one band of averaging circuit 150,
`form, the output in one band of averaging circuit 150,
`i.e., a smoothed signal whose variation with time is due
`Le, a smoothed signal whose variation with time is due
`solely to information in the input signal. As indicated in
`solely to information in the input signal. As indicated in
`FIG. 3B, noise begins between intervals 10 and 12 on
`FIG. 3B, noise begins between intervals 10 and 12 on
`the time scale. Microprocessor 170 might sample the
`the time scale. Microprocessor 170 might sample the
`smoothed signal at the times indicated in the time scale
`smoothed signal at the times indicated in the time scale
`in FIG. 3B with the result that by interval 15, three
`in FIG. 3B with the result that by interval 15, three
`minima would have occurred, namely the minima at the
`minima would have occurred, namely the minima at the
`second, ninth, and fifteenth time interval. At that in(cid:173)
`second, ninth, and ?fteenth time interval. At that in
`stant, three registers (not shown) would contain the
`stant, three registers (not shown) would contain the
`values A V(2), A V(9) and A V(15) as indicated. A com-
`values AV(2), AV(9) and AV(15) as indicated. A com
`parison of these values with the threshold would show
`parison of these values with the threshold would show
`that only one out of the three minima exceeds the
`that only one out of the three minima exceeds the
`threshold. Under the criterion established above, a deci(cid:173)
`threshold. Under the criterion established above, a deci
`sion would be reached at the 15th interval that no noise
`sion would be reached at the 15th interval that no noise
`is present in the input. When the next minima occurs at
`is present in the input. When the next minima occurs at
`interval 20, the three registers in the microprocessor
`interval 20, the three registers in the microprocessor
`would contain the minima A V(9), A V(15) and A V(20)
`would contain the minima AV(9), AV(15) and AV(20)
`associated with the ninth, fifteenth, and twentieth inter(cid:173)
`associated with the ninth, ?fteenth, and twentieth inter
`vals. When the test is made at the twentieth interval,
`vals. When the test is made at the twentieth interval,
`two out of the three amplitudes would exceed the
`two out of the three amplitudes would exceed the
`threshold with the result that the microprocessor would
`threshold with the result that the microprocessor would
`conclude that noise was present in the input signal.
`conclude that noise was present in the input signal.
`Alternative to the above criterion, a value of the
`Alternative to the above criterion, a value of the
`inverse signal-to-noise ratio (SNR -1) in a band in ex(cid:173)
`inverse signal-to-noise ratio (SNR—1) in a band in ex
`cess of a threshold can be used as a test for the presence
`cess of a threshold can be used as a test for the presence
`of noise in the input signal. In order to compute
`of noise in the input signal. In order to compute
`SNR -1, the smoothed signal in a band is sampled at
`SNR—l, the smoothed signal in a band is sampled at
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`BRIEF DESCRIPTION OF THE DRAWINGS
`Embodiments of the present invention are illustrated
`Embodiments of the present invention are illustrated
`by way of example in the accompanying drawings
`by Way of example in the accompanying drawings
`wherein:
`wherein:
`FIG. 1 is a block diagram of the preferred form of an
`FIG. 1 is a block diagram of the preferred form of an
`adaptive filter according to the present invention;
`adaptive ?lter according to the present invention;
`FIG. 2 is a frequency diagram showing the different
`FIG. 2 is a frequency diagram showing the different
`bands into which an input signal containing information
`bands into which an input signal containing information
`.
`. d· ·d d
`an nOIse IS
`IVI e ;
`d
`and noise is divided;
`FIG. 3A is a plot showing the time-variation of a
`FIG. 3A is a plot showing the time-variation of a
`running average of the absolute value ofa signal in a
`running average of the absolute value of ‘a signal in a
`band in the absence of noise;
`band in the absence of noise;
`FIG. 3B is a plot like FIG. 3A showing the vertical 25
`FIG. 3B is a plot like FIG. 3A showing the vertical
`shifting of the running average of the absolute value of
`shifting of the running average of the absolute value of
`a signal when near-stationary noise is present;
`a signal when near-stationary noise is present;
`FIG. 4 is a block diagram of a second form of an
`FIG. 4 is a block diagram of a second form of an
`adaptive filter according to the present invention;
`adaptive ?lter according to the present invention;
`FIG. 5 is a detail of one of a typical notch-filter of the 30
`FIG. 5 is a detail of one of a typical notch-?lter of the
`noise-elimination circuit shown in FIG. 4;
`noise-elimination circuit shown in FIG. 4;
`FIG. 6 is a block diagram of a threshold logic circuit
`FIG. 6 is a block diagram of a threshold logic circuit
`shown in FIG. 4; and
`shown in FIG. 4; and
`FIG. 7 is a block diagram of a hearing aid incorporat(cid:173)
`FIG. 7 is a block diagram of a hearing aid incorporat
`ing the first embodiment of the adaptive filter but pro- 35
`ing the ?rst embodiment of the adaptive ?lter but pro
`viding for suppression of acoustic feedback.
`viding for suppression of acoustic feedback.
`DETAILED DESCRIPTION
`DETAILED DESCRIPTION
`Referring now to the drawing, reference numeral 100
`Referring now to the drawing, reference numeral 100
`designates the preferred embodiment of an adaptive 40
`designates the preferred embodiment of an adaptive
`filter according to the present invention. Filter 100
`?lter according to the present invention. Filter 100
`comprises a noise-analysis circuit 102 and noise-reduc(cid:173)
`comprises a noise-analysis circuit 102 and noise-reduc
`tion circuit 104. Each of circuits 102 and 104 comprises
`tion circuit 104. Each of circuits 102 and 104 comprises
`a plur