CN100550951C - A kind of echo cancelling device and method - Google Patents
A kind of echo cancelling device and method Download PDFInfo
- Publication number
- CN100550951C CN100550951C CNB2005100982114A CN200510098211A CN100550951C CN 100550951 C CN100550951 C CN 100550951C CN B2005100982114 A CNB2005100982114 A CN B2005100982114A CN 200510098211 A CN200510098211 A CN 200510098211A CN 100550951 C CN100550951 C CN 100550951C
- Authority
- CN
- China
- Prior art keywords
- signal
- echo signal
- main
- residual echo
- filter factor
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
Abstract
The invention discloses a kind of echo cancelling device, comprising: adaptive-filtering module, main filtration module and first control module.The present invention discloses a kind of echo cancel method, echogenicity signal the last period according to present segment carries out the self adaptation adjustment to the filter factor of current preservation, obtain once adjusted filter factor, according to present segment echogenicity signal this once adjusted filter factor is carried out the adjustment of secondary self adaptation again, if adjust the predetermined condition that the result satisfies the filter factor of the current preservation of renewal for this twice, then upgrade the filter factor of current preservation with adjusted filter factor once.The present invention passes through the further checking to the predetermined condition of upgrading filter factor, has guaranteed the reliability of the filter factor of preservation, thereby has guaranteed the effect that echo is eliminated.Simultaneously, filter factor must satisfy the predetermined condition that continues certain number of times, could further be preserved, and has more guaranteed the accuracy of the filter factor preserved and the reliability that echo is eliminated.
Description
Technical field
The present invention relates to the echo field of voice communication system, be specifically related to a kind of echo cancelling device and method.
Background technology
When making a phone call especially toll telephone, oneself the sound after we can hear from telephone receiver sometimes and postpone through a period of time, Here it is echo.Echo can make the people feel under the weather, when serious even conversation can't normally be carried out.
Echo is divided into two kinds according to its mechanism of production: electric echo and acoustic echo.As shown in Figure 1, wherein, R
InThe expression remote end input signal, R
OutThe remote signaling that the expression near-end receives, S
InThe expression near end input signal, S
OutThe near end signal that the expression far-end receives.With the far-end is example, and the mechanism of production of electric echo is as follows: far-end speech signal R
InAt public switch telephone network (PSTN, when transmitting Public SwitchedTelephone Network), need to use the mixing transformation device to finish of the conversion of four lines of switch to the two wires of user side, understanding some signal in transfer process leaks on the far-end RX path from the far-end transmit path, the signal of this part " leakage " transfers back to far-end again, remote subscriber has just been heard the sound of oneself like this, Here it is electric echo.The mechanism of production of acoustic echo is as follows: the far-end speech signal R that loud speaker receives
Out, through the reflection in different path or without any reflection, after being picked up by the near-end microphone, resending to far-end, remote subscriber is heard the voice of oneself as a result, Here it is acoustic echo.
Echo can be felt out to satisfy intensity and two conditions of time delay by the people: the signal that the signal that, feeds back promptly leaks or the signal of reflection should reach certain amplitude; Two, enough propagation delay times will be arranged, promptly speech range reaches certain-length.Along with development of Communication Technique, speech range is more and more longer, and the extensive use of ip voice (VOIP, Voice Over IP) in recent years more makes signal delay increase greatly, thereby echoing is also just more serious.The existence of echo has influenced communication quality, makes the communication system cisco unity malfunction when serious, and the echo that therefore must take effective measure to suppress is improved voice communication quality.
Now, use usually based on the Echo Canceller of adaptive-filtering and eliminate echo, its basic principle is as follows:
Suppose: R
In, R
Out, S
In, S
OutImplication identical with the implication of corresponding character among Fig. 1, H represents real echo path unit impulse response, G represents real near end echo signal, S represents the voice signal of near-end user, N represents the background noise of near-end, and h represents the echo path unit impulse response that sef-adapting filter estimates, and g represents the analog echo signal that sef-adapting filter estimates, e represents the residual echo signal after echo is eliminated, simultaneously as can be seen: S
In=S+G+N.
With the far-end is example, for acoustic echo, and R
OutBe the source signal that produces acoustic echo, G is real acoustic echo signal, and when near-end does not have voice and is S=0, S
InBe approximately equal to G, therefore, the basic thought of eliminating based on the acoustic echo of adaptive-filtering is: sef-adapting filter is by making R
OutApproach S
InAdaptive algorithm to obtain H, when error meets the demands, obtain the echo path h of a simulation, thereby obtain analog echo signal g, last subtracter is from S
InIn deduct g and just realized the acoustic echo elimination.The principle that electric echo is eliminated is the same with acoustic echo.
Since echo path H normally unknown with the time become, therefore, the effect key that echo is eliminated depends on sef-adapting filter by the adaptive algorithm simulation and the quality of following the tracks of echo path, promptly whether g can approach G well.For acoustic echo, when having only remote subscriber to speak, owing to S this moment
InBe approximately equal to G, so the effect of sef-adapting filter can accept generally, but when remote subscriber and near-end user are talked (double-talk) state simultaneously, this moment S
In=S+G+N is because near-end voice signals S and far-end speech signal R
OutBe incoherent, therefore, then easily cause algorithm to be dispersed if sef-adapting filter is proceeded the self adaptation adjustment this moment, and it is ineffective that acoustic echo is eliminated.Same reason can make when the double-talk state, and electric echo is eliminated ineffective.In view of above reason, in based on the Echo Canceller of adaptive-filtering, usually need to add the double-talk testing mechanism, sef-adapting filter just stops the self adaptation adjustment when being in the double-talk state when detecting.
Fig. 2 is the structured flowchart that adds the echo cancelling device of double-talk mechanism in the prior art, as shown in Figure 2, this device is to increase a senior filter in based on the Echo Canceller of adaptive-filtering, it mainly comprises: sef-adapting filter 21, first subtracter 22, senior filter 23, second subtracter 24 and control module 25, and operation principle is as follows:
At first, sef-adapting filter 21 sends the request of obtaining filter factor to senior filter 23, senior filter 23 returns the filter coefficient alpha 1 of self preserving to sef-adapting filter 21 after receiving this request, after sef-adapting filter 21 is received this coefficient, be reference signal, be error signal with current echogenicity signal x (i) with the residual echo signals of the last output of first subtracter 22, according to adaptive algorithm α 1 is adjusted, write down adjusted filter coefficient alpha
i, and with α
iOutput to control module 25, and the adaptive modeling echo signal that estimates according to filter factor and x (i) in the adjustment process outputed to first subtracter 22,22 calculating of first subtracter are mixed with the difference of the present segment input signal d (i) and the adaptive modeling echo signal that sef-adapting filter 21 is exported of echo signal, this difference is self adaptation residual echo signal, and this self adaptation residual echo signal is outputed to control module; Simultaneously, senior filter 23 is according to the filter coefficient alpha 1 of self preserving, present segment echogenicity signal x (i) is carried out filtering to be calculated, and the main analog echo signal that will obtain outputs to second subtracter 24, second subtracter 24 calculates the difference of the main analog echo signal of d (i) and senior filter 23 outputs, this difference is main residual echo signal, and should output to control module 25 by main residual echo signal; Control module 25 is calculated the self adaptation residual echo signal that receives and the average power p0 (i) and the p1 (i) of main residual echo signal correspondence respectively, and whether judges p0 (i) less than p1 (i), if think the current double-talk state that is not in, with α
iOutput to senior filter 23, senior filter 23 receives that the back is with α
iUpgrade the α 1 that self preserves; Otherwise, think not to be for further processing the current double-talk of being in state.At last, the signal after control module 25 is eliminated main residual echo signal as echo outputs to far-end.
The basic process that the echo of adding double-talk mechanism is eliminated in the prior art is as follows:
01: with current echogenicity signal x (i) is reference signal, self adaptation residual echo signal with the last time is an error signal, filter coefficient alpha 1 to current preservation is carried out the self adaptation adjustment, and in the self adaptation adjustment process, obtain the adaptive modeling echo signal, calculate the current input signal d (i) of echo and the difference of this adaptive modeling echo signal of being mixed with, obtain self adaptation residual echo signal, establish and become α after α 1 adjusts
iAccording to α 1, x (i) is carried out filtering calculate simultaneously, obtain main analog echo signal, calculate the difference of d (i) and this main analog echo signal then, obtain main residual echo signal.
02: the average power p0 (i) and the p1 (i) that calculate self adaptation residual echo signal and main residual echo signal correspondence respectively.
03: judge whether p0 (i)≤β p1 (i) sets up, and wherein β is a constant, if think the current double-talk state that is not in, with α
iUpgrade the α 1 of current preservation; Otherwise, think to keep α 1 constant by the current double-talk of being in state.
By as can be seen above, the condition of upgrading current filter coefficient alpha 1 is: system is not in the double-talk state.Accuracy to the double-talk state-detection will directly have influence on the effect that echo is eliminated.In said system, the basis that decision-making system is not in the double-talk state is: p0 (i) is less than p1 (i).When satisfying p0 (i) less than p1 (i), system might not be in the double-talk state but in fact.This be because: in practice, in order to improve the accuracy of adaptive algorithm, usually the echogenicity signal of getting a period of time is that reference signal is carried out the self adaptation adjustment, may double-talk appear at last several points of this signal, and mistuning has appearred in the filter factor of this moment, but still satisfy p0 (i) less than p1 (i), according to above-mentioned condition, think that system is not in the double-talk state, the update condition that meets filter factor, the result has preserved the filter factor of mistuning get off.If when filter factor generation mistuning, in time do not stop the adjustment of filter factor, may make adaptive-filtering coefficient adjustment after this always tanglewrack so, so the effect that echo is eliminated can't be guaranteed.
Summary of the invention
In view of this, main purpose of the present invention is to provide a kind of echo cancelling device and method, with raising double-talk status detection accuracy, thereby improves validity and the reliability that echo is eliminated.
For achieving the above object, technical scheme of the present invention is achieved in that
A kind of echo cancelling device, this device comprises: adaptive-filtering module 31, main filtration module 32 and first control module 33,
Adaptive-filtering module 31, when the present segment echogenicity signal of receiving from the outside, the filter factor that main filtration module 32 is preserved carries out a self adaptation adjustment, and will output to first control module 33 according to the preceding self adaptation residual echo signal that the last period of present segment, the echogenicity signal obtained;
Wherein, after adaptive-filtering module 31 is received and is continued the filtering triggering signal, to once adjust the back filter factor and output to main filtration module 32, and carry out the adjustment of secondary self adaptation to once adjusting the back filter factor according to present segment echogenicity signal, the existing self adaptation residual echo signal that obtains is outputed to first control module 33; Main filtration module 32 will carry out the existing main residual echo signal that filtering calculates and output to first control module 33 according to once adjusting back filter factor, present segment echogenicity signal; First control module 33 judges whether the average power of existing self adaptation residual echo signal and existing main residual echo signal satisfies predetermined relationship, if, send the update coefficients triggering signal to main filtration module 32, main filtration module 32 upgrades the filter factor of self preserving once to adjust the back filter factor, and will show main residual echo signal output; Otherwise, sending the retention factor triggering signal to main filtration module 32, main filtration module 32 will be exported according to the existing main residual echo signal that filter factor, the present segment echogenicity signal self preserved obtain.
Described adaptive-filtering module (31) comprising:
Sef-adapting filter (311) is used for obtaining filter factor to main filtration module (32) when the present segment echogenicity signal of receiving from the outside; When receiving the filter factor of main filtration module (32) output, with echogenicity signal the last period from the present segment of outside is reference signal, self adaptation residual echo signal with the last output of first subtracter (312) is an error signal, filter factor to this main filtration module (32) output carries out a self adaptation adjustment, according to the filter factor in the self adaptation adjustment process echogenicity signal the last period of present segment being carried out filtering calculates, adaptive modeling echo signal before obtaining, should before adaptive modeling echo signal and preceding adaptive echo identify and output to first subtracter (312); Be used for after receiving the continuation filtering triggering signal of first control module (33), with present segment echogenicity signal is reference signal, preceding self adaptation residual echo signal with the last output of first subtracter (312) is an error signal, adjusted filter factor is once carried out the adjustment of secondary self adaptation, according to the filter factor in this secondary self adaptation adjustment process present segment echogenicity signal being carried out filtering calculates, obtain existing adaptive modeling echo signal, should output to first subtracter (312) with existing adaptive echo sign by existing adaptive modeling echo signal;
First subtracter (312), before receiving being designated of sef-adapting filter (311) output during the analog echo signal of adaptive echo, calculating receive from the input signal that is mixed with echo the last period of the present segment of outside and should before the difference of adaptive modeling echo signal obtain before self adaptation residual echo signal, should before self adaptation residual echo signal output to sef-adapting filter (311), and should before self adaptation residual echo signal and preceding self adaptation residual echo identify and output to first control module (33); When receive sef-adapting filter (311) output be designated the analog echo signal of existing adaptive echo the time, the present segment input signal that is mixed with echo signal from the outside that calculating receives obtains existing self adaptation residual echo signal with the difference of being somebody's turn to do existing adaptive modeling echo signal, should output to sef-adapting filter (311) by existing self adaptation residual echo signal, and will show self adaptation residual echo signal and existing self adaptation residual echo sign outputs to first control module (33).
Described main filtration module (32) comprising:
Senior filter (321), be used for providing filter factor to adaptive-filtering module (31), and according to this filter factor echogenicity signal the last period from the present segment of outside is carried out filtering and calculate, main analog echo signal before obtaining, should before main analog echo signal and preceding main echo sign output to second subtracter (322); Be used for behind filter factor after the once adjustment of receiving adaptive-filtering module (31) output, once adjust the back filter factor according to this present segment echogenicity signal from the outside is carried out filtering calculating, obtain now main analog echo signal, should output to second subtracter (322) with existing main echo sign by existing main analog echo signal; Then, when receiving the update coefficients triggering signal of first control module (33) output, the back filter factor of once adjusting with adaptive-filtering module (31) output upgrades the filter factor of self preserving, and sends the existing main residual echo triggering signal of output for second subtracter (322); When receiving the retention factor triggering signal of first control module (33) output, according to the filter factor of self preserving, present segment echogenicity signal is carried out filtering again calculate, analog echo signal and the existing main residual echo sign of renewal that obtains outputed to second subtracter (322);
Second subtracter (322), before receiving being designated of senior filter (321) output during the analog echo signal of main echo, calculating receive from the input signal that is mixed with echo the last period of the present segment of outside and should before the difference of main analog echo signal obtain before main residual echo signal, and should before main residual echo signal and preceding main residual echo identify and output to first control module (33); When receive senior filter (321) output be designated the analog echo signal of existing main echo the time, the input signal that the present segment from the outside that calculating receives is mixed with echo obtains now main residual echo signal with the difference of being somebody's turn to do existing main analog echo signal, and will show main residual echo signal and existing main residual echo sign outputs to first control module (33), and when the existing main residual echo triggering signal of the output that receives senior filter (321), will show main residual echo signal output; When receive senior filter (321) output be designated the analog echo signal that upgrades existing main residual echo the time, the present segment that calculating receives is mixed with the difference of the input signal and the analog echo signal of echo, be mixed with the input signal of echo and the existing main residual echo signal of difference renewal of analog echo signal with this present segment, and the existing main residual echo signal output that will upgrade.
Described device further comprises:
Backup filtration module (34) is used for according to the filter factor of self preserving present segment echogenicity signal being carried out filtering and calculates, and the backup residual echo signal that obtains is outputed to second control module (35) and comparison module (36); Be used for when receiving the filter factor of second control module (35) output, upgrading the filter factor of self preserving with this filter factor;
Second control module (35), be used for calculating respectively the average power of the existing main residual echo signal and the backup residual echo signal that backup filtration module (34) is exported of main filtration module (32) output that receives, when this two average power satisfied predetermined relationship, the filter factor of main filtration module (32) output that then will receive outputed to backup filtration module (34);
Comparison module (36) is used for the size of the backup residual echo signal of the main residual echo signal of more main filtration module (32) output and backup filtration module (34) output, the signal output after signal less in the two is eliminated as echo;
Main filtration module (32) is further used for, and when receiving the update coefficients triggering signal of first control module (33), will show main residual echo signal and output to second control module (35) and comparison module (36); When receiving the retention factor triggering signal of first control module (33), the filter factor that will preserve according to self carries out the existing main residual echo signal that filtering obtains to present segment echogenicity signal and outputs to second control module (35) and comparison module (36).
Described backup filtration module (34) comprising:
Backup filter (341) is used for according to the filter factor of self preserving the present segment echogenicity signal from the outside being carried out filtering and calculates, and the analog echo signal that obtains is outputed to the 3rd subtracter (342); Be used for when receiving the filter factor of second control module (35) output, upgrading the filter factor of self preserving with this filter factor;
The 3rd subtracter (342), be used to calculate the difference that present segment from the outside is mixed with the analog echo signal of the input signal of echo and backup filter (341) output and obtain backing up the residual echo signal, should back up the residual echo signal and output to second control module (35) and comparison module (36).
A kind of echo cancel method, this method comprises:
A, according to the main filter factor of current preservation, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before main residual echo signal; While is according to echogenicity signal the last period of present segment, main filter factor to current preservation carries out a self adaptation adjustment, according to the filter factor in the self adaptation adjustment process, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before self adaptation residual echo signal, and the once adjusted filter factor of record;
Whether the average power of main residual echo signal and the average power of preceding self adaptation residual echo signal satisfy predetermined relationship before B, the judgement, if, execution in step C; Otherwise, execution in step D.
C, the input signal that is mixed with echo according to once adjusted filter factor, present segment echogenicity signal and present segment in the steps A obtain now main residual echo signal; Simultaneously according to present segment echogenicity signal, this is once adjusted the back filter factor carry out the adjustment of secondary self adaptation, the input signal that is mixed with echo according to the filter factor in this secondary adjustment process, present segment echogenicity signal and present segment obtains existing self adaptation residual echo signal, judge whether the average power of existing main residual echo signal and the average power of existing self adaptation residual echo signal satisfy predetermined relationship, if, once to adjust the main filter factor of the current preservation among the filter factor step of updating A of back, execution in step E then in the steps A; Otherwise, execution in step D;
D, according to the main filter factor of current preservation, the input signal that present segment echogenicity signal and present segment are mixed with echo obtains now main residual echo signal;
E, will show the signal after main residual echo signal is eliminated as echo.
The described main filter factor of steps A, present segment according to current preservation the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before the step of main residual echo signal comprise:
Main filter factor according to current preservation, echogenicity signal the last period to present segment carries out filtering calculating, main analog echo signal before obtaining deducts main analog echo signal before this, main residual echo signal before obtaining with the input signal that is mixed with echo the last period of present segment;
Steps A described according to the filter factor in the self adaptation adjustment process, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before the step of self adaptation residual echo signal be:
According to the filter factor in the self adaptation adjustment process, echogenicity signal the last period to present segment carries out filtering calculating, adaptive modeling echo signal before obtaining, the input signal that is mixed with echo the last period of present segment is deducted this preceding adaptive modeling echo signal, self adaptation residual echo signal before obtaining;
The step that the described input signal that is mixed with echo according to once adjusted filter factor, present segment echogenicity signal and present segment in the steps A of step C obtains now main residual echo signal is:
According to the once adjusted filter factor in the steps A, present segment echogenicity signal is carried out filtering calculate, obtain now main analog echo signal, the input signal that present segment is mixed with echo deducts this existing main analog echo signal, obtains now main residual echo signal;
The step that the described input signal that is mixed with echo according to the filter factor in the secondary adjustment process, present segment echogenicity signal and present segment of step C obtains existing self adaptation residual echo signal is:
According to the filter factor in the secondary adjustment process, present segment echogenicity signal is carried out filtering to be calculated, obtain existing adaptive modeling echo signal, the input signal that present segment is mixed with echo deducts this existing adaptive modeling echo signal, obtains existing self adaptation residual echo signal.
The current backup filter factor of preserving, and, further comprise after the described step D, before the step e:
The backup filter factor of D1, this current preservation of basis carries out filtering to present segment echogenicity signal and calculates, obtain the analog echo signal, the input signal that present segment is mixed with echo deducts this analog echo signal and obtains backing up the residual echo signal, and calculates the average power of this backup residual echo signal correspondence; Judge whether continuously certain number of times satisfies predetermined relationship for the average power of backup residual echo signal and the average power of existing main residual echo signal, if, upgrade the backup filter factor of this current preservation, execution in step D2 then with adjusted filter factor in the steps A; Otherwise, direct execution in step D2;
D2, judge backup residual echo signal average power whether less than the average power of existing main residual echo signal, if, with the existing main residual echo signal of backup residual echo signal update; Otherwise, keep existing main residual echo signal constant.
The predetermined relationship that the average power of the average power of the described backup residual echo of step D1 signal and existing main residual echo signal satisfies is: the average power of existing main residual echo signal and the product of predetermined constant are less than or equal to the average power of backup residual echo signal.
The predetermined relationship that the average power of the average power of self adaptation residual echo signal and preceding main residual echo signal need satisfy before step B was described is: the average power of main residual echo signal and the product of predetermined constant before the average power of preceding self adaptation residual echo signal is less than or equal to;
The predetermined relationship that the average power of the described existing self adaptation residual echo signal of step C and the average power of existing main residual echo signal satisfy is: the average power of existing self adaptation residual echo signal is less than or equal to the product of the average power and the predetermined constant of now main residual echo signal.
Compared with prior art, main filtration module is according to the filter factor of current preservation, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before main residual echo, the adaptive-filtering module is a reference signal with echogenicity signal the last period of present segment simultaneously, the filter factor of current preservation is carried out a self adaptation adjustment obtain once adjusted filter factor, and according to the filter factor in the self adaptation adjustment process, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before self adaptation residual echo signal, after the average power of preceding main residual echo signal and preceding self adaptation residual echo signal satisfies predetermined condition, main filtration module is again according to adjusted filter factor once, the input signal that present segment echogenicity signal and present segment are mixed with echo obtains now main residual echo signal, the adaptive-filtering module is a reference signal with present segment echogenicity signal simultaneously, adjusted filter factor is once carried out the adjustment of secondary self adaptation, obtain the adjusted filter factor of secondary, and according to the filter factor in the self adaptation adjustment process, the input signal that present segment echogenicity signal and present segment are mixed with echo obtains now main residual echo signal, if the average power of existing main residual echo signal and existing self adaptation residual echo signal still satisfies predetermined condition, then upgrade the filter factor of current preservation with adjusted filter factor once.The present invention passes through the further checking to the predetermined condition of update coefficients, has guaranteed the reliability of the filter factor of preservation, thereby has guaranteed the effect that echo is eliminated.Simultaneously, filter factor must continue certain number of times and satisfy predetermined relationship, could finally further be preserved, guaranteed the accuracy of the filter factor preserved more, and will show less signal after eliminating as echo in main residual echo signal and the backup residual echo signal, further guarantee the reliability that echo is eliminated.
Description of drawings
Fig. 1 is an echo mechanism of production schematic diagram;
Fig. 2 is the structured flowchart that adds the Echo Canceller of double-talk testing mechanism in the prior art;
Fig. 3 is the structured flowchart of the specific embodiment one of realization echo provided by the invention elimination;
Fig. 4 is the structured flowchart of the specific embodiment two of realization echo provided by the invention elimination;
Fig. 5 is the structured flowchart of the specific embodiment three of realization echo provided by the invention elimination;
Fig. 6 is the structured flowchart of the specific embodiment four of realization echo provided by the invention elimination;
Fig. 7 is the flow chart of the specific embodiment one of realization echo provided by the invention elimination;
Fig. 8 is the flow chart of the specific embodiment two of realization echo provided by the invention elimination.
Embodiment
The present invention is further described in more detail below in conjunction with drawings and the specific embodiments.
Fig. 3 is the structured flowchart of the specific embodiment one of echo cancelling device provided by the invention, and the x among the figure (i) is echogenic signal, is called the echogenicity signal, i.e. the signal from far-end of near-end reception; G (i) is an echo signal; V (i) is a voice signal, and promptly near-end will send to the voice signal of far-end; D (n) is the voice signal that is mixed with echo, i.e. d (i)=v (i)+g (i), as can be seen, and when v (i)=0, d (i)=g (i), as shown in Figure 3, it mainly comprises:
Adaptive-filtering module 31: be used for when the present segment echogenicity signal x (i) that receives from the outside, preserve x (i), and to main filtration module 32 transmission request coefficient triggering signals, when receiving the filter coefficient alpha 1 of main filtration module 32 outputs, according to echogenicity signal x the last period (i-1) that has preserved α 1 is carried out a self adaptation adjustment from the present segment of outside, obtain preceding adaptive modeling echo signal according to the filter factor in x (i-1) and the self adaptation adjustment process, self adaptation residual echo signal before obtaining according to this analog echo signal with from the input signal d (i-1) that is mixed with echo the last period of the present segment of outside then, and preceding self adaptation residual echo signal and preceding self adaptation residual echo sign 00 outputed to first control module 33, establish α 1 and after adjusting, become α
I-1
When receiving the continuation filtering triggering signal 0 of first control module 33, with α
I-1Output to main filtration module 32, and according to present segment echogenicity signal x (i), to the filter coefficient alpha of record
I-1Carry out the adjustment of secondary self adaptation, and obtain existing adaptive modeling echo signal according to the filter factor in x (i) and the self adaptation adjustment process, obtain existing self adaptation residual echo signal according to this analog echo signal with from the input signal d (i) that the present segment of outside is mixed with echo, and should output to first control module 33 with existing self adaptation residual echo sign 01 by existing self adaptation residual echo signal, establish α
I-1After adjusting, become α
i
The concrete length of every segment signal can be determined according to actual needs, but should satisfy: the adaptive-filtering module is identical with the length of every segment signal that main filtration module is provided with.
Main filtration module 32: after the request coefficient triggering signal that receives adaptive-filtering module 31, the filter coefficient alpha 1 of self preserving is outputed to adaptive-filtering module 31;
When the present segment echogenicity signal x (i) that receives from the outside, preserve x (i), and carry out filtering according to signal x the last period (i-1) that 1 pair of the filter coefficient alpha that self preserves has been preserved and calculate from the present segment of outside, main analog echo signal before obtaining, main residual echo signal before obtaining according to this analog echo signal and d (i-1) then, and should before main residual echo signal and preceding main residual echo identify 10 and output to first control module 33;
Be used to be received from the filter coefficient alpha of adaptive filtering module output
I-1, and according to α
I-1The present segment echogenicity signal x (i) from the outside that receives and preserve is carried out filtering to be calculated, obtain now main analog echo signal, obtain now main residual echo signal according to this analog echo signal and d (i) then, and should output to first control module 33 with existing main residual echo sign 11 by existing main residual echo signal;
When receiving the update coefficients triggering signal 0 of first control module, 33 outputs, with the filter coefficient alpha of adaptive-filtering module 31 outputs
I-1Upgrade the filter coefficient alpha 1 of self preserving, and the signal that will show after main residual echo signal is eliminated as echo outputs to far-end or near-end;
When receiving the retention factor triggering signal 1 of first control module, 33 outputs, carrying out filtering according to 1 pair of present segment echogenicity of the filter coefficient alpha signal x (i) that self preserves calculates, obtain new existing main analog echo signal, obtain new existing main residual echo signal according to this analog echo signal and d (i) then, and the signal that will be somebody's turn to do after new existing main residual echo signal is eliminated as echo with the existing main residual echo sign 12 of renewal outputs to far-end.
First control module 33: be used for calculating respectively the preceding self adaptation residual echo signal of adaptive-filtering module 31 outputs and the average power p0 (i-1) and the p1 (i-1) of the preceding main residual echo signal that main filtration module 32 is exported, if p0 (i-1) and p1 (i-1) satisfy predetermined relationship, send for adaptive-filtering module 31 and continue filtering triggering signal 0, otherwise, send retention factor triggering signal 1 for main filtration module 32; Be used for calculating respectively the existing self adaptation residual echo signal of adaptive-filtering module 31 outputs and the average power p0 (i) and the p1 (i) of the existing main residual echo signal that main filtration module 32 is exported, if p0 (i) and p1 (i) satisfy predetermined relationship, send update coefficients triggering signal 0 for main filtration module 32; Otherwise, send retention factor triggering signal 1 for main filtration module 32.
Fig. 4 is the structured flowchart of the specific embodiment two of echo cancelling device provided by the invention, and as shown in Figure 4, this figure compares with Fig. 3,
Adaptive-filtering module 31 comprises:
Sef-adapting filter 311:
One, when the present segment echogenicity signal x (i) that receives from the outside, preserves x (i), and send request coefficient triggering signals to senior filter 321;
Two, when receiving the filter coefficient alpha 1 of senior filter 321 outputs, be reference signal, be error signal with echogenicity signal x the last period (i-1) that has preserved with the self adaptation residual echo signals of first subtracter, 312 the last outputs from the present segment of outside, α 1 is carried out a self adaptation adjustment, go out preceding adaptive modeling echo signal according to the filter coefficient estimation in x (i-1) and the self adaptation adjustment process, this analog echo signal and preceding adaptive echo sign 00 are outputed to first subtracter 312, establish α 1 and after adjusting, become α
I-1
Three, after the continuation filtering triggering signal 0 that receives first control module 33, with α
I-1Outputing to senior filter 321, and be reference signal with present segment echogenicity signal x (i), is error signal with the preceding self adaptation residual echo signals of first subtracter, the 312 the last outputs that receive, to α
I-1Carry out the adjustment of secondary self adaptation, the adaptive modeling echo signal occurs, this analog echo signal and existing adaptive echo sign 01 are outputed to first subtracter 312, establish α according to the filter coefficient estimation in x (i) and the self adaptation adjustment process
I-1After adjusting, become α
i
First subtracter 312:
One, when receive sef-adapting filter 311 output be designated 00 preceding adaptive modeling echo signal the time, calculating receive from the input signal d (i-1) that is mixed with echo the last period of the present segment of outside and the difference of this analog echo signal, this difference is preceding self adaptation residual echo signal, should before self adaptation residual echo signal output to sef-adapting filter 311, and should before self adaptation residual echo signal and preceding self adaptation residual echo identify 00 and output to first control module 33;
Two, when receive sef-adapting filter 311 output be designated 01 existing adaptive modeling echo signal the time, the present segment from the outside that calculating receives is mixed with the input signal d (i) of echo and the difference of this analog echo signal, this difference is existing self adaptation residual echo signal, should output to sef-adapting filter 311 by existing self adaptation residual echo signal, and should output to first control module 33 with existing self adaptation residual echo sign 01 by existing self adaptation residual echo signal.
Senior filter 321:
One, when receiving the request coefficient triggering signal of sef-adapting filter 311, the filter coefficient alpha 1 of self preserving is outputed to sef-adapting filter 311;
Two, when the present segment echogenicity signal x (i) that receives from the outside, preserve x (i), and carry out filtering according to echogenicity signal x the last period (i-1) from the present segment of outside that 1 pair of the filter coefficient alpha that self preserves has been preserved and calculate, preceding main analog echo signal and the preceding main echo sign 10 that obtains outputed to second subtracter 322;
Three, be used to receive and preserve the filter coefficient alpha of sef-adapting filter 311 outputs
I-1, according to this filter factor the present segment echogenicity signal x (i) from the outside that receives and preserve is carried out filtering and calculate, existing main analog echo signal and the existing main echo sign 11 that obtains outputed to second subtracter 322; And when receiving the update coefficients triggering signal 0 of first control module, 33 outputs, with filter coefficient alpha
I-1Upgrade the filter coefficient alpha 1 of self preserving, and send the existing main residual echo triggering signal 0 of output for second subtracter 322.
Four, when receiving the retention factor triggering signal 1 of first control module, 33 outputs, according to the filter coefficient alpha 1 of self preserving, present segment echogenicity signal x (i) is carried out filtering calculate, new existing main analog echo signal and the existing main residual echo sign 12 of renewal that obtains outputed to second subtracter 322.
Second subtracter 322:
One, when receive senior filter 321 output be designated 10 preceding main analog echo signal the time, calculating receive from the input signal d (i-1) that is mixed with echo the last period of the present segment of outside and the difference of this analog echo signal, this difference is preceding main residual echo signal, and should before main residual echo signal and preceding main residual echo sign 10 output to first control module 33;
Two, when receive senior filter 321 output be designated 11 existing main analog echo signal the time, the present segment from the outside that calculating receives is mixed with the input signal d (i) of echo and the difference of this analog echo signal, this difference is now main residual echo signal, and should output to first control module 33 with existing main residual echo sign 11 by existing main residual echo signal, and when the existing main residual echo triggering signal 0 of the output that receives senior filter 321, the signal after existing main residual echo signal eliminated as echo outputs to far-end.
Three, when receive senior filter 321 output be designated 12 new existing main analog echo signal the time, the present segment that calculating receives is mixed with the input signal d (i) of echo and the difference of this analog echo signal, this difference is new existing main residual echo signal, and the signal that will be somebody's turn to do after new existing main residual echo signal is eliminated as echo outputs to far-end.
First control module 33: be used for calculating respectively the preceding self adaptation residual echo signal of first subtracter, 312 outputs that receive and the average power p0 (i-1) and the p1 (i-1) of the preceding main residual echo signal that second subtracter 322 is exported, and judge whether p0 (i-1) and p1 (i-1) satisfy predetermined relationship, if send for sef-adapting filter 311 and continue the filtering triggering signal; Otherwise, send retention factor triggering signal 1 for senior filter 321; Be used for calculating respectively the existing self adaptation residual echo signal of first subtracter, 312 outputs that receive and the average power p0 (i) and the p1 (i) of the existing main residual echo signal correspondence that second subtracter 322 is exported, and judge whether p0 (i) and p1 (i) satisfy predetermined relationship, if send update coefficients triggering signal 0 for senior filter 321; Otherwise, send retention factor triggering signal 1 for senior filter 321.
Fig. 5 is the structured flowchart of the specific embodiment three of echo cancelling device provided by the invention, and as shown in Figure 5, this figure compares with Fig. 3, further comprises:
Backup filtration module 34: be used for when receiving the filter coefficient alpha 1 of second control module, 35 outputs, upgrading the filter coefficient alpha 2 of self preserving with α 1; Being used for carrying out filtering according to the present segment echogenicity signal x (i) from the outside that 2 couples of α have received calculates, obtain backing up the analog echo signal, obtain backing up the residual echo signal according to this analog echo signal with from the d (i) of outside, should back up the residual echo signal and output to second control module 35 and comparison module 36.
The length of every segment signal that the backup filtration module is provided with is identical with main filtration module with the adaptive-filtering module.
Second control module 35: the filter coefficient alpha 1 that is used to receive main filtration module 32 outputs, calculate the existing main residual echo signal of main filtration module 32 outputs that receive and the average power p1 (i) and the p2 (i) of the backup residual echo signal correspondence that backup filtration module 34 is exported respectively, when p1 (i) and p2 (i) continuous N time satisfy predetermined relationship, then filter coefficient alpha 1 is outputed to backup filtration module 34.
Here, M is predefined value.
Comparison module 36: be used for the size of the main residual echo signal and the backup residual echo signal that backup filtration module 34 is exported of more main filtration module 32 outputs, the signal after signal less in the two is eliminated as echo outputs to far-end.
Simultaneously, main filtration module 32 is further used for:
One, when receiving the update coefficients triggering signal 0 of first control module 33, will show main residual echo signal and output to second control module 35 and comparison module 36.
Two, when receiving the retention factor triggering signal 1 of first control module 33, newly existing main residual echo signal is outputed to second control module 35 and comparison module 36.
Fig. 6 is the structured flowchart of the specific embodiment four of echo cancelling device provided by the invention, and as shown in Figure 6, the adaptive-filtering module 31 of this figure, main filtration module 32, first control module 33 are identical with Fig. 4, compare with Fig. 5,
Backup filter 341: be used for when receiving the filter coefficient alpha 1 of second control module, 35 outputs, upgrading the filter coefficient alpha 2 of self preserving with α 1; Be used for carrying out filtering and calculate, obtain backing up the analog echo signal, this analog echo signal is outputed to the 3rd subtracter 342 according to 2 couples of present segment echogenicity signal x (i) of α from the outside.
The 3rd subtracter 342: be used to calculate the difference that is mixed with the input signal d (i) and the backup analog echo signal that backup filter 341 is exported of echo from the present segment of outside, this difference is backup residual echo signal, should back up the residual echo signal and output to second control module 35 and comparison module 36.
Fig. 7 is the flow chart of the specific embodiment one of realization echo provided by the invention elimination, and as shown in Figure 7, its concrete steps are as follows:
701: echogenicity signal x the last period (i-1) with present segment is a reference signal, self adaptation residual echo signal with the last time is an error signal, filter coefficient alpha 1 to current preservation is carried out a self adaptation adjustment, obtain preceding adaptive modeling echo signal according to filter factor in the self adaptation adjustment process and x (i-1), the difference of calculating d (i-1) and this analog echo signal obtains preceding self adaptation residual echo signal, and establish after the self adaptation adjustment finishes, α 1 becomes α
I-1According to α 1, x (i-1) is carried out filtering calculate simultaneously, main analog echo signal before obtaining calculates the input signal d (i-1) that is mixed with echo the last period of present segment and the difference of this analog echo signal then, main residual echo signal before obtaining.
The concrete length of every segment signal can set in advance according to actual needs.
702: the average power p0 (i-1) and the p1 (i-1) of self adaptation residual echo signal and preceding main residual echo signal correspondence before calculating respectively.
703: judge whether p0 (i-1)≤β 1p1 (i-1) sets up, and wherein, β 1 is a constant, if, execution in step 704; Otherwise, execution in step 708.
Step 704: with present segment echogenicity signal x (i) is reference signal, is error signal with the self adaptation residual echo signal of the last time, to α
I-1Carry out the adjustment of secondary self adaptation, obtain existing adaptive modeling echo signal according to filter factor in the self adaptation adjustment process and x (i), the difference of calculating d (i) and this analog echo signal obtains existing self adaptation residual echo signal.After if the self adaptation adjustment finishes, α
I-1Become α
iSimultaneously according to filter coefficient alpha
I-1, x (i) is carried out filtering calculate, obtain now main analog echo signal, calculate present segment then and be mixed with the input signal d (i) of echo and the difference of this analog echo signal, obtain now main residual echo signal.
705: the average power p0 (i) and the p1 (i) that calculate existing self adaptation residual echo signal and existing main residual echo signal correspondence respectively.
706: judge whether p0 (i)≤β 2p1 (i) sets up, and wherein, β 2 is a constant, if judge the current double-talk state that is not in, execution in step 707; Otherwise, execution in step 708.
707: with α
I-1Upgrade the α 1 of current preservation, and will show the signal after main residual echo signal is eliminated as echo, this flow process finishes.
708: again x (i) is carried out filtering according to filter coefficient alpha 1 and calculate, obtain newly existing main analog echo signal, the difference of calculating d (i) and this analog echo signal obtains new existing main residual echo signal, the signal after this newly existing main residual echo signal is eliminated as echo.
As can be seen from the above step, after judging that p0 (i-1)≤β 1p1 (i-1) sets up, not immediately with α
I-1Upgrade the filter coefficient alpha 1 of current preservation, this be because: only can not whether be in the double-talk state by full confirmation x (i-1), at this moment might be in the double-talk state but still satisfy p0 (i-1)≤β 1p1 (i-1) at last several points of x (i-1) according to p0 (i-1)≤β 1p1 (i-1).Therefore, continue α according to x (i) here
I-1Carry out the self adaptation adjustment, by judging whether p0 (i)≤β 2p1 (i) sets up, confirm further whether x (i-1) is not in the double-talk state then.Owing to carried out back check, can reduce the probability that mistake is preserved filter factor effectively.
The method that provides from Fig. 7 as can be seen, the present invention has carried out further checking before the filter factor that upgrades current preservation, can reduce the probability that the filter factor mistake is preserved like this.Reasoning according to this can be learnt: if further the filter factor that satisfies update condition is verified, can further reduce the probability that the filter factor mistake is preserved so.With Fig. 8 is that example specifies this process, and Fig. 8 compares with Fig. 7:
Step 801~802 are identical with step 701~702.
Step 803 is: judge whether p0 (i-1)≤β 1p1 (i-1) sets up, if, execution in step 804; Otherwise, execution in step 808.
Step 804~805 are identical with step 704~705.
Step 806 is: judge whether p0 (i)≤β 2p1 (i) sets up, if judge the current double-talk state that is not in, execution in step 807; Otherwise, execution in step 808.
Step 807 with the difference of step 707 is: will not show the signal after main residual echo signal is eliminated as echo, but go to step 810.
Step 808 with the difference of step 708 is: after calculating new existing main residual echo signal, signal after it not being eliminated as echo, but replace existing main residual echo signal with this newly existing main residual echo signal, and calculate its corresponding average power p1 (i), go to step 810 then.
809: according to the filter coefficient alpha 2 of current preservation, present segment echogenicity signal x (i) is carried out filtering to be calculated, obtain backing up the analog echo signal, and the difference of calculating d (i) and this analog echo signal obtains backing up the residual echo signal, calculates the average power P2 (i) of this backup residual echo signal correspondence then.
Step 809 and step 801~808 can be carried out simultaneously.
810: judge whether γ p1 (i)<p2 (i) sets up, and wherein, γ is a constant, if, execution in step 812; Otherwise, execution in step 811.
811: make m=0, execution in step 815 then.
M is used for the number of times that inequality γ p1 (the i)<p2 (i) of markers step 810 sets up continuously.
812: write down the number of times m that this inequality is set up continuously.
When calculating the number of times of the continuous establishment of this inequality, to judge at first whether the last inequality is set up, if, continue the number of times that the inquiry inequality has been set up, this number of times is added 1 just obtain m; Otherwise, m=1.Preserve the value of m and the sign whether inequality is set up, use when inquiring about in order to next time.
813: judge whether m<M sets up, if, execution in step 815; Otherwise, execution in step 814.
M is predefined integer.
814: make m=0, upgrade α 2 with filter coefficient alpha 1 simultaneously.
815: whether judge p1 (i) less than p2 (i), if, execution in step 816; Otherwise, execution in step 817.
816: will show the signal after main residual echo signal is eliminated as echo, this flow process finishes.
817: will back up the signal after the residual echo signal is eliminated as echo.
By above step 810~817 as can be seen, only when continuing to satisfy for M time predetermined condition γ p1 (i)<p2 (i), could upgrade α 2 with filter coefficient alpha 1, this has further reduced the probability that α 2 mistakes are preserved, to show less signal after eliminating as echo in main residual echo signal and the backup residual echo signal simultaneously, further guarantee the effect that echo is eliminated.
The present invention can be applicable to the electric echo elimination and acoustic echo is eliminated.
The above only is process of the present invention and method embodiment, in order to restriction the present invention, all any modifications of being made within the spirit and principles in the present invention, is not equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1, a kind of echo cancelling device is characterized in that, this device comprises: adaptive-filtering module (31), main filtration module (32) and first control module (33),
Adaptive-filtering module (31), when the present segment echogenicity signal of receiving from the outside, the filter factor that main filtration module (32) is preserved carries out a self adaptation adjustment, and will output to first control module (33) according to the preceding self adaptation residual echo signal that the last period of present segment, the echogenicity signal obtained;
Main filtration module (32) when the present segment echogenicity signal of receiving from the outside, will carry out the preceding main residual echo signal that filtering calculates to echogenicity signal the last period of present segment and output to first control module (33);
First control module (33), whether the average power of self adaptation residual echo signal and preceding main residual echo signal satisfies predetermined relationship before judging, if send for adaptive-filtering module (31) and continue the filtering triggering signal, otherwise, send the retention factor triggering signal for main filtration module (32);
Wherein, after adaptive-filtering module (31) is received and is continued the filtering triggering signal, to once adjust the back filter factor and output to main filtration module (32), and carry out the adjustment of secondary self adaptation to once adjusting the back filter factor according to present segment echogenicity signal, the existing self adaptation residual echo signal that obtains is outputed to first control module (33); Main filtration module (32) will carry out the existing main residual echo signal that filtering calculates and output to first control module (33) according to once adjusting back filter factor, present segment echogenicity signal; First control module (33) judges whether the average power of existing self adaptation residual echo signal and existing main residual echo signal satisfies predetermined relationship, if, send the update coefficients triggering signal to main filtration module (32), main filtration module (32) upgrades the filter factor of self preserving once to adjust the back filter factor, and will show main residual echo signal output; Otherwise, sending the retention factor triggering signal to main filtration module (32), main filtration module (32) will be exported according to the existing main residual echo signal that filter factor, the present segment echogenicity signal self preserved obtain.
2, device as claimed in claim 1 is characterized in that, described adaptive-filtering module (31) comprising:
Sef-adapting filter (311) is used for obtaining filter factor to main filtration module (32) when the present segment echogenicity signal of receiving from the outside; When receiving the filter factor of main filtration module (32) output, with echogenicity signal the last period from the present segment of outside is reference signal, self adaptation residual echo signal with the last output of first subtracter (312) is an error signal, filter factor to this main filtration module (32) output carries out a self adaptation adjustment, according to the filter factor in the self adaptation adjustment process echogenicity signal the last period of present segment being carried out filtering calculates, adaptive modeling echo signal before obtaining, should before adaptive modeling echo signal and preceding adaptive echo identify and output to first subtracter (312); Be used for after receiving the continuation filtering triggering signal of first control module (33), with present segment echogenicity signal is reference signal, preceding self adaptation residual echo signal with the last output of first subtracter (312) is an error signal, adjusted filter factor is once carried out the adjustment of secondary self adaptation, according to the filter factor in this secondary self adaptation adjustment process present segment echogenicity signal being carried out filtering calculates, obtain existing adaptive modeling echo signal, should output to first subtracter (312) with existing adaptive echo sign by existing adaptive modeling echo signal;
First subtracter (312), before receiving being designated of sef-adapting filter (311) output during the analog echo signal of adaptive echo, calculating receive from the input signal that is mixed with echo the last period of the present segment of outside and should before the difference of adaptive modeling echo signal obtain before self adaptation residual echo signal, should before self adaptation residual echo signal output to sef-adapting filter (311), and should before self adaptation residual echo signal and preceding self adaptation residual echo identify and output to first control module (33); When receive sef-adapting filter (311) output be designated the analog echo signal of existing adaptive echo the time, the present segment input signal that is mixed with echo signal from the outside that calculating receives obtains existing self adaptation residual echo signal with the difference of being somebody's turn to do existing adaptive modeling echo signal, should output to sef-adapting filter (311) by existing self adaptation residual echo signal, and will show self adaptation residual echo signal and existing self adaptation residual echo sign outputs to first control module (33).
3, device as claimed in claim 1 is characterized in that, described main filtration module (32) comprising:
Senior filter (321), be used for providing filter factor to adaptive-filtering module (31), and according to this filter factor echogenicity signal the last period from the present segment of outside is carried out filtering and calculate, main analog echo signal before obtaining, should before main analog echo signal and preceding main echo sign output to second subtracter (322); Be used for behind filter factor after the once adjustment of receiving adaptive-filtering module (31) output, once adjust the back filter factor according to this present segment echogenicity signal from the outside is carried out filtering calculating, obtain now main analog echo signal, should output to second subtracter (322) with existing main echo sign by existing main analog echo signal; Then, when receiving the update coefficients triggering signal of first control module (33) output, the back filter factor of once adjusting with adaptive-filtering module (31) output upgrades the filter factor of self preserving, and sends the existing main residual echo triggering signal of output for second subtracter (322); When receiving the retention factor triggering signal of first control module (33) output, according to the filter factor of self preserving, present segment echogenicity signal is carried out filtering again calculate, analog echo signal and the existing main residual echo sign of renewal that obtains outputed to second subtracter (322);
Second subtracter (322), before receiving being designated of senior filter (321) output during the analog echo signal of main echo, calculating receive from the input signal that is mixed with echo the last period of the present segment of outside and should before the difference of main analog echo signal obtain before main residual echo signal, and should before main residual echo signal and preceding main residual echo identify and output to first control module (33); When receive senior filter (321) output be designated the analog echo signal of existing main echo the time, the input signal that the present segment from the outside that calculating receives is mixed with echo obtains now main residual echo signal with the difference of being somebody's turn to do existing main analog echo signal, and will show main residual echo signal and existing main residual echo sign outputs to first control module (33), and when the existing main residual echo triggering signal of the output that receives senior filter (321), will show main residual echo signal output; When receive senior filter (321) output be designated the analog echo signal that upgrades existing main residual echo the time, the present segment that calculating receives is mixed with the difference of the input signal and the analog echo signal of echo, be mixed with the input signal of echo and the existing main residual echo signal of difference renewal of analog echo signal with this present segment, and the existing main residual echo signal output that will upgrade.
4, device as claimed in claim 1 is characterized in that, described device further comprises:
Backup filtration module (34) is used for according to the filter factor of self preserving present segment echogenicity signal being carried out filtering and calculates, and the backup residual echo signal that obtains is outputed to second control module (35) and comparison module (36); Be used for when receiving the filter factor of second control module (35) output, upgrading the filter factor of self preserving with this filter factor;
Second control module (35), be used for calculating respectively the average power of the existing main residual echo signal and the backup residual echo signal that backup filtration module (34) is exported of main filtration module (32) output that receives, when this two average power satisfied predetermined relationship, the filter factor of main filtration module (32) output that then will receive outputed to backup filtration module (34);
Comparison module (36) is used for the size of the backup residual echo signal of the main residual echo signal of more main filtration module (32) output and backup filtration module (34) output, the signal output after signal less in the two is eliminated as echo;
Main filtration module (32) is further used for, and when receiving the update coefficients triggering signal of first control module (33), will show main residual echo signal and output to second control module (35) and comparison module (36); When receiving the retention factor triggering signal of first control module (33), the filter factor that will preserve according to self carries out the existing main residual echo signal that filtering obtains to present segment echogenicity signal and outputs to second control module (35) and comparison module (36).
5, device as claimed in claim 4 is characterized in that, described backup filtration module (34) comprising:
Backup filter (341) is used for according to the filter factor of self preserving the present segment echogenicity signal from the outside being carried out filtering and calculates, and the analog echo signal that obtains is outputed to the 3rd subtracter (342); Be used for when receiving the filter factor of second control module (35) output, upgrading the filter factor of self preserving with this filter factor;
The 3rd subtracter (342), be used to calculate the difference that present segment from the outside is mixed with the analog echo signal of the input signal of echo and backup filter (341) output and obtain backing up the residual echo signal, should back up the residual echo signal and output to second control module (35) and comparison module (36).
6, a kind of echo cancel method is characterized in that, this method comprises:
A, according to the main filter factor of current preservation, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before main residual echo signal; While is according to echogenicity signal the last period of present segment, main filter factor to current preservation carries out a self adaptation adjustment, according to the filter factor in the self adaptation adjustment process, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before self adaptation residual echo signal, and the once adjusted filter factor of record;
Whether the average power of main residual echo signal and the average power of preceding self adaptation residual echo signal satisfy predetermined relationship before B, the judgement, if, execution in step C; Otherwise, execution in step D.
C, the input signal that is mixed with echo according to once adjusted filter factor, present segment echogenicity signal and present segment in the steps A obtain now main residual echo signal; Simultaneously according to present segment echogenicity signal, this is once adjusted the back filter factor carry out the adjustment of secondary self adaptation, the input signal that is mixed with echo according to the filter factor in this secondary adjustment process, present segment echogenicity signal and present segment obtains existing self adaptation residual echo signal, judge whether the average power of existing main residual echo signal and the average power of existing self adaptation residual echo signal satisfy predetermined relationship, if, once to adjust the main filter factor of the current preservation among the filter factor step of updating A of back, execution in step E then in the steps A; Otherwise, execution in step D;
D, according to the main filter factor of current preservation, the input signal that present segment echogenicity signal and present segment are mixed with echo obtains now main residual echo signal;
E, will show the signal after main residual echo signal is eliminated as echo.
7, method as claimed in claim 6 is characterized in that,
The described main filter factor of steps A, present segment according to current preservation the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before the step of main residual echo signal comprise:
Main filter factor according to current preservation, echogenicity signal the last period to present segment carries out filtering calculating, main analog echo signal before obtaining deducts main analog echo signal before this, main residual echo signal before obtaining with the input signal that is mixed with echo the last period of present segment;
Steps A described according to the filter factor in the self adaptation adjustment process, present segment the last period the echogenicity signal and the input signal that is mixed with echo the last period of present segment obtain before the step of self adaptation residual echo signal be:
According to the filter factor in the self adaptation adjustment process, echogenicity signal the last period to present segment carries out filtering calculating, adaptive modeling echo signal before obtaining, the input signal that is mixed with echo the last period of present segment is deducted this preceding adaptive modeling echo signal, self adaptation residual echo signal before obtaining;
The step that the described input signal that is mixed with echo according to once adjusted filter factor, present segment echogenicity signal and present segment in the steps A of step C obtains now main residual echo signal is:
According to the once adjusted filter factor in the steps A, present segment echogenicity signal is carried out filtering calculate, obtain now main analog echo signal, the input signal that present segment is mixed with echo deducts this existing main analog echo signal, obtains now main residual echo signal;
The step that the described input signal that is mixed with echo according to the filter factor in the secondary adjustment process, present segment echogenicity signal and present segment of step C obtains existing self adaptation residual echo signal is:
According to the filter factor in the secondary adjustment process, present segment echogenicity signal is carried out filtering to be calculated, obtain existing adaptive modeling echo signal, the input signal that present segment is mixed with echo deducts this existing adaptive modeling echo signal, obtains existing self adaptation residual echo signal.
8, method as claimed in claim 6 is characterized in that, the current backup filter factor of preserving, and, further comprise after the described step D, before the step e:
The backup filter factor of D1, this current preservation of basis carries out filtering to present segment echogenicity signal and calculates, obtain the analog echo signal, the input signal that present segment is mixed with echo deducts this analog echo signal and obtains backing up the residual echo signal, and calculates the average power of this backup residual echo signal correspondence; Judge whether continuously certain number of times satisfies predetermined relationship for the average power of backup residual echo signal and the average power of existing main residual echo signal, if, upgrade the backup filter factor of this current preservation, execution in step D2 then with adjusted filter factor in the steps A; Otherwise, direct execution in step D2;
D2, judge backup residual echo signal average power whether less than the average power of existing main residual echo signal, if, with the existing main residual echo signal of backup residual echo signal update; Otherwise, keep existing main residual echo signal constant.
9, method as claimed in claim 8, it is characterized in that the predetermined relationship that the average power of the average power of the described backup residual echo of step D1 signal and existing main residual echo signal satisfies is: the average power of existing main residual echo signal and the product of predetermined constant are less than or equal to the average power of backup residual echo signal.
10, method as claimed in claim 6, it is characterized in that the predetermined relationship that the average power of the average power of self adaptation residual echo signal and preceding main residual echo signal need satisfy before step B was described is: the average power of main residual echo signal and the product of predetermined constant before the average power of preceding self adaptation residual echo signal is less than or equal to;
The predetermined relationship that the average power of the described existing self adaptation residual echo signal of step C and the average power of existing main residual echo signal satisfy is: the average power of existing self adaptation residual echo signal is less than or equal to the product of the average power and the predetermined constant of now main residual echo signal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2005100982114A CN100550951C (en) | 2005-09-01 | 2005-09-01 | A kind of echo cancelling device and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2005100982114A CN100550951C (en) | 2005-09-01 | 2005-09-01 | A kind of echo cancelling device and method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1925526A CN1925526A (en) | 2007-03-07 |
CN100550951C true CN100550951C (en) | 2009-10-14 |
Family
ID=37817980
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2005100982114A Expired - Fee Related CN100550951C (en) | 2005-09-01 | 2005-09-01 | A kind of echo cancelling device and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100550951C (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9154635B2 (en) | 2008-09-26 | 2015-10-06 | Nec Corporation | Signal processing method, signal processing device, and signal processing program |
CN105323395B (en) * | 2014-07-31 | 2020-01-14 | 锐迪科(重庆)微电子科技有限公司 | Echo eliminator and echo eliminating method |
CN104954595B (en) * | 2015-05-15 | 2017-07-25 | 百度在线网络技术(北京)有限公司 | Residual echo removing method and device |
CN107786755B (en) * | 2016-08-30 | 2020-12-15 | 合肥君正科技有限公司 | Double-end call detection method and device |
CN107277209B (en) * | 2017-07-27 | 2020-05-29 | 维沃移动通信有限公司 | Call adjustment method and mobile terminal |
CN108712585A (en) * | 2018-05-09 | 2018-10-26 | 质音通讯科技(深圳)有限公司 | Call terminal, the initial method of filter, device, equipment and storage medium |
CN111277718B (en) * | 2020-01-21 | 2021-10-08 | 上海推乐信息技术服务有限公司 | Echo cancellation system and method thereof |
-
2005
- 2005-09-01 CN CNB2005100982114A patent/CN100550951C/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN1925526A (en) | 2007-03-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP0898816B1 (en) | Echo suppressor and non-linear processor of echo canceller | |
US6507653B1 (en) | Desired voice detection in echo suppression | |
US5631900A (en) | Double-Talk detector for echo canceller | |
CN100550951C (en) | A kind of echo cancelling device and method | |
JP4568439B2 (en) | Echo suppression device | |
US20090028354A1 (en) | Echo Canceller Employing Dual-H Architecture Having Split Adaptive Gain Settings | |
US5764753A (en) | Half-duplex controller | |
JP4582562B2 (en) | Method and apparatus for estimating and suppressing echo | |
JP2002033683A (en) | Network echo cancelation system and echo cancelation method | |
KR20080106122A (en) | Double talk detection method based on spectral acoustic properties | |
US5675644A (en) | Method and apparatus for canceling echo accounting for delay variations | |
CN100593287C (en) | Echo eliminating method, echo eliminating device and system equipment | |
US6757384B1 (en) | Robust double-talk detection and recovery in a system for echo cancelation | |
US8121260B2 (en) | Method and apparatus for retraining of echo cancellation | |
US7856087B2 (en) | Circuit method and system for transmitting information | |
JP2001044896A (en) | Speech unit and speech method | |
US6839427B2 (en) | Method and apparatus for echo canceller automatic gain control | |
JP3220979B2 (en) | Voice switch | |
JP2002280938A (en) | Echo canceller | |
JPH07226697A (en) | Echo canceler, echo canceling method and transmitter/ receiver having echo canceler | |
JPH07121027B2 (en) | Loud phone | |
JP3303524B2 (en) | Echo canceller learning method | |
JP3187716B2 (en) | Echo canceller | |
KR20080102712A (en) | Method and apparatus for cancelling echo of communication system | |
CN113921029A (en) | Double-end sounding detection method applied to echo cancellation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20091014 |
|
CF01 | Termination of patent right due to non-payment of annual fee |