CN1251959A - Method and device for determining adjacent speech - Google Patents
Method and device for determining adjacent speech Download PDFInfo
- Publication number
- CN1251959A CN1251959A CN 99111988 CN99111988A CN1251959A CN 1251959 A CN1251959 A CN 1251959A CN 99111988 CN99111988 CN 99111988 CN 99111988 A CN99111988 A CN 99111988A CN 1251959 A CN1251959 A CN 1251959A
- Authority
- CN
- China
- Prior art keywords
- coefficient
- bilateral
- variance
- signal
- single session
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M9/00—Arrangements for interconnection not involving centralised switching
- H04M9/08—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic
- H04M9/082—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic using echo cancellers
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
- Filters That Use Time-Delay Elements (AREA)
Abstract
A double-talk detection device detects the presence of near-end speech signals for an echo canceller including an adaptive filter to generate an echo estimate. A control device generates the coefficients for the adaptive filter. The control device includes a coefficient updating circuit. A buffer holds the previous filter coefficients and a buffer holds the current filter coefficients. The double-talk detection device detects the presence of near-end speech signals based upon a detected variation of the coefficients. The detection device is used to detect a double-talk condition during a single-talk condition and a single-talk condition during the double-talk condition.
Description
The present invention is usually about communication system, and detects about the bilateral words of an adaptability echo cancelling system being used for communication system in particular.
Demand for the echo elimination has appearred in many duplex communication systems.Existence is to a duplexing portable operation that challenging especially environment is cellular radio device and video conferencing device of the demand of reliable echo elimination.In the portable operating period of these devices, be delayed before being fed the reply tube and arriving loud speaker once more by voice path from the signal of loud speaker, thereby they can be felt by remote subscriber as echo signal.These echo signals are commonly referred to acoustic echo, are very irritating and are difficult to remove for the participant who comprises two-way communication.
General is the Echo Canceller that utilization has a sef-adapting filter with one of effective method that removes echo signal.Lowest mean square (LMS) sef-adapting filter is by the filter of the general type of usefulness.A LMS filter is a finite impulse response (FIR) (FIR) filter, and it simulates an echo path by the coefficient of self adaptation adjustment.The coefficient of filter is used the remote signaling of a portable communications device of driving and trains adaptively from the near end signal that the microphone of portable communications device is exported.In a portable device, sef-adapting filter comprehensively comes out adaptively from the duplicating of the echo of remote signaling, and is duplicated from deduct this from the signal of the reception of the microphone output of near-end.As a result, the signal that further is sent to far-end does not have echo basically.
For two-way communication device, 4 operating states are arranged, i.e. situation, these states here will be described with reference to the portable communications device: the near-end single session state that takes place when having only near-end sound to occur; The far-end single session state that when having only far-end sound to occur, takes place; The non-talking state of bilateral speech phase that when near-end and far-end sound all occur, takes place and generation when near-end and far-end sound do not occur.Everybody understands, and during far-end single session state, the self adaptation of filter coefficient should realize preferably.
When a bilateral speech phase occurred, self adaptation must stop.If self adaptation does not stop, the coefficient of filter will be disagreed with their optimal value.On the other hand, during the single session condition, a high false detection rate can not be carried out the coefficient self adaptation, the convergence rate of the adaptive filter coefficient that so seriously slowed down continually.This resembles in portable radiotelephone and the video conferencing environment in the application that comprises a fast-changing echo path and high-noise environment, is the problem of an especially severe.Therefore, reliable bilateral words detector is a pith for the echo elimination of success.
Some bilateral words detectors measure based on signal energy or the correlation of measured signal realizes.Detect bilateral words condition based on the method for signal energy according to the different relation that transmits and receives the signal energy in the path during bilateral speech phase and single session state.For example, in the simplest realization is the energy of distant end signal and the energy of near end signal.Generally, the energy of near end signal is being to increase the bilateral speech phase from a single session transition between states.During being a bilateral speech phase from a single session transition between states, echo residual signal energy also increases.In energy by research remote signaling and near end signal or the echo residual signal one can detect the bilateral speech phase.
The echo return loss is than being used to detect the bilateral speech phase.Echo return loss ratio is generally defined as the ratio of echo residual signal energy and near-end echo signal energy.During being a bilateral speech phase from a single session transition between states, the echo return loss is than also increasing.Other method merges various signal energies and is used for the detection of bilateral words.The problem that occurs based on the method for energy comprises that they are sensitive inadequately for the low level near end sound signal, if and near-end is the environment of making an uproar, they will produce the error detection of a very high number, and (near-end is the portable environments that device is positioned at, far-end is by a communication channel, for example a participant's linking to each other of telephone wire position).
Carrying out the bilateral speech phase based on relevant method according to the signal correction of far-end and near end signal detects.The relevant of far-end and near end signal compared at bilateral speech phase height at the single session state usually.By the associated level of remote signaling and near end signal and a thresholding are compared, state can be marked as bilateral speech phase or single session state.The relevant of remote signaling and echo residual signal used in another kind of realization.In such system, when echo residual signal and remote signaling are not relevant well, be recognized as the bilateral speech phase.Under any situation, the bilateral speech phase is based on when a transition from the single session state to the bilateral speech phase, in the relevant supposition that can reduce that transmits and receives the signal in the path of two-way communication device.
The also known relevant bilateral speech phase that detects by monitoring estimated echo signal and echo residual signal.Normally, this relevant can increasing during transition from the bilateral speech phase to the single session state.Yet because relevant to noise-sensitive, these technology can not be used to the environment of making an uproar.
Therefore, exist a kind ofly to can be, comprise that a kind of more healthy and stronger and more accurate bilateral of working in the environment that is subjected to significantly with the background noise of great dynamic range and the influence of fast-changing echo path talks about the demand of detector at all environment.
Fig. 1 is the circuit diagram of the duplex communication device of the self-adaptive echo eliminator that uses of signal utilization the present invention.
Fig. 2 is the circuit diagram that signal is used for the control device of self-adaptive echo eliminator.
Fig. 3 is the circuit diagram of signal bilateral words detector.
Fig. 4 is the flow chart that is shown in the work of bilateral words detector under the single session condition.
Fig. 5 is the flow chart that is shown in the work of single session detector under the bilateral words condition.
Bilateral words detector is to be associated with the echo of the sef-adapting filter with the band adjustability coefficients device that disappears.Echo Canceller comprises a sef-adapting filter that produces an estimated echo, the subtracter of the cancelling out echo signal that generation is used to transmit and the coefficient of a renewal sef-adapting filter and carry out the control device that the bilateral words detect.Control device comprises a buffer that is used to store the current coefficient of sef-adapting filter, the buffer of the coefficient of the front of a storage sef-adapting filter, the filter coefficient of a utilization front produces the filter of an estimated echo, a generation is used for upgrading the estimated echo error signal of filter coefficient, and filter coefficient update device and selective updating are used for the bilateral words detection means of the filter coefficient that echo eliminates.
Bilateral words detection means comprises a bilateral words detector and single session detector.Bilateral is talked about the detector monitors index variation, and for example detects the difference between the coefficient of current coefficient and front.If too greatly above a predetermined cycle, detecting, difference is the bilateral speech phase.Can use a self adaptation variance thresholding.The single session detector detects difference between the coefficient of when current coefficient and front and falls a variance thresholding when following.If difference is fallen a variance thresholding with the predetermined cycle of the next one, detect and be the single session state.Can use an adaptive threshold, and can produce in an identical manner for the thresholding of single session and the detection of bilateral words.
Fig. 1 represents to use the duplex communication device 101 of the Echo Canceller 100 that the present invention uses.Duplex communication device 101 is illustrated as a portable device, can be a portable radiotelephone, portable video conferencing device, portable satellite phone, portable cordless telephone, people's computer (PC) multimedia communication device or any other suitable communication device one by one.In addition, those one skilled in the art can recognize that here the bilateral words detection means of announcing can be used in other environment with echo signal, resemble repeater, and 2 lines are to 4 line transducers etc.
The transmit path of duplex communication device 101 comprises the microphone 104 that picks up this locality (near-end) voice signal, one is adjacent speech signal from analog format conversion that 106, one of analog to digital (A/D) transducers of number format deduct the subtracter 112 of an estimated echo and one free echo is sent signal is converted to analog format from number format digital-to-analog (D/A) transducer 126 from digitlization adjacent speech signal.
The RX path of duplex communication device 101 comprises that 124, one of an A/D converter that the far-end voice signal are converted to number format from analog format is converted to the far-end voice signal D/A converter 108 of analog format and long-range (far-end) voice signal is sent to local hearer's loud speaker 102 from number format.
Those one skilled in the art will recognize that A/D converter 124 and D/A converter 126 are used to analogue system.Additionally, in some applications, A/D converter 124 can replace with a vocoder, and D/A converter 126 can replace with a voice decoder.For example, transducer 124 and 126 can be the voice decoder and the vocoder of a digital interface that is used for the communication device of global system for mobile communications (GSM) or Integrated Service Digital Network.The merging that is further noted that antenna 122 and transceiver 120 can be replaced by a network interface in some applications, resembles the application that is used for wired or optical communication system.
Far-end voice signal x (n) at the number format at output 109 places of A/D converter 124 is imported into a D/A converter 108, thereby produces the analog signal of a driving loud speaker 102.Detect by microphone 104 with the adjacent speech signal and in A/D converter 106, be converted to number format by the part of the remote signaling of loud speaker 102 output.The y of near end signal as a result (n) at output 103 is imported into subtracter 112.The estimated echo signal z (n) of number format is exported from sef-adapting filter 110 at output 107, and is input to subtracter 112.The echo of number format eliminates voice signal e (n) or r (n) exports and be input to D/A converter 126 at the output 105 of subtracter 112.
Self-adaptive echo eliminator 100 can be at digital signal processor (DSP), realizes in the microprocessor, a programmable logic device etc.Echo Canceller 100 comprises that one is illustrated as one and has adjustable integral coefficient W
B(n) the finite impulse response (FIR) that is used for producing estimated echo z (n) (FIR) self adaptation of filter filter 110.At current sampling time n, a far-end sound samples x (n) receives from output 109 as the output of A/D converter 124, and near-end sound sampling y (n) receiving from output 103 as the output of A/D converter 106.Signal x (n) and y (n) are synchronous, and reason is that A/D converter 106 uses identical clock with D/A converter 108.Estimated echo z (n) can calculate based on following formula:
Wherein subscript T representative vector or transpose of a matrix, L is the exponent number (i.e. the number of the coefficient positions in sef-adapting filter 110) of FIR sef-adapting filter 110, X (n)=[x (n) x (n-1) ... x (n-L+1)]
TKeep L the sampling of far-end sound recently, W
B(n)=[W
B 0(n) W
B 1(n) ... W
B L-1(n)] T is used for the filter coefficient that echo is eliminated.W
B(n)
TX (n) is defined as 2 following vector W
B(n) and the dot product of X (n):
Self-adaptive echo eliminator 100 further is included in output 105 and produces the subtracter 112 that is used for the further echo cancellation signal e (n) that sends, and by deduct the estimated echo z (n) at output 107 places from the near end signal y (n) of output 103, gets
e(n)=y(n)-z(n) (2)
Self-adaptive echo eliminator 100 further comprise based on from the far-end voice signal x (n) of output 109 and from the near end sound signal y (n) of output 103 at the output 407 output factor W of place
B(n) and carry out the control device 300 that bilateral words detect.
With reference to figure 2, control device 300 is linked output 109 and is received remote signaling x (n), links output 103 and receives near end signal y (n) and comprise updated filter coefficient W is provided
B(n) output 407.Control device 300 comprises a coefficient update circuit 320.The current coefficient of first buffer 318 storages, the current coefficient that during the sampling interval recently, produces.Second buffer 314 keeps the coefficient of fronts, promptly just at the coefficient in sampling interval of nearest sampling interval front.
The coefficient W of the sef-adapting filter 110 (Fig. 1) of bilateral words detection means 400 output response buffer 314 and 318 output
B(n).Sef-adapting filter 310 is second sef-adapting filter and produces an estimated echo of delivering to subtracter 312.The error signal that subtracter 312 outputs are used by coefficient update circuit 320.
W
C(n)=W
P(n)+μe
1(n)X(n)[X(n)
TX(n)]
-1 (3)
Wherein μ is a step-length, W
C(n)=[W
C 0(n) W
C 1(n) ... W
C L-1(n)]
TIt is current coefficient.During the single session condition, coefficient W
B(n) be continuously updated coefficient W
C(n) be used for being input to sef-adapting filter 110 by output 407.
In order to adapt to, before just in coefficient update circuit 320, upgrading, current coefficient W
C(n) be used as front coefficient W
P(n) exist in the buffer 314.After adapting to, the coefficient after the new renewal is used as current coefficient W
C(n) exist in the buffer 318. Buffer 314 and 318 can use any suitable memory devices to realize, resembles the parallel loads shift register, incoming memory (RAM) at random, the erasable read-only memory (EEPROM) etc. that removes.
The filter coefficient W of control device 300 utilization fronts
P(n) produce an estimated echo z
1(n).Estimated echo is the coefficient W by the sampling period of sef-adapting filter 310 utilization fronts
P(n) and current remote signaling x (n) produce following generation estimated echo z
1(n):
W wherein
P(n)=[W
P 0(n) W
P 1(n) ... W
P L-1(n)]
TIt is the filter coefficient of front.
e
1(n)=y(n)-z
1(n) (5)
Bilateral words detection means 400 (Fig. 3) receive the filter coefficient W from the front of output 403
P(n) with from the current filter coefficient W of output 401
C(n) and at the filter coefficient of output 413 or output 415 output adaptive filters 110.During the bilateral speech phase, the coefficient in buffer 411, and the coefficient of therefore delivering to sef-adapting filter 110 can not change.
Bilateral words detection means 400 comprises the switch of selecting according to the state of bilateral words sign DF (shown in Fig. 4 and Fig. 5) 402 between output 501 and output 503.DF is the bilateral words sign of the current state of an indication bilateral words detection means 400, wherein, DF is 1 when being in the bilateral speech phase, is in the single session state, and DF is 0 (the single session state that is used for the purpose of this sign is non-existent all moment of bilateral speech phase).If the bilateral speech phase is not detected, the filter coefficient W of front then
P(n) and current filter coefficient W
C(n) at output 501.If the bilateral speech phase is detected, the filter coefficient W of the selected and output front of output 503
P(n) and current filter coefficient W
C(n).
Bilateral words detection means 400 (Fig. 3) comprise the filter coefficient W of storage sef-adapting filter 110 (Fig. 1)
B(n) buffer 411.Coefficient W in buffer 411
B(n), and,, all give sef-adapting filter 110 (Fig. 1) continuously no matter single session still is the bilateral speech phase through output 407 outputs by bilateral words detector 600 and single session detector 800 selective updatings.
Bilateral words detection means 400 (Fig. 3) further comprise the coefficient W that carries out in detection of bilateral words and the renewal buffer 411
B(n) bilateral words detector 600 and fill order converse and detect and upgrade coefficient W in the buffer 411
B(n) single session detector 800.As described here, when detecting to the bilateral speech phase, bilateral words detector is talked about sign and bilateral words sign DF is set in bilateral of output 409 outputs is 1.When detecting to the single session state, it is 0 that single session detector 800 output bilateral words sign DF give switch 402 and the bilateral of resetting words sign DF.Bilateral words detector 600 upgrades the coefficient W of the buffer 411 that is used for sef-adapting filter 110 (Fig. 1) by the current coefficient of output 413 outputs in the single session state
B(n).In the bilateral speech phase, in a single day single session detector 800 detects the single session state by output 415 just passes through current coefficient W
C(n) duplicate and upgrade the adaptive filter coefficient W that is used for sef-adapting filter 110 (Fig. 1) at buffer 411
B(n).Coefficient W
B(n) always identical coefficient in the bilateral speech phase, and bilateral words detector frequent coefficient of upgrading of output in the single session state.
Below with reference to Fig. 4, the work of bilateral words detector 600 will be described now.Detect in order to carry out the bilateral words under the single session state, bilateral words detector 600 receives the filter coefficient W of front from output 501 (Fig. 3)
P(n) and current filter coefficient W
C(n).The change rate of bilateral words detector 600 variance by following calculating filter coefficient in step 610:
D(n)=||W
C(n)-W
P(n)||
2 (6)
All can followingly represent wherein:
L is the number of the coefficient among sef-adapting filter 110 (Fig. 1) and 310 (Fig. 2), so as to be created in the coefficient in each filter location difference and.Although filter has the tap of similar number, i.e. coefficient, those one skilled in the art will appreciate that they can have the tap of different numbers, under the sort of situation, L is the number of the tap in the sef-adapting filter 310.Those one skilled in the art also will appreciate that by other method also can determine variance.The present invention can serious vibration come reliable Detection bilateral speech phase by affirmation coefficient during the self adaptation of bilateral speech phase, and the variance of the coefficient of sef-adapting filter is no matter near-end noise can become invariant after restraining in the single session state.Like this, even talking about detector, bilateral of the present invention in the environment of making an uproar, also can effectively work.
Whether determine D (n) greater than variance thresholding ATH (n) in step 612, wherein ATH (n) is an adaptive threshold.Step of updating for example can be a low pass filter in the single session state by using the following formula of asking ATH (n):
ATH(n)=(1-α)ATH(n)+αD(n) (8)
Wherein α is the constant between 0 and 1, for example can be a value 0.01.Additionally, the mean value of D (n) can be used as adaptive threshold.Yet, wish the thresholding of low-pass filtering.Because no matter the appearance of near-end noise, the variance of coefficient almost is a constant at the single session state after convergence, use adaptive threshold.Such adaptive threshold will be from an initial value convergence, and reason is that sef-adapting filter is all restrained regardless of background noise.In case convergence occurs, thresholding can almost be a constant.In addition, if background noise has a higher average level, adaptive threshold will allow a higher bilateral words thresholding.This allows bilateral words detector reliably working in various noise conditions.
Again with reference to figure 4, if all square D (n) is bigger than adaptive threshold ATH (n), bilateral words detector 600 adds 1 to bilateral words counter DT_CNT and indicates all square D (n) to surpass the number in the moment of adaptive threshold successively in step 616 so.If as definite bilateral words counter DT_CNT surpasses a bilateral words thresholding DTH in step 618, detecting so is the bilateral speech phase.In step 620, bilateral words sign DF is set as high logic level, and counter ST_CNT is set as 0.In step 624, bilateral words sign DF is output to output 409, filter coefficient W
B(n) be output to output 413.
As determining in step 618,, detect and be non-bilateral speech phase if bilateral words counter DT_CNT is big unlike bilateral words threshold level DTH.In the case, in step 630, bilateral words sign DF is set as 0.Can envision bilateral words thresholding can be very short, 50 sampling intervals for example, and promptly 0.01 second, so that the bilateral speech phase is by very fast detection.In step 624, bilateral words sign DF is output to output 409, filter coefficient W
B(n) output to output 413.
Get back to Fig. 4 once more, determine that in step 612 all square D (n) is unlike under the big situation of ATH (n), bilateral words detector 600 is 0 to upgrade bilateral words counter DT_CNT by counter is set, and upgrades self adaptation variance thresholding ATH (n) and by duplicating W according to formula (8)
C(n) or W
P(n) upgrade filter coefficient W
B(n), as indicated in step 628.Bilateral words detector 600 further is provided with bilateral words sign DF=0 and from output 409 output bilaterals words sign DF with from output 413 output factor W in step 630
B(n).
In a word, bilateral words counter DT_CNT is used for the integer counter that the bilateral words detect, and DTH is a bilateral words threshold level that is used for bilateral words counter DT_CNT.That describe below is a single session counter ST_CNT, and this is one and is used for the integer counter that single session detects that STH is a single session threshold level that is used for single session counter ST_CNT.
Below with reference to Fig. 5, the converse work of the single session detector 800 that detects of fill order will be described during bilateral words condition.Single session detector 800 receives the filter coefficient W of front from output 503
P(n) and current filter coefficient W
C(n).Single session detector 800 is talked about the calculating W that detector 600 defines in step 610 as about bilateral
C(n) and W
PAll sides of difference (n).
With reference to figure 5, if determined that in step 712 all square D (n) is littler than adaptive threshold ATH (n), single session counter ST_CNT adds 1 in step 806 again.Determine in step 807 afterwards whether single session counter ST_CNT is bigger than single session thresholding STH.If single session counter ST_CNT is bigger than single session thresholding STH, detects and be the single session state.It is relatively very big to envision the single session thresholding, like this, for example can be 2000 sampling intervals, promptly 0.25 second.Be the single session state in case detect, bilateral words sign DF is set as 0 in step 810, and bilateral words counter DT_CNT is set as 0.In step 811, bilateral words sign DF is output to output 405, and by duplicating the current coefficient W through output 415 outputs
C(n), filter coefficient W
P(n) be updated so that these current coefficients can be existed in the buffer 411.
If determine that in step 807 single session counter ST_CNT is big unlike single session thresholding STH, detect to be non-single session state.In this case, bilateral words sign DF is set as 1 in step 812.Bilateral words sign DF is output to output 405, filter coefficient W
B(n) do not output to output 415.
Refer again to Fig. 5, if bigger than ATH (n) as determined all square D (n) in step 712, single session counter ST_CNT is made as in step 804 and equals 0 so, and bilateral words sign DF is set as 1 in step 812.Bilateral words sign DF produces filter coefficient W at output 405
B(n) without output 415 outputs.
Visible so a kind of new bilateral words detector uses the measurement of variance of the coefficient of sef-adapting filter to do the judgement of bilateral words.In addition, variable threshold of utilization adapts to the various noise conditions at near-end.Therefore, bilateral words detector is very insensitive to near-end noise, and very healthy and strong in the environment of making an uproar.Because bilateral words detection means is directly measured variance from sef-adapting filter, it is accurately with reliably.
Claims (11)
1, a kind of method that in the Echo Canceller that is used for offsetting in the echo of the remote signaling that near end signal occurs, detects adjacent speech, the method comprising the steps of:
Upgrade the filter coefficient that connects in order to the sef-adapting filter (310) that receives remote signaling and error signal;
The variance of calculating filter coefficient (610); With
If the variance of filter coefficient surpasses a variance thresholding during the single session state, detect and be bilateral speech phase (618).
2, the method for claim 1 further is included in during the bilateral speech phase, falls the variance thresholding when following when the variance of filter coefficient, detects the step into single session state (618).
3, the method for claim 2 is wherein fallen the variance thresholding with next preset time during the cycle (616,618) when variance, detects to be the single session state.
4, as the method for definition in claim 1 or 2, further comprise the step that produces the variance thresholding adaptively.
5, as the method for definition in claim 2 or 4, the step that wherein detects the single session state comprises the step of the quadratic sum signal that produces the filter coefficient difference.
6, as the method for definition in the claim 5, wherein the variance thresholding is by producing averaging with signal.
7, as the method for definition in the claim 5, wherein the variance thresholding is by producing carrying out low-pass filtering with signal.
8, as the method for definition in the claim 1, wherein when the variance rate of the change of filter coefficient above preset time of variance thresholding during the cycle, detect and be bilateral speech phase (806,807).
9, as the method for definition in the claim 8, the step that wherein detects the bilateral speech phase comprises square summation to the difference between the coefficient of one of an a coefficient positions nearest coefficient and a front.
10, as the method for definition in the claim 9, wherein summation comprises:
Calculate the difference of each coefficient positions of sef-adapting filter;
Each difference is carried out square; With
Square summation to the difference of all coefficient positions.
11, a kind of control device (300) is carried out according to any one method among the claim 1-10.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US12907698A | 1998-08-04 | 1998-08-04 | |
US09/129,076 | 1998-08-04 |
Publications (1)
Publication Number | Publication Date |
---|---|
CN1251959A true CN1251959A (en) | 2000-05-03 |
Family
ID=22438352
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 99111988 Pending CN1251959A (en) | 1998-08-04 | 1999-08-04 | Method and device for determining adjacent speech |
Country Status (3)
Country | Link |
---|---|
CN (1) | CN1251959A (en) |
DE (1) | DE19935587A1 (en) |
GB (1) | GB2342832B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100466665C (en) * | 2005-09-28 | 2009-03-04 | 华为技术有限公司 | Method and device for detecting circuit one-side connecting |
CN100530998C (en) * | 2002-10-30 | 2009-08-19 | 冲电气工业株式会社 | Echo Eliminator with echo path change detector |
CN1926841B (en) * | 2004-03-05 | 2011-05-18 | 汤姆森许可公司 | Acoustic echo canceller with multimedia training signal |
CN1668058B (en) * | 2005-02-21 | 2011-06-15 | 南望信息产业集团有限公司 | Recursive least square difference based subband echo canceller |
CN101719969B (en) * | 2009-11-26 | 2013-10-02 | 美商威睿电通公司 | Method and system for judging double-end conversation and method and system for eliminating echo |
CN106033673A (en) * | 2015-03-09 | 2016-10-19 | 电信科学技术研究院 | Near-end speech signal detecting method and near-end speech signal detecting device |
CN108353107A (en) * | 2015-11-13 | 2018-07-31 | 伯斯有限公司 | The double talk detection eliminated for acoustic echo |
CN111654585A (en) * | 2020-03-26 | 2020-09-11 | 紫光展锐(重庆)科技有限公司 | Echo sound field state determination method and device, storage medium and terminal |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2946486B1 (en) * | 2009-06-09 | 2012-04-20 | Parrot | METHOD FOR DETECTING A DOUBLE SPEECH SITUATION FOR HANDS-FREE TELEPHONE DEVICE |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1578025A1 (en) * | 1994-05-06 | 2005-09-21 | NTT Mobile Communications Network Inc. | Double talk detecting method, double talk detecting apparatus, and echo canceler |
-
1999
- 1999-08-02 DE DE1999135587 patent/DE19935587A1/en not_active Withdrawn
- 1999-08-02 GB GB9917920A patent/GB2342832B/en not_active Expired - Fee Related
- 1999-08-04 CN CN 99111988 patent/CN1251959A/en active Pending
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100530998C (en) * | 2002-10-30 | 2009-08-19 | 冲电气工业株式会社 | Echo Eliminator with echo path change detector |
CN1926841B (en) * | 2004-03-05 | 2011-05-18 | 汤姆森许可公司 | Acoustic echo canceller with multimedia training signal |
CN1668058B (en) * | 2005-02-21 | 2011-06-15 | 南望信息产业集团有限公司 | Recursive least square difference based subband echo canceller |
CN100466665C (en) * | 2005-09-28 | 2009-03-04 | 华为技术有限公司 | Method and device for detecting circuit one-side connecting |
CN101719969B (en) * | 2009-11-26 | 2013-10-02 | 美商威睿电通公司 | Method and system for judging double-end conversation and method and system for eliminating echo |
CN106033673A (en) * | 2015-03-09 | 2016-10-19 | 电信科学技术研究院 | Near-end speech signal detecting method and near-end speech signal detecting device |
CN106033673B (en) * | 2015-03-09 | 2019-09-17 | 电信科学技术研究院 | A kind of near-end voice signals detection method and device |
CN108353107A (en) * | 2015-11-13 | 2018-07-31 | 伯斯有限公司 | The double talk detection eliminated for acoustic echo |
CN111654585A (en) * | 2020-03-26 | 2020-09-11 | 紫光展锐(重庆)科技有限公司 | Echo sound field state determination method and device, storage medium and terminal |
CN111654585B (en) * | 2020-03-26 | 2021-08-03 | 紫光展锐(重庆)科技有限公司 | Echo sound field state determination method and device, storage medium and terminal |
Also Published As
Publication number | Publication date |
---|---|
GB9917920D0 (en) | 1999-09-29 |
GB2342832B (en) | 2001-03-21 |
GB2342832A (en) | 2000-04-19 |
DE19935587A1 (en) | 2000-02-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7203308B2 (en) | Echo canceller ensuring further reduction in residual echo | |
CN101878637B (en) | A method and arrangement for echo cancellation of voice signals | |
US5390244A (en) | Method and apparatus for periodic signal detection | |
EP0366584B1 (en) | Full-duplex digital speakerphone | |
US6904146B2 (en) | Full duplex echo cancelling circuit | |
US5473686A (en) | Echo cancellation apparatus | |
CN101562669B (en) | Method of adaptive full duplex full frequency band echo cancellation | |
US6192126B1 (en) | Double talk detector, method for double talk detection and device incorporating such a detector | |
EP0597201A1 (en) | Apparatus and method for noise reduction for a full-duplex speakerphone or the like | |
US20050286714A1 (en) | Echo canceling apparatus, telephone set using the same, and echo canceling method | |
CN1842110B (en) | Echo eliminating device and method | |
EP1391101B1 (en) | Telephone apparatus | |
PL173748B1 (en) | Network echo eliminator | |
JP2000512466A (en) | Nonlinear processor for echo suppression and echo cancellation. | |
JPS61127234A (en) | Control system of echo-canceller | |
US5675644A (en) | Method and apparatus for canceling echo accounting for delay variations | |
CN1706174B (en) | Integrated noise cancellation and residual echo suppression | |
CN1653713A (en) | Echo canceller with double-talk detector | |
US5978473A (en) | Gauging convergence of adaptive filters | |
CN1533665A (en) | Echo canceller having nonlinear echo suppressor for harmonics calculations | |
US6816592B1 (en) | Echo cancellation in digital data transmission system | |
CN1251959A (en) | Method and device for determining adjacent speech | |
CN100550951C (en) | A kind of echo cancelling device and method | |
WO1991012683A1 (en) | Speaking device having hand-free conversational function | |
CN101826328A (en) | Echo offset method in embedded wireless visual doorbell |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C06 | Publication | ||
PB01 | Publication | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |