AC-center clipper for noise and echo suppression in a

AC-center clipper for noise and echo suppression in a
US006301357B1
(12)
United States Patent
(10) Patent N0.:
US 6,301,357 B1
(45) Date of Patent:
*Oct. 9, 2001
Romesburg
(54)
AC-CENTER CLIPPER FOR NOISE AND
5,796,819 *
ECHO SUPPRESSION IN A
6,160,886 * 12/2000 Romesburg et al. .............. .. 379/410
COMMUNICATIONS SYSTEM
8/1998 Romesburg ........................ .. 379/410
* Cited by examiner
(75) Inventor: Eric Douglas Romesburg, Chapel Hill,
NC (Us)
(73) Assignee: Ericsson Inc., Research Triangle Park,
NC (US)
Primary Examiner_wing F Chan
(74) Attorney, Agent, or Firm—Burns, Doane, Swecker &
Mathls’ L'L'P'
(57)
ABSTRACT
( * ) Notice:
Subject to any disclaimer, the term of this
patent is extended or adjusted under 35
echo_ and noise Suppression device _for Processing an
information signal to suppress unwanted signal components.
U_S_C_ 154(k)) by 0 days
In an exemplary embodiment, the echo suppression device
This patent is subject to a terminal dis_
Chimes
(21)
Appi NO _ 08/77 5 797
'
"
(22) Filed:
’
the unwanted components are suppressed even when the
Dec. 31, 1996
information signal contains a signi?cant noise component.
7
In another exemplary embodiment, a mobile radio station is
(51)
Int‘ Cl‘
(52)
US Cl- ~~~~~~~~~~~~~~~~ "
"
""""""""""" " H04B 3/20
379/406-06; 379/406~01
Of Search ................................... ..
(56)
provided including a microphone for receiving a near-end
audio input and for producing a near-end audio output which
is to be transmitted to a far-end user_ The exemplary mobile
379/407, 408, 409, 411, 389, 390, 40601,
40605, 406-08; 455/422, 426, 439; 375/232
station also includes a loudspeaker for broadcasting, to a
near-end user of the mobile station, a far-end audio signal
which is generated by the far-end user and then received at
the mobile station. An echo suppression circuit within the
References Cited
5,263,019
mcludes an AC-center clipper havmg a chppmg wmdow
with an adjustable window center. A clipping threshold of
the AC-center clipper is set to attenuate the unwanted signal
components, and the center of the clipping window is varied
in accordance with the value of the information signal so that
Us PATENT DOCUMENTS
mobile station is used to attenuate an echo component of the
* 11/1993
near-end audio signal which results from the mobile station
microphone receiving output from the mobile station loud
5J274J7O5
ch11 ------------- -
12/1993 Younce ct a1~
i
379/406
379/410
(8351mm """"" "
5j5s7j99s * 12/1996 Velardo, Jr. et a1. .
379/410
5,668,831 * 9/1997 Claydon et al.
375/232
5,721,771 * 2/1998 Higuchi et al. .................... .. 379/410
40
1
10
///—/ 70 I)
11/0 70 //
20
speaker. The echo suppression circuit includes an AC-center
clipper having a clipping window with an adjustable clip
pmg Center‘
41 Claims, 9 Drawing Sheets
U.S. Patent
0a. 9, 2001
Sheet 2 0f 9
US 6,301,357 B1
U.S. Patent
0a. 9, 2001
0.5
Sheet 6 0f 9
0
US 6,301,357 B1
-0.5—
U.S. Patent
0a. 9, 2001
Sheet 7 0f 9
US 6,301,357 B1
.w
m.
_
_
_
_
_
_
_
_
_
_
5%
10/71/700” UJZ/W/"MA’
pgkQ
U.S. Patent
0a. 9, 2001
Sheet 9 0f 9
US 6,301,357 B1
_
_
r
_
_
_
_
_
_
hem
US 6,301,357 B1
1
2
AC-CENTER CLIPPER FOR NOISE AND
ECHO SUPPRESSION IN A
COMMUNICATIONS SYSTEM
contexts. For example, as is described in more detail beloW,
residual echo suppression circuits Within such systems may
be relatively ineffective When ambient noise arises at the
microphone input. Ambient noise is commonplace and may
BACKGROUND
The present invention relates to communications systems,
and more particularly, to echo and noise suppression in a
bi-directional communications link.
In many communications systems, for example landline
and Wireless telephone systems, voice signals are often
occur, for example, due to road and traffic noise in the case
of an automobile telephone. Therefore, it Would be advan
tageous if a system Were available in Which all of the echo
suppression features of the system could function effectively
even in the presence of ambient noise.
10
Additionally, certain aspects of available systems are not
transmitted betWeen tWo system users via a bi-directional
communications link. In such systems, speech of a near-end
user is typically detected by a near-end microphone at one
end of the communications link and then transmitted over
the lin to a far-end loudspeaker for reproduction and pre
sentation to a far-end user. Conversely, speech of the far-end
user is detected by a far-end microphone and then transmit
ted via the communications link to a near-end loudspeaker
for reproduction and presentation to the near-end user. At
Well suited for double-talk situations in Which a near-end
user and a far-end user are speaking simultaneously. For
15
example, because residual echo suppression circuits Within
available systems may intolerably distort near-end signals
from a far-end-user perspective, they are typically deacti
vated during double-talk situations. By deactivating all or
part of the echo suppression features, hoWever, a conven
tional system may be susceptible to other problems. For
example, as is described in more detail beloW, the echo
either end of the communications link, loudspeaker output
detected by a proximate microphone may be inadvertently
suppression provided by a conventional system employing a
adaptive-?lter echo canceler may be insufficient, absent
transmitted back over the communications link, resulting in
residual echo suppression, due to non-linearities introduced
by the components used to process information signals.
What may be unacceptably disruptive feedback, or echo,
from a user perspective. Furthermore, if the round-trip gain
of a near-end microphone is greater than unity at any aiblq
Thus, it Would be advantageous if a system Were available
frequency, then the system Will tend to “bowl” as is Well
knoWn in the art.
Therefore, in order to avoid transmission of such unde
could be used even during double-talk situations. In sum,
there is a real need for an improved technique for suppress
sirable echo signals, microphone input should be isolated
ing echo signals in a tWo-Way communications link.
from loudspeaker output. With a conventional telephone
handset, in Which the handset microphone is situated close
to the user’s mouth While the handset speaker essentially
covers the user’s ear, the requisite isolation is easily
achieved. HoWever, as the physical siZe of portable tele
phones has decreased, and as hands-free speaker-phones
in Which all of the echo suppression aspects of the system
SUMMARY
35
The present invention ful?lls the above-described and
other needs by providing an echo suppression device for
processing an input signal containing a time-varying pri
have become more popular, manufacturers have moved
toWard designs in Which a microphone and a loudspeaker
may be situated physically close to one another, yet rela
tively far aWay from the user. As a result, the need for more
mary component and a time-varying secondary component
to produce an output signal in Which the time-varying
secondary component is substantially suppressed. In an
exemplary embodiment, the echo suppression device
sophisticated echo suppression techniques has become para
includes an input node for receiving the input signal, and a
center clipper connected to the input node for processing the
mount in modern systems.
The need is particularly pronounced in the case of hands
free automobile telephones, Where the closed vehicular
input signal to produce the output signal. The center clipper
employs a clipping WindoW having a variable center and a
environment can cause multiple re?ections of a loudspeaker 45 clipping threshold Which is set to attenuate the secondary
signal to be coupled back to a high-gain hands-free micro
phone. Movement of the vehicle and changes in the relative
directions and strengths of the user and echo signals, for
component of the input signal.
In another exemplary embodiment, a mobile station
includes a microphone for receiving a near-end audio input
and for producing a near-end audio output Which is to be
example as WindoWs are opened and closed or as the user
moves his head While driving, further complicate the task of
echo suppression in the automobile environment.
transmitted to a far-end user. The exemplary mobile station
also includes a loudspeaker for broadcasting, to a near-end
user of the mobile station, a far-end audio signal Which is
Additionally, more recently developed digital telephones
process speech signals through vocoders Which introduce
signi?cant signal delays and create non-linear signal distor
tions. As is Well knoWn, these prolonged delays tend to
magnify the problem of signal echo from a user perspective,
55
generated by the far-end user and received at the mobile
station. In the mobile station, an echo suppression circuit is
used to attenuate an echo component of the near-end audio
signal resulting from the microphone receiving output from
and the additional non-linear distortions can make echo
suppression difficult.
Traditionally, echo suppression has been accomplished
using echo canceling circuits designed to approximate and
the loudspeaker. The echo suppression circuit includes an
AC-center clipper having a clipping WindoW With an adjust
able clipping center.
remove echo signals from microphone output so that only
near-end speech is transmitted over the communications
link. These systems are described, for example, in US. Pat.
The above described and additional features of the present
invention are explained in greater detail hereinafter With
reference to the illustrative examples shoWn in the accom
panying draWings. Those skilled in the art Will appreciate
No. 5,475,731, Which is incorporated herein by reference.
While the systems described in the cited reference are
generally effective in suppressing echo signals, certain
aspects of those systems make them impractical in some
65
that the described embodiments are provided for purposes of
illustration and understanding and that numerous equivalent
embodiments are contemplated herein.
US 6,301,357 B1
3
4
BRIEF DESCRIPTION OF THE DRAWINGS
speaker 20 and picked up by the microphone 10, and a noise
signal, corresponding to ambient noise existing at the mobile
FIG. 1 depicts a conventional echo suppression circuit.
FIG. 2 depicts a conventional echo canceling system
employing a digital ?lter and a residual echo suppression
circuit.
FIG. 3 depicts operation of a conventional center clipper
employed, for example, in the system of FIG. 2.
FIG. 4 depicts operation of an AC-center clipper con
structed in accordance With the teachings of the present
invention.
station site. Where the mobile station is a mobile telephone
located in an automobile, sources of ambient noise include
traffic, movement by the near-end user, and movement of the
microphone itself (e.g., When the microphone is attached to
a sun visor).
As described above, echo signals can be extremely both
ersome to users of a communications system. In fact, if echo
signals are alloWed to pass unattenuated betWeen users of a
communications system, the system may be virtually unus
able in many real World applications. Therefore, conven
tional echo suppression circuitry, such as that depicted in
FIG. 1, has been used to prevent echo signals from passing
FIG. 5 depicts an exemplary embodiment of an echo
suppression system constructed in accordance With the
teachings of the present invention.
FIG. 6 depicts a second exemplary embodiment of an
15
The con?guration of FIG. 1 implements a Well-knoWn and
echo suppression system constructed in accordance With the
teachings of the present invention.
straightforWard approach to echo suppression in Which the
speech of only one user is transmitted at any given time. In
other Words, When the near-end user is speaking, the trans
mit path of the far-end user is attenuated or muted, and vice
FIG. 7 depicts a third exemplary embodiment of an echo
suppression system constructed in accordance With the
teachings of the present invention.
FIG. 8 depicts a far-end audio signal Which may be
versa.
received, for example, at a near-end mobile station trans
ceiver employed in the embodiment of FIG. 6.
FIG. 9 depicts a near-end audio signal Which may arise,
for example, at an output of a microphone employed in the
embodiment of FIG. 6.
FIG. 10 depicts an output audio signal Which may arise,
for example, at an output of an AC-center clipper employed
in the embodiment of FIG. 6.
FIG. 11 depicts a noisy near-end audio signal Which may
arise, for example, at an output of a microphone employed
in the embodiment of FIG. 6.
FIG. 12 depicts an output audio signal Which may arise,
for example, at an output of an AC-center clipper employed
in the embodiment of FIG. 6.
25
For example, When a voice signal from a far-end user is
received at the radio transceiver 30 of FIG. 1, the voice
activity detector 140 senses that the far-end user is speaking
and indicates that fact to the decision block 120. The
decision block 120 then controls the attenuator 110 to
attenuate, or even mute, the audio signal output by the
microphone 10, thereby preventing the near-end input from
being transmitted to the far-end user. Similarly, an output of
the voice activity detector 100 indicates Whether or not the
near-end user of the mobile station is speaking. If so, then
the decision block 120 controls the attenuator 130 to
attenuate, or mute, sound transmissions from the far-end
35
user. In this Way, neither the near-end user, nor the far-end
user, hears a delayed echo of his or her oWn voice When
speaking. HoWever, the system of FIG. 1 may create abrupt
transitions in transmitted signals and may unsettlingly cut
off the speech of a far-end user When a near-end user
DETAILED DESCRIPTION
inadvertently makes noise or simply intends to reassure the
far-end user With a quick “OK”. Therefore, the system of
FIG. 1 may be unsatisfactory in many contexts.
To alleviate the problems associated With the conven
FIG. 1 depicts a conventional echo suppression circuit
implemented, for example, in a mobile station in a cellular
radio communications system. As shoWn in FIG. 1, a radio
transceiver 30 is connected to an antenna for receiving and
transmitting information signals to and from the mobile
station, respectively. In the receive path, the radio trans
back and forth betWeen users of a communications system.
tional system of FIG. 1, more sophisticated echo suppression
circuits have been developed. For example, FIG. 2 depicts
45
ceiver 30 is connected to a voice activity detector 140 Which
is in turn connected to an attenuator 130. An output of the
attenuator 130 is connected to a loudspeaker 20. In the
transmit path, a microphone 10 is connected to a voice
activity detector 100 Which is in turn connected to an
attenuator 110. An output of the attenuator 110 is connected
to an input of the radio transceiver 30. Adecision block 120
an echo suppression circuit such as that disclosed in US.
Pat. No. 5,475,731. As shoWn, an echo suppression circuit
210 includes an echo canceler, comprising a ?lter 280 and a
summing device 250, as Well as a residual echo suppressor
225, comprising a center clipper 205 and an envelope
detector 215. As shoWn in FIG. 2, the echo suppression
circuit 210 may be incorporated into a mobile station in a
cellular radio communications system. In operation, a voice
receives inputs from the voice activity detectors 100, 140
signal from a far-end user is received at a transceiver of the
and produces outputs Which are fed to the attenuators 110,
130.
In operation, an information signal received at the radio
transceiver 30, corresponding to a voice signal transmitted
by a far-end user in the communications system, is pro
mobile station (not shoWn). Depending upon the type of
55
the mobile station transceiver output is assumed to be
digital. As shoWn, a received signal is passed through an
optional receive digital signal processor 290 to produce a
cessed and broadcast to a near-end user of the mobile station
via the loudspeaker 20. At the same time, sound picked up
at the microphone 10 is processed and transmitted to the
far-end user via the radio transceiver 30. The audio signal
output by the microphone 10 may consist of several com
ponents. For example, it may contain a primary component,
corresponding to speech of the near-end user of the mobile
station, as Well as secondary components Which may include
an echo signal, corresponding to sound output by the loud
transmission channel used, information signals received at
the transceiver, and then passed to the circuitry of FIG. 2,
may be analog or digital. In the circuit of FIG. 2, hoWever,
digital loudspeaker signal LD. The digital loudspeaker signal
LD is then passed through a digital-to-analog converter 240
to produce an analog loudspeaker signal LA. The analog
65
loudspeaker signal LA is in turn used to drive the loud
speaker 20 to produce a loudspeaker audio signal AL. At the
same time, a near-end audio signal AM is picked up at the
microphone 10 producing an analog microphone signal MA.
US 6,301,357 B1
5
6
The microphone audio signal MA is passed through an
1-to-1. Thus, loW-amplitude error signals are clipped, and
residual echo suppression is achieved.
As shoWn in FIG. 2, the output 270 of the ?lter 280 is
input to the envelope detector 215 and an output of the
envelope detector 215 is in turn fed to the center clipper 205.
The output of the envelope detector 215 is used to adjust the
clipping thresholds 310, 320 of the center clipper 205 such
that the center clipper is active only When the far-end user
is speaking. In other Words, the clipping thresholds 310, 320
analog-to-digital converter 220 to produce a digital micro
phone signal MD. As described above, the near-end audio
signal AM, as Well as the digital microphone signal MD, may
comprise several signal components, including near-end
voice, near-end noise, and far-end echo. The echo suppres
sion circuit 210 is used to cancel the far-end echo component
of the microphone signal MD.
As shoWn in FIG. 2, the microphone signal MD is input to
one node of the summing device 250. At the same time, the
are increased and decreased as the output of the ?lter 280
digital loudspeaker signal LD is fed through the ?lter 280,
increases and decreases, respectively. This is necessary so
that important loW-amplitude components of the micro
and an output 270 of the ?lter 280 is input to another node
phone signal MD are not suppressed When the near-end user
is spealing and the far-end user is silent.
of the summing device 250. An output 260 of the summing
device 250 is fed to the center clipper 205, and at the same
time is fed back, as described beloW, to the ?lter 280. The
output 270 of the ?lter 280 is also input to the envelope
detector 215, and an output of the envelope detector 215 is
input to the center clipper 205. An output of the center
clipper 205 is passed to the mobile station transceiver (not
shoWn). The ?lter 280 is a multiple-tap ?lter, as is Well
knoWn in the art, having a transfer function, or impulse
response, approximating that of a path from the loudspeaker
20 to the microphone 10. The true transfer function associ
ated With the path from the loudspeaker 20 to the micro
phone 10 is a function of frequency and Will depend upon,
among other things, the relative physical placement of the
loudspeaker 20 and the microphone 10, as Well as the
position of the near-end user of the mobile station. Thus, the
transfer function of the ?lter 280 should be continually
updated. Assuming that the transfer function of the ?lter 280
15
the near-end user and the far-end user are speaking simul
taneously (i.e., during double-talk). Such loW-amplitude
clipping during double-talk may be unacceptable in many
applications, and therefore conventional systems typically
deactivate the residual echo suppression during double-talk.
As described above, hoWever, doing so may not be desirable.
Note also that the residual echo suppression circuit 225 of
FIG. 2 may fail should the residual echo signal be super
imposed on other components of the microphone signal
25
does represent a reasonable approximation of the true trans
fer function, then the output 270 of the ?lter 280 Will
represent a close approximation of the echo component of
MD.In other Words, noise or near-end voice components of
the microphone signal MD may cause the error signal 260 to
have an amplitude Which lies outside the clipping thresholds
310, 320 of the center clipper 205. If so, then the residual
echo suppressing circuit 225 of FIG. 2 Will be ineffective.
Advantageously, the present invention teaches methods
and apparatus for suppressing noise and echo components in
information signals, Wherein the problems described above
the microphone signal MD. Therefore, the output 260 of the
summing device 250 Will represent an echo-canceled ver
Note hoWever, that the important loW-amplitude portions
of the near-end voice signal are suppressed, or clipped, When
35
sion of the microphone signal MD. Assuming loW near-end
noise, and assuming that the near-end user is not speaking,
the output 260 of the summing device 250 should then be
With respect to FIGS. 1, 2, and 3 are overcome. More
particularly, the present invention teaches a center clipper
having a clipping WindoW With an adjustable WindoW center.
Such a center clipper may properly be referred to as an
AC-center clipper, in that the clipping WindoW of such a
center clipper may be set to track a time-varying, or
Zero Whether or not the far-end user is speaking. Therefore,
alternating-current (AC), input signal. Such an AC center
the output 260 of the summing device 250 is sometimes
clipper may be implemented, for example, by the folloWing
referred to in the art as an error signal. The error signal 260
pseudocode:
may be used, as in knoWn in the art, to update ?lter
coef?cients of the ?lter 280 such that the transfer function of
the ?lter 280 represents a reasonable approximation of the
true transfer function betWeen the loudspeaker 20 and the
microphone 10, even When the true transfer function is
aciclip (AC-center clipper)
45
aciclip(input, delta, oldiout) returns an
AC-center clipped version of the input value.
changing (e.g., due to movement of the near-end user).
The echo cancellation provided by the ?lter 280 and the
summing device 250 may be insufficient in certain applica
tions. For example, in practice the transfer function of the
input = input value
delta = +/— clipping threshold
(i.e., 1/2 of clipping Window)
oldiout = output from the last call to aciclip
The center of the clipping Window starts at the
?lter 280 may never fully converge to the true transfer
initial value of the input signal and remains there,
function due to non-linearities in the loudspeaker 20 and
other components Which are used to process signals (e.g.,
as does the output of the AC-center clipper, until
the input signal moves outside the clipping Window.
At that time, the center of the clipping Window,
signal converters, ampli?ers, or transducers). Therefore,
residual echo suppression, in addition to that provided by the
?lter 280 and the summing device 250, may be necessary. In
the circuit of FIG. 2, residual echo suppression is provided
by the center clipper 205 and the envelope detector 215, both
and the output of the AC-center clipper, are shifted
55
to roughly track the amplitude of the input signal.
function neWiout=aciclip(input, delta, oldiout)
of Which are Well knoWn in the art. Operation of the center
if oldiout<input—delta
clipper 205 is depicted in FIG. 3 by an output-versus-input
function 300. When the amplitude of the signal 260, Which
elseif oldiout>input+delta
neWiout=input—delta;
neWiout=input+delta;
is input to the center clipper, lies Within a clipping WindoW
else
de?ned by the clipping thresholds 310, 320, the output of the
center clipper 205 remains Zero. HoWever, When the ampli
tude of the signal 260, Which is input to the center clipper
205, exceeds the clipping thresholds 310, 320, the output of
the center clipper 205 tracks the input of the center clipper
I1€WiOllI=OldiOll?
65
end
An example of the behavior of an AC-center clipper, as
implemented using the above listed pseudocode to move the
US 6,301,357 B1
8
7
center of the clipping WindoW, is depicted in FIG. 4. In FIG.
Thus, FIG. 6 depicts an exemplary embodiment of the
4, a solid line 410 represents an information signal com
prising a summation of a ?rst sinusoid having a period of 10
units and an amplitude of 25 units and a second sinusoid
having a period of 100 units and an amplitude of 100 units.
present invention in Which an AC-center clipper, used to
suppress noise and echo components of a near-end voice
signal, is controlled in a manner Which minimiZes impact on
In the simulation depicted in FIG. 4, the clipping threshold
output from a microphone 10 is input to an AC-center
a near-end speech signal. As shoWn, a microphone signal M
of the AC-center clipper is ?xed at 25 units. The information
clipper 40. An output of the AC-center clipper 40 is coupled
signal 410 may represent, for example, a microphone signal
MD such as that depicted in FIG. 2. The ?rst, loW-frequency
to a mobile station transceiver (not shoWn). A received
sinusoidal component of the information signal 410 may
represent, for example, loW-frequency noise or voice signals
picked up at the microphone 10 of FIG. 2. The second,
higher-frequency sinusoidal component of the information
signal 410 may represent, for example, an echo signal
to a loudspeaker 20. As in the system of FIG. 5, the
generated by the loudspeaker 20 and picked up by the
signal L, output from the mobile station transceiver, is input
microphone signal M and the loudspeaker signal L may be
analog or digital, as appropriate, and may be normaliZed to
a range of —1 to 1.
In FIG. 6, the loudspeaker signal L is input to an envelope
15
exponential-decay peak detector. A time constant of the
microphone 10 of FIG. 2.
detector 50 is set such that the decay rate of an output P of
the detector 50 is no faster than the decay rate associated
With an acoustic path betWeen the loudspeaker 20 and the
In FIG. 4, a dashed line 400 represents an output of the
AC-center clipper implemented using the above listed
pseudocode. As shoWn, the high-amplitude, loW-frequency
component of the information signal passes through the
AC-center clipper, While the loW-amplitude, high-frequency
component of the information signal is substantially sup
pressed by the AC-center clipper. Thus, an echo signal
component “riding on top of” another signal component is
easily attenuated by the AC-center clipper. As described
beloW, this aspect of the AC-center clipper can be used
detector 50 Which may be constructed, for example, as an
microphone 10. In other Words, a peak in the detector output
P, Which is induced by a peak in the loudspeaker signal L,
should fall off no faster than the corresponding echo signal
(including reverberations) Which is picked up at the micro
phone 10. The decay rate of the peak detector should not be
25
made so sloW, hoWever, that near-end transmissions are
signi?cantly distorted. In FIG. 6, the detector output P,
advantageously, not only to suppress echo signals even in
the presence of noise, but also to achieve residual echo
suppression even during double-talk situations.
By Way of contrast, a conventional center clipper such as
that shoWn in FIG. 2 Would fail to suppress the echo signal
of FIG. 4 should the overall amplitude of the information
Which may also be normaliZed to a range of 0 to 1, is coupled
to a multiplier 600 Where it is multiplied by a parameter
Hpeak, Which is an estimator of the true transfer function H
signal 410 exceed the ?xed-center clipping WindoW of the
upon testing performed on experimental systems or,
alternatively, may be set for each particular system during
of the acoustic path from the loudspeaker 20 to the micro
phone 10. The transfer function estimator HpmK, as Well as
the time constant of the peak detector, may be preset based
conventional center clipper. Thus, in order to achieve the
level of echo suppression depicted in FIG. 4, the clipping
35
WindoW of the conventional center clipper Would have to be
installation and calibration. An output A of the multiplier
600 is used as a clipping threshold for the AC-center clipper
40.
In operation, When a far-end user of the system of FIG. 6
made large enough to capture both the echo signal and the
loW-frequency signal component. Doing so, hoWever, Would
so severely distort the information signal that it Would be
is silent, the amplitude of the loudspeaker signal L Will be
bothersome, if not intolerable, from a far-end user perspec
tive. As a result, a conventional center clipper cannot be used
effectively, if it can be used at all, during ambient noise or
double-talk situations.
FIG. 5 depicts an AC-center clipper 40 used as a stand
alone noise and echo suppressor in a mobile station 45
Zero, as Will the output P of the envelope detector 50. Thus,
the output A of the multiplier 600 Will be Zero and the
AC-center clipper 40 Will act as a pass-through, having no
effect on the microphone signal M. Alternatively, When the
far-end user is speaking, the loudspeaker signal L Will be
non-Zero, as Will the output P of the envelope detector 50.
Therefore, the output A of the multiplier 600 Will be non
employed, for example, in a cellular radio communications
system. As shoWn, output from a microphone 10 is input to
the AC-center clipper 40, and an output of the AC-center
clipper 40 is passed to a mobile station transceiver (not
shoWn). A clipping threshold A of the AC-center clipper is
Zero, and the AC-center clipper 40 Will behave as Was
described With respect to FIG. 4. In this Way, the AC-center
clipper 40 is active only When necessary. In other Words, it
is active only When the far-end user is speaking. As a result,
set equal to a constant. Output from the mobile station
the AC-center clipper 40 serves to suppress echo in a
transceiver, corresponding to voice signals received at the
single-talk situation in Which only the far-end user is
speaking, but it does not distort the near-end voice signal in
a single-talk situation in Which only the near-end user is
speaking. Note that any DC offset Which may be present in
the far-end signal, or Which may be introduced during
analog-to-digital conversion of the far-end signal, may be
removed from the signal input to the envelope detector 50
transceiver from a far-end user, is coupled to a loudspeaker
20. As shoWn in FIG. 5, a microphone signal M, output from
the microphone 10, and a loudspeaker signal L, input to the
loudspeaker 20, may be normaliZed to lie Within a range of
—1 to 1. Additionally, the clipping threshold A of the
AC-center clipper 40 may also be normaliZed to a range of
0 to 1. It Will be understood that the microphone signal M
and the loudspeaker signal L may be digital or analog, as
appropriate. The AC-center clipper 40 of FIG. 5 behaves as
is described above With respect to FIG. 4, and thus provides
utility as a stand-alone noise and echo suppressor. HoWever,
because the AC-center clipper may introduce some
distortion, as shoWn in FIG. 4, it may be advantageous to
reduce the effects of the AC-center clipper When a far-end
user of the system of FIG. 5 is not speaking.
55
(e.g., by AC-coupling or high-pass ?ltering, as appropriate)
so that the clipping threshold of the AC-center clipper 40 is
not made arti?cially, and unnecessarily, high.
During a double-talk situation, the AC-center clipper 40 is
active due to the non-Zero amplitude of the loudspeaker
65
signal L. Therefore, in addition to suppressing echoes, the
AC-center clipper 40 partially distorts the near-end voice
signal. HoWever, because the distortion introduced by the
AC-center clipper 40 is slight compared to that Which Would
US 6,301,357 B1
10
FIG. 9 depicts a near-end signal Which Would arise, for
example, at the output of the microphone 10 of FIG. 6. As
shoWn in FIG. 9, the near-end signal may comprise alter
nating bursts of near-end and far-end speech. In FIG. 9, a
be introduced by a conventional center clipper having a
clipping WindoW large enough to achieve echo suppression
during double-talk, the AC-center clipper 40 of the present
invention need not be turned off. As a result, the AC-center
clipper 40 may serve as an effective technique for full-time
echo suppression, even in the absence of a “true” echo
canceler comprising an adaptive ?lter such as that shoWn in
FIG. 2. By Way of contrast, a conventional center clipper
cannot be used as effectively, if it can be used at all, in a
far-end burst 900 and a near-end burst 910 are identi?ed by
brackets. The far-end burst 900 represents, for example, an
echo signal component arising at the output of the micro
phone 10 of FIG. 6 due to output of the loudspeaker 20. The
near-end burst 910 represents, for example, a near-end
con?guration such as that depicted in FIG. 6.
The con?guration of FIG. 6 may be simulated, for
speech signal component arising at the output of the micro
phone 10 of FIG. 6 due to a near-end user speaking into the
example, using the folloWing pseudocode, When used in
conjunction With the previously listed pseudocode:
microphone 10.
15
FIG. 10 then depicts a processed signal corresponding, for
example, to the output of the AC-center clipper 40 of FIG.
6. As shoWn, the processed signal may comprise bursts of
near-end speech separated by periods of silence. A burst of
peakidet (peak detector)
peakidet(input, pole, oldiout) returns the peak
magnitude of the input and the decayed previous
near-end speech 1010 and a period of silence 1000 are
identi?ed by brackets in FIG. 10. The near-end burst 1010
corresponds to the near-end burst 910 of FIG. 9. The period
of silence 1000 indicates that the far-end burst 900, present
output value. Thus, the output rises With the
input, but falls off sloWly after an input peak.
input = input value
pole = location in the Z-plane of a real pole
in an IIR exponential-decay ?lter
(note: 0 < pole < 1 for stability)
oldiout = output from the last call to peakidet
in the input to the AC-center clipper 40, has been sup
pressed. In other Words, only the near-end voice bursts are
an AC-center clipped version of the neariend signal.
alloWed to pass to the far-end user, While the far-end bursts,
or echoes, are suppressed. Note that FIGS. 8, 9 and 10
represent a situation in Which there exists little near-end
noise.
FIGS. 11 and 12, by Way of contrast, represent a situation
in Which there exists considerable near-end noise. As
described above, such near-end noise may result, in the case
of a mobile automobile telephone, from road noise or from
physical movement by the near-end user. FIG. 11 depicts an
The clipping Window threshold (or delta) is adjusted
exemplary near-end signal comprising alternating bursts of
25
function neWiout=peakidet(input, pole, oldiout)
neWiout=max(abs(input), pole * oldiout);
echoisup (echo suppressor)
echoisup(fariend, neariend, pole, Hpeak) returns
in accordance With an envelope of the fariend signal.
fariend = far end user signal
35
pole = pole of peak detector
Hpeak = acoustic path estimator
Assuming the near-end signal of FIG. 11 is fed into the
AC-center clipper 40 of FIG. 6, FIG. 12 then depicts an
function [clip, peak]
exemplary output of the AC-center clipper 40. As shoWn, the
AC-center clipper effectively suppresses echo components
=echoisup(fariend, neariend, pole, Hpeak)
if length (fariend)~=length(neariend)
of the near-end signal even in the presence of signi?cant
near-end noise.
In the preceding discussion, it Was assumed that the
error(‘Vectors must be the same length.’) end
N=length(fariend);
oldipeak=0;
oldiclip=0;
near-end and far-end speech, superimposed upon near-end
noise. A far-end burst 1110, a near-end burst 1100, and a
period of noise 1120 are identi?ed in FIG. 11 by brackets.
neariend = near end user signal
45
peak=Zeros(1,N);
clip=Zeros(1,N);
far-end signal Was relatively noise free. Note, hoWever, that
if the far-end signal is noisy, then the envelope detector
output, and thus the clipping threshold of the AC-center
clipper, Will generally be non-Zero even When the far-end
user is not speaking. As a result, the AC-center clipper Will
undesirably introduce a level of distortion on the near-end
for i=1:N
peak(i)=peakidet(fariend(i), pole, oldipeak);
oldipeak=peak(i);
signal in near-end single-talk and no-talk situations.
Therefore, the present invention teaches that it is advanta
clip(i)=ac clip(neariend(i), Hpeak * peak(i), oldi
end
FIGS. 8—12 illustrate the performance of the AC-center
geous in certain contexts to reduce the clipping threshold of
the AC-center clipper by an amount proportional to the level
of far-end noise. For example, in the embodiment of FIG. 6,
the output of the envelope detector 50 can be used to provide
clipper con?guration of FIG. 6. FIG. 8 depicts an exemplary
far-end voice signal Which may be received, for example, at
the output of the multiplier 600 to provide the clipping
the mobile station transceiver of FIG. 6. As shoWn in FIG.
threshold A for the AC-center clipper 40. The indicator of
far-end noise can be derived from the output of the envelope
clip);
oldiclip=clip(i);
55
an indicator of far-end noise Which can be subtracted from
8, the far-end signal may comprise bursts of speech inter
mixed With periods of silence, corresponding to a far-end
user alternately speaking and then Waiting for a response
from the near-end user. One such burst of far-end speech 800
is identi?ed by a bracket in FIG. 8. The far-end signal of
FIG. 8 Would be output, for example, via the loudspeaker 20
of FIG. 6.
detector 50, for example, by gradually increasing a noise
indicator variable from Zero and capping it using the detec
tor output so that it tracks the loWer limit of the far-end
65
signal-envelope. Such an approach is simulated, for
example, using the folloWing modi?cation of the echo
suppression pseudo-code provided above.
US 6,301,357 B1
11
12
combine the noise and echo suppression characteristics of
the AC-center clipper of the present invention with a tradi
tional echo canceler. FIG. 7 depicts an exemplary embodi
ment of such a combination. As shown, a loudspeaker signal
echoisup (echo suppressor)
echoisup(fariend, neariend, pole, Hpeak) returns
L corresponding to a sound transmission received from a
an AC-center clipped version of the neariend signal.
far-end user, is coupled to a loudspeaker 20. A microphone
signal M arising at the output of a microphone 10 is coupled
The clipping window threshold (or delta) is adjusted
in accordance with an envelope of the fariend signal
and reduced by a level of noise in the fariend signal.
to an input of an echo canceler 700, and an output E of the
echo canceler 700 is coupled to an input of an AC-center
fariend = far end user signal
near end = near end user signal
pole = pole of peak detector
10
Hpeak = acoustic path estimator
noise = indicator of fariend noise
scale = multiplier used to increase noise indicator
offset = offset used to increase noise indicator
scale and offset are determined ernperically and set to
establish a desired rate of increase for the fariend
noise indicator. The values shown below cause the
signal L is also coupled to a second input of the echo
canceler 700 and to an input of an envelope detector 50. An
15
noise indicator to rise from Zero to 2A(—8) in about
65536 samples and to double in about 65536 samples
(65536 samples corresponds to about 8 seconds for an
8 kHz sample rate). Note that the time constant is
independent of the magnitude of the fariend noise.
20
function [clip, peak]
=echoisup(fariend, neariend, pole, Hpeak)
if length(fariend)~=length(neariend)
error(‘Vectors must be the same length’) end
25
N=length(fariend);
In operation, the echo canceler 700 behaves as is
described above with respect to FIG. 2. In brief, the echo
canceler 700 comprises a summing device 710, a ?lter 720,
720 is used to produce an estimator of the echo component
30
of the microphone signal M. The estimator is then combined
with the microphone signal M at the summing device 710 to
produce an error signal E. Filter coef?cients of the ?lter 720
are adjusted in time such that an impulse response of the
offset=2A(—24);
peak=Zeros(1,N);
clip=Zeros(1,N);
for i=1:N
output P of the envelope detector 50 is coupled to an input
of a multiplier 750 where it is multiplied by an estimator H.
of the acoustic transfer function H between the loudspeaker
20 and the microphone 10 to produce an output which is in
turn input to a second multiplier 740. A convergence output
C of the echo canceler 700 is input to a MAX logic block
760. Aparameter F is coupled to a second input of the MAX
logic block 760 and an output of the MAX logic block 760
is input to the second multiplier 740. As shown, an output A
of the second multiplier 740 is used as a clipping threshold
for the AC-center clipper 40.
and a least-mean-square (LMS) logic block 730. The ?lter
oldipeak=0;
oldiclip=0;
noise=0;
scale=1+1/65536;
clipper 40. An output of the AC-center clipper 40 is coupled
to a mobile station transceiver (not shown). The loudspeaker
?lter 720 approximates the acoustic transfer function H
existing between the loudspeaker 20 and the microphone 10.
35
As is known in the art, the coef?cients of ?lter 720 may be
peak(i)=peakidet(fariend(i), pole, oldipeak);
oldipeak=peak(i);
updated using the error signal E, in conjunction with an
LMS algorithm, implemented for example in the LMS block
noise=noise * scale+offset;
730.
Because the true transfer function H may change over
time, for example due to changes in noise conditions at the
near-end mobile station, the coef?cients of ?lter 720 are
noise=min(noise, peak(i));
threshold=Hpeak * (peak(i)—noise);
40
clip(i)=aciclip(neariend(i), threshold, oldiclip);
oldiclip=clip(i); end
continually updated. When the mobile station is ?rst pow
Note that when the clipping threshold of an AC-center
clipper is adjusted based on a level of noise in the source
signal (i.e., the echo-producing signal), then AC-center
ered up, or when a relatively stable prevailing transfer
45
clipper circuits can be used effectively to provide network
echo suppression in which both near-end and far-end echoes
are suppressed. In other words, an AC-center clipper circuit
function H. However, as the ?lter coef?cients are updated in
response to the error signal E, the transfer function of the
?lter 720 will converge toward the true transfer function H.
Thus, the echo canceler 700 is said to be converged or
can be used as shown in FIG. 6 to suppress far-end signal
echo from the near-end signal, and an analogous AC-center
clipper circuit can be used to suppress near-end signal echo
from the far-end signal. For example, in an alternative
unconverged depending upon whether the transfer function
of the ?lter 720 is, or is not, a good approximation of the true
transfer function H, respectively.
embodiment, the loudspeaker signal L of FIG. 6 is passed
through a second AC-center clipper (not shown) prior to
being fed to the loudspeaker 20 and the envelope detector
50, and a second envelope detector (not shown) is used to
provide a clipping threshold for the second AC-center clip
55
per based on the output of the ?rst AC-center clipper 40.
Because the clipping threshold of the second AC-center
clipper is reduced in proportion to a level of noise in the
near-end signal in a manner analogous to that described
above with respect to the ?rst AC-center clipper 40, the
60
As described in US. patent application Ser. No. 08/578,
While the con?guration of FIG. 6 may be extremely
useful in certain contexts, it may also be advantageous to
944, entitled “Gauging Convergence of Adaptive Filters”
and ?led Dec. 27, 1995, the LMS logic block 730 may be
used to produce an output C indicating a relative level of
convergence of the echo canceler 700. Though the echo
canceler 700 of FIG. 7 is shown to be a traditional LMS-type
echo canceler, other more sophisticated echo cancelers, as
well as other devices for measuring the convergence of those
echo cancelers, are contemplated by the present invention.
See, for example, the above-mentioned U.S. patent applica
tion Ser. No. 08/578,944, which is incorporated herein by
far-end signal produced at the loudspeaker 20 is not unduly
distorted, and the dual AC-center clipper circuits provide
effective two-way (acoustic and network) echo suppression.
function H changes abruptly, there will exist a ?nite period
of time during which the transfer function of the ?lter 720
is a relatively poor approximation of the true transfer
65
reference.
In the system of FIG. 7, the convergence output C of the
echo canceler 700 is used to adjust the clipping threshold A
US 6,301,357 B1
13
14
of the AC-enter clipper 40 so that the AC-center clipper 40
are used to produce the clipping threshold A, as Well as other
serves as a residual echo suppressor and provides echo
techniques for producing the parameters C, F, HPEAK, P
suppression above and beyond that provided by the echo
themselves, are possible and contemplated herein.
Accordingly, those skilled in the art Will appreciate that the
present invention is not limited to the speci?c exemplary
canceler 700, as necessary. As described in more detail
below, the effect of the AC-enter clipper 40 is maximized
When the echo canceler 700 is unconverged, and is then
embodiments Which have been described herein for pur
poses of illustration. The scope of the invention is de?ned by
the claims Which are appended hereto, rather than the
reduced as the echo canceler 700 converges so that any
distortion introduced by the AC-center clipper 40 is mini
miZed. Recall, hoWever, that since the echo canceler 700
may never fully converge due to non-linearities in the
foregoing description, and all equivalents Which are consis
10
loudspeaker 20 and other signal processing components, it
may be desirable to keep the AC-center clipper 40 active, at
tent With the meaning of the claims are intended to be
embraced therein.
What is claimed is:
1. A signal processing device, comprising:
least to some degree, even When the echo canceler 700 is
largely converged. Furthermore, as is described With respect
to FIG. 6, the AC-center clipper 40 need not be entirely
an input node for receiving an input signal; and
a signal processor connected to said input node for
15
processing the input signal to produce an output signal,
Wherein the output signal is produced by center clip
ping the input signal using a clipping WindoW having a
deactivated at any time, even during double-talk situations.
Once again, this represents a signi?cant advantage over
conventional ?xed-center center clippers.
As shoWn in FIG. 7, the convergence output C is input to
the MAX logic block 760, as is a ?xed parameter F. The
2. The device of claim 1, Wherein the variable center of
the clipping WindoW varies in accordance With a value of the
convergence output C may be normaliZed to a range of 0 to
input signal.
1, Where 1 represents a completely unconverged state and 0
represents a fully converged state. The MAX logic block 760
then provides an output corresponding to either the conver
includes a ?xed clipping threshold.
gence output C of the echo canceler 700 or the ?oor
variable center.
3. The device of claim 1, Wherein the clipping WindoW
4. The device of claim 1, Wherein the clipping WindoW
25
includes a variable clipping threshold.
parameter F, Whichever is greater. Thus, as is described
5. The device of claim 4, Wherein the variable clipping
further beloW, the parameter F serves as a “?oor” for the
threshold varies in accordance With a value of a component
clipping threshold of the AC-center clipper 40 and prevents
the AC-center clipper 40 from being completely deactivated
of the input signal Which is to be suppressed in the output
signal.
6. The device of claim 1, Wherein the clipping WindoW
even When the echo canceler 700 is largely converged. The
parameter F is set in practice to yield a desired minimum
includes an upper clipping threshold and a loWer clipping
threshold, Wherein the upper and loWer clipping thresholds
AC-center clipper effect. Note that although the parameter F
are set independently of one another.
is shoWn as a ?xed constant in FIG. 7, it may be adjusted
7. The device of claim 6, Wherein the upper and loWer
dynamically based upon prevailing system conditions.
As shoWn, the output of the MAX logic block 760 of FIG.
7 is multiplied at the second multiplier 740 by the output of
the ?rst multiplier 750. The ?rst multiplier 750, the envelope
35
8. The device of claim 1, Wherein said signal processor is
implemented in real time using a digital signal processing
integrated circuit.
above With respect to FIG. 6. Therefore, the output A of the
second multiplier 740 represents a combination of the state
of convergence of the echo canceler 700 and the level of
input being received from the far-end user. Because the
output A of the multiplier 740 is used as the clipping
9. The device of claim 1, Wherein said signal processor is
implemented in softWare using a computer.
10. The device of claim 1, Wherein said signal processor
is implemented in real time using an analog circuit.
11. A method of processing an information signal, com
45
ate level.
With a value of the information signal.
12. The method of claim 11, Wherein the information
signal is a speech signal in a bidirectional communications
system.
13. The method of claim 11, Wherein the information
signal is a speech signal input at a mobile station in a cellular
55
tion signal.
760 equals ?oor parameter F, and the effect of the AC-center
clipper 40 is at a minimum. Assuming, hoWever, that the
?oor parameter F is non-Zero, the AC-center clipper 40 Will
continue to operate as a residual echo and noise suppressor
Note that FIG. 7 depicts but one useful con?guration.
Other combinations of the parameters C, F, HPEAK, P, which
radio system.
14. The method of claim 11, comprising the additional
step of adjusting a clipping threshold of the clipping WindoW
in accordance With a value of a component of the informa
output C approaches 0), the output of the MAX logic block
as desired.
prising the steps of:
center clipping the information signal using a clipping
WindoW having an adjustable center; and
adjusting the center of the clipping WindoW in accordance
For example, if the echo canceler 700 is completely
unconverged (i.e., if the convergence output C is 1), then the
output of the MAX logic block 760 is 1, the clipping
threshold A is equal to the output of the ?rst multiplier 750,
and the effect of the AC-center clipper is maximiZed. As is
described With respect to FIG. 6, hoWever, the output of the
?rst multiplier 750 is non-Zero only When the far-end signal
is non-Zero. When the echo canceler 700 converges (i.e.,
When the convergence output C drops from 1 toWard 0), the
output of the MAX logic block 760 tracks the convergence
output C and drops beloW 1 so that the effect of the
AC-center clipper 40 is diminished. When the echo canceler
700 becomes highly converged (i.e., as the convergence
component of the input signal Which is to be suppressed in
the output signal.
detector 50, and the parameter HPEAK operate as is described
threshold for the AC-center clipper 40, the effect of the
AC-center clipper 40 is continually adjusted to an appropri
clipping thresholds vary in accordance With at least one
65
15. The method of claim 14, Wherein the information
signal is a speech signal in a bidirectional communications
system, and Wherein the component of the information
signal used to adjust the clipping threshold is an echo signal.
16. A method of processing an input signal to produce an
output signal in Which a signal component present in the
input signal is substantially suppressed, comprising the steps
of:
US 6,301,357 B1
15
16
receiving the input signal;
Wherein the output of the envelope detector is used to adjust
the clipping threshold of said center clipper.
center clipping the input signal to produce the output
27. The device of claim 26, Wherein said envelope detec
signal using a clipping WindoW having a variable center
tor is an exponential-decay peak detector.
and a variable clipping threshold;
5
28. The device of claim 27, Wherein said exponential
adjusting the center of the clipping WindoW in time and in
decay peak detector is realiZed as an in?nite impulse
accordance With a value of the input signal;
response digital ?lter having a pole of about 255/256.
29. The device of claim 26, further comprising a multi
adjusting the clipping threshold in time and in accordance
plier disposed betWeen said envelope detector and said
With a value of the signal component present in the
center clipper, Wherein the output of the envelope detector is
10
input signal; and
outputting the output signal.
multiplied by an estimator of an effective transfer function
of a path from the source of the secondary component of the
input signal to said input node, and Wherein an output of said
rnultiplier is used to adjust the clipping threshold of said
center clipper.
17. The method of claim 16, Wherein said step of adjust
ing the clipping threshold includes the steps of:
feeding a source of the signal component present in the
input signal to an input of a peak detector; and
using an output of the peak detector as a basis for
15
31. The device of claim 22, Wherein said input node is
adjusting the clipping threshold.
connected to an output of a microphone of a mobile station
in a communications system, Wherein said output node is
connected to a transceiver of the mobile station, Wherein the
18. The method of claim 16, Wherein the input signal is a
speech signal in a bi-directional communications system,
primary component of the input signal is a voice signal
and Wherein the signal component present in the input signal
generated by a near-end user of the mobile station speaking
is an echo signal.
19. A mobile station, comprising:
an input for receiving at least an information signal; and
a center clipper for processing the information signal,
Wherein a clipping WindoW of the center clipper
25
into the microphone, and Wherein the secondary component
of the input signal is an echo signal generated by a loud
speaker of the mobile station broadcasting a voice signal
generated by a far-end user of the mobile station.
32. A method for processing an input signal, the input
signal including a time-varying prirnary component and a
time-varying secondary component, to produce an output
signal in Which the time-varying secondary component is
includes an adjustable center.
20. The mobile station of claim 19, Wherein the adjustable
center of the clipping WindoW is varied in accordance With
substantially suppressed, comprising the steps of:
receiving the input signal;
center clipping the input signal to produce the output
a value of the information signal being processed.
21. The mobile station of claim 19, Wherein the clipping
WindoW includes an adjustable clipping threshold, and
Wherein the clipping threshold is varied in accordance With
a value of a disruptive component of the information signal
30. The device of claim 29, Wherein the estimator is set
equal to a constant, and Wherein the constant is about 2.
signal using a clipping WindoW having a variable center
and a clipping threshold set to suppress the secondary
35
being processed in order to substantially suppress the dis
component of the input signal;
ruptive component.
adjusting the center of the clipping WindoW in time based
upon a value of the input signal; and
22. An echo and noise suppressing device for processing
an input signal, the input signal including a time-varying
prirnary component and a time-varying secondary
33. The method of claim 32, further comprising the steps
varying secondary component is substantially suppressed,
of:
comprising:
sensing a value of a source of the secondary component
of the input signal; and
an input node for receiving the input signal;
a center clipper connected to said input node for process
45
ing the input signal to produce the output signal,
34. A mobile station in a communications system, corn
prising:
having a variable center and a clipping threshold set to
a microphone for receiving near-end audio input at the
mobile station and for producing a near-end audio
attenuate the secondary component of the input signal;
and
an output node connected to said center clipper for
signal Which is to be transmitted to a far-end user in the
communications system;
outputting the output signal.
a loudspeaker for broadcasting, to a near-end user of the
55
of the input signal.
24. The device of claim 22, Wherein the clipping WindoW
mobile station, a far-end audio signal Which is gener
ated by the far-end user and Which is received at the
mobile station; and
includes a ?xed clipping threshold.
25. The device of claim 22, Wherein the clipping threshold
of said center clipper is variable and adjusted in accordance
adjusting the clipping threshold of the clipping WindoW in
time based upon a result of said sensing step.
Wherein said center clipper includes a clipping WindoW
23. The device of claim 22, Wherein the variable center of
the clipping WindoW is adjusted in accordance With a value
outputting the output signal.
40
component, to produce an output signal in Which the time
an echo suppression circuit for attenuating an echo corn
60
ponent of the near-end audio signal Which results from
With a value of a source of the secondary component of the
the microphone receiving output from the loudspeaker,
input signal.
Wherein said echo suppression circuit includes a center
26. The device of claim 22, further comprising an enve
lope detector disposed betWeen a source of the secondary
component of the input signal and said center clipper,
Wherein an output of the envelope detector is proportional to
a value of the source of the secondary component, and
clipper having a clipping WindoW With an adjustable
clipping center.
35. The mobile station of claim 34, Wherein the adjustable
clipping center is adjusted based upon a value of the
near-end audio signal.
US 6,301,357 B1
17
18
36. The mobile station of claim 34, wherein a variable
39. The mobile station of claim 21, Wherein the clipping
clipping threshold of the center clipper is adjusted based
threshold is reduced by a value indicating a level of noise in
a source of the disruptive component of the information
upon a value of the far-end audio signal.
37. The mobile station of claim 36, Wherein the echo
signal.
suppression circuit includes a peak detector receiving the
far-end audio signal and producing an output Which is
is reduced by a value indicating a level of noise in the source
40. The device of claim 25, Wherein the clipping threshold
of the secondary component of the input signal.
proportional to a value of the far-end audio signal, and
Wherein the clipping threshold of the center clipper is
adjusted based upon the output of the peak detector.
38. The method of claim 17, Wherein the step of adjusting
the clipping threshold includes the step of using, as an
additional basis for adjusting the clipping threshold, a value
indicating a level of noise in the source of the signal
component present in the input signal.
41. The device of claim 26, Wherein the output of the
envelope detector is used to produce a noise value indicating
10
a level of noise in the source of the secondary component of
the input signal, and Wherein the clipping threshold is
reduced by an amount proportional to said noise value.
*
*
*
*
*
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement