CN1701528A - Speech communication apparatus - Google Patents

Abstract

A speech communication apparatus not only for preventing echo suppression capability from degrading even when the positional relationship between a speaker and a microphone is changed but also for allowing a high quality speech communication to be performed regardless of the positional relationship between a near-end person and the microphone. A current value determining part (131) is provided that determines the current value of a parameter defining the positional relationship between a remote-end signal converting part (112) and one of first (121) and second (122) near-end voice converting parts. First (133) and second (134) near-end signal delaying parts are used to delay near-end signals in accordance with the current value determined by the current value determining part (131). The difference between the near-end signals as delayed is outputted, as a near-end signal as echo-suppressed, from a first output part (135).

Description

Communicator

Technical field

The present invention relates to communicator, particularly under the situation that the relative position of microphone and loud speaker has changed, also can suppress the communicator of echo reliably.

Background technology

As existing communicator, there is and possesses a microphone, suppress the communicator of echo as everyone knows by the sef-adapting filter of the acoustics transmission characteristic between analog speakers and the microphone.

For example, (Japan) spy drives in the 2000-244670 communique disclosed existing communicator as shown in figure 25, near-end person side at the voice switch 51 handled conversed simultaneously comprises first echo eliminator 52, is comprising second echo eliminator 53 across voice switch 51 side opposite with first echo eliminator 52.

First echo eliminator 52 is by generating false echo according to the transfer function of setting in first echo eliminator 52 of the propagation path of 55 voice from loud speaker 54 to microphone, from the near end signal of microphone 55 output, deduct false echo with subtracter 58, suppress to be picked up by microphone 55 and the echo that causes from the voice of loud speaker 54 outputs.

Second echo eliminator 53 is used under the situation of carrying out " conversion of 2-4 line " suppressing voice that microphone picks up around the echo that produces to the loud speaker side.Wherein, in " 4 line formula communicator ", do not need second echo eliminator 53 and voice switch 51.

The transfer function coefficients of setting in first echo eliminator 52 and second echo eliminator 53 is upgraded in time change (changing in time) according to propagation path one by one.

Moreover, as another existing communicator, possess as everyone knows the communicator that suppresses the echo that comprises in the output of another microphone from different first microphone of the distance of loud speaker and second microphone, according to the output of a microphone effectively in addition.

For example, (Japan) spy opens that disclosed existing communicator detects the output of loud speaker 54 as shown in figure 26 in the flat 8-223275 communique with first microphone 55 and second microphone 56.

In the learning process before using beginning, the echo suppression characteristic of the echo eliminator 57 that is provided with according to the back level of setting first microphone 55 from the transfer function of the propagation path of the voice of loud speaker 54 to first microphones 55 and second microphone 56 makes the output of subtracter 58---the power minimum of the difference of the output of first microphone 55 and the output of second microphone 56.

Therefore, under user mode,, also can from the output of subtracter 58, suppress echo even the output of loud speaker 54 is picked up by first microphone 55 and second microphone 56.

As mentioned above, open the 2000-244670 communique and the spy opens in the flat 8-223275 communique, disclose the echo suppression characteristic that decides echo eliminator according to the propagation path of the voice from the loud speaker to the microphone the spy.

Therefore as can be known, above-mentioned spy is being opened the 2000-244670 communique and special opening that disclosed echo eliminator is applied to have with hinged two housings in the flat 8-223275 communique, in a housing, holding loud speaker, under the situation of communicator that the mobile phone that holding microphone in another housing can change the relative position between loud speaker and the microphone like that, need change the echo suppression characteristic according to the relative position between loud speaker and the microphone.

Yet, open the 2000-244670 communique and the spy opens in the flat 8-223275 communique above-mentioned spy, unexposed, hint changes the echo suppression characteristic according to the relative position between loud speaker and the microphone, if the spy is opened the 2000-244670 communique and special drives the communicator that disclosed echo eliminator in the flat 8-223275 communique is applied to change the relative position between loud speaker and the microphone, then under the situation that has changed the relative position between loud speaker and the microphone, produce following problems: the propagation path of the voice that the echo suppression characteristic of echo eliminator is not suitable for relative position between loud speaker and the microphone after changing, the echo rejection worsens, and speech quality worsens.

The present invention proposes in order to address the above problem, its purpose is to provide a kind of communicator, even under the situation that the relative position between loud speaker and the microphone has been changed, also can keep the echo rejection, carries out high-quality conversation.

Summary of the invention

Communicator of the present invention has following structure, comprising: housing has datum mark; The voice output parts are exported remote signaling as far-end speech; The phonetic entry parts, the input near-end speech; And echo inhibition parts, suppress owing to the said distal ends voice from above-mentioned voice output parts output are imported into the echo that produces the above-mentioned phonetic entry parts; Above-mentioned voice output parts comprise: the remote signaling transformation component, be provided with said reference point predetermined first distance of being separated by, and be the said distal ends voice with said distal ends signal transformation; Above-mentioned phonetic entry parts comprise: the first near-end speech transformation component, be provided with the said reference point predetermined second distance of being separated by, and the above-mentioned near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component, being set at the above-mentioned first near-end speech transformation component is on the preposition of benchmark, and the above-mentioned near-end speech of input is transformed to second near end signal; Above-mentioned echo suppresses parts and comprises: the currency test section, detect the currency of the parameter of the relative position between some in regulation said distal ends signal transformation portion and above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component; The first propagation time calculating part, calculate first propagation time and second propagation time corresponding according to first function and second function with the detected above-mentioned currency of above-mentioned currency test section, this first function and second function with above-mentioned parameter as independent variable, the propagation time that the said distal ends voice that this first function goes out the conversion of said distal ends signal transformation portion propagate into the above-mentioned first near-end speech transformation component, the said distal ends voice that this second function goes out the conversion of said distal ends signal transformation portion propagated into the propagation time of the above-mentioned second near-end speech transformation component as the subordinate variable as the subordinate variable; And first suppress back near end signal efferent, and above-mentioned first propagation time of calculating according to the above-mentioned first propagation time calculating part and above-mentioned second propagation time are used in output and the echo suppression characteristic that changed has suppressed first of echo and suppresses the back near end signal.

According to this structure, even under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can suppress owing to far-end speech is sneaked into the echo that produces in the near end signal.

In addition, communicator of the present invention has following structure, above-mentioned first suppresses back near end signal efferent comprises: the first near end signal delay portion, and above-mentioned first propagation time of calculating according to the above-mentioned first propagation time calculating part postpones above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out; The second near end signal delay portion, above-mentioned second propagation time of calculating according to the above-mentioned first propagation time calculating part postpones above-mentioned second near end signal that the above-mentioned second near-end speech transformation component conversion goes out; And first efferent, that exports that the above-mentioned first near end signal delay portion postponed first postpones the poor of the second delay near end signal that near end signal and the above-mentioned second near end signal delay portion postponed.

According to this structure, the time that propagates into the first near-end speech transformation component and the second near-end speech transformation component by the far-end speech that goes out according to the conversion of remote signaling transformation component postpones near end signal, can suppress echo.

Moreover, communicator of the present invention has following structure, above-mentioned echo suppresses parts and comprises: the first balanced portion will be and the corresponding equalization characteristic of the detected above-mentioned currency of above-mentioned currency test section from the above-mentioned first above-mentioned first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

By this structure, even, can prevent that also speech quality from worsening owing to the variation of the relative position between remote signaling transformation component and the near-end speech transformation component worsens the frequency distribution that suppresses the back near end signal.

Moreover, communicator of the present invention has following structure, above-mentioned echo suppresses parts and comprises: the first near-end person position detection part, cross-correlation function between above-mentioned second near end signal that above-mentioned first near end signal that goes out according to the above-mentioned first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out, detecting with some in above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component is the near-end person's of benchmark position---the first near-end person position; And the second balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned first near-end person position of the above-mentioned first near-end person position detection part from the above-mentioned first above-mentioned first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

By this structure,, can prevent that also speech quality from worsening even be under the situation of near-end person position of benchmark with some in the first near-end speech transformation component and the second near-end speech transformation component at near-end person position deviation.

Moreover, communicator of the present invention has following structure, above-mentioned echo suppresses parts and comprises: the second near-end person position detection part, according to the detected above-mentioned currency of above-mentioned currency test section, detecting with above-mentioned remote signaling transformation component is the above-mentioned near-end person's of benchmark current position---the second near-end person position; And the 3rd balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned second near-end person position of the above-mentioned second near-end person position detection part from the above-mentioned first above-mentioned first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

According to this structure,, can prevent that also speech quality from worsening, and can make the designs simplification of the structure of the second near-end person position detection part than the first near-end person position detection part even not only near-end person position deviation is the near-end person of benchmark with the remote signaling transformation component.

Moreover communicator of the present invention has following structure, comprising: housing has datum mark; The voice output parts are exported remote signaling as far-end speech; The phonetic entry parts, the input near-end speech; And echo inhibition parts, suppress owing to the said distal ends voice from above-mentioned voice output parts output are imported into the echo that produces the above-mentioned phonetic entry parts; Above-mentioned voice output parts comprise: the remote signaling transformation component, be provided with said reference point predetermined first distance of being separated by, and be the said distal ends voice with said distal ends signal transformation; Above-mentioned phonetic entry parts comprise: the first near-end speech transformation component, be provided with the said reference point predetermined second distance of being separated by, and the above-mentioned near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component, being set at the above-mentioned first near-end speech transformation component is on the preposition of benchmark, and the above-mentioned near-end speech of input is transformed to second near end signal; Above-mentioned echo suppresses parts and comprises: the currency test section, detect the currency of the parameter of the relative position between some in regulation said distal ends signal transformation portion and above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component; The second propagation time calculating part, calculate three propagation time and four propagation time corresponding according to the 3rd function and the 4th function with the detected above-mentioned currency of above-mentioned currency test section, the 3rd function and the 4th function with above-mentioned parameter as independent variable, the 3rd propagation time that the 3rd function propagates into the above-mentioned first near-end speech transformation component with above-mentioned near-end speech, the 4th function propagated into the 4th propagation time of the above-mentioned second near-end speech transformation component as the subordinate variable with above-mentioned near-end speech as the subordinate variable; Residual voice signal extraction unit is used above-mentioned the 3rd propagation time of calculating according to the above-mentioned second propagation time calculating part and above-mentioned the 4th propagation time and the residual voice signal extraction property that changed extracts residual voice signal; And second suppress back near end signal efferent, deducts above-mentioned residual voice signal that above-mentioned residual voice signal extraction unit extracts in some above-mentioned second near end signal that above-mentioned first near end signal that goes out from the above-mentioned first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out and export second and suppress the back near end signal.

By this structure, can be suppressed at the echo and the background noise that mix far-end speech in the near-end speech that comprises in the near end signal and produce reliably according to the relative position between remote signaling transformation component and the near-end speech transformation component.

Moreover, communicator of the present invention has following structure, above-mentioned residual voice signal extraction unit comprises: the 3rd near end signal delay portion, and above-mentioned the 3rd propagation time of calculating according to the above-mentioned second propagation time calculating part postpones above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out; Nearby place's end signal delay portion, above-mentioned the 4th propagation time of calculating according to the above-mentioned second propagation time calculating part postpones above-mentioned second near end signal that the above-mentioned second near-end speech transformation component conversion goes out; And residual voice signal efferent, that exports that above-mentioned the 3rd near end signal delay portion postponed the 3rd postpones difference between the 4th delay near end signal that near end signal and above-mentioned nearby place's end signal delay portion postponed as residual voice signal; Above-mentioned second suppresses back near end signal efferent comprises: interference tones signal estimation portion, according to above-mentioned residual voice signal from above-mentioned residual voice signal efferent output, signal---the interference tones signal beyond the signal of estimating to comprise in some in above-mentioned second near end signal that above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out or the above-mentioned second near-end speech transformation component conversion go out corresponding with above-mentioned near-end speech; The 5th near end signal delay portion, the above-mentioned interference tones signal of the some delays estimation portion in above-mentioned second near end signal that above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion is gone out or the above-mentioned second near-end speech transformation component conversion go out estimates the interference tones required time of signal; Second efferent, the above-mentioned near end signal that postponed from above-mentioned the 5th near end signal delay portion---the 5th postpones the near end signal, deducts above-mentioned interference tones signal that above-mentioned interference tones signal estimation portion estimates and exports second and suppress the back near end signal; And renewal portion, the interference tones signal evaluation characteristic of above-mentioned interference tones signal estimation portion is updated to one by one makes the above-mentioned second Optimal Jamming tone signal evaluation characteristic that suppresses square time average minimum of back near end signal.

According to this structure, by near end signal, extracting residual voice signal, from near end signal, deduct the residual voice signal that extracts, even under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can suppress echo and background noise in the near end signal.

Moreover communicator of the present invention has following structure, and above-mentioned echo suppresses parts and comprises: the initial characteristic configuration part, according to the detected above-mentioned currency of above-mentioned currency test section, set the initial interference tone signal evaluation characteristic of above-mentioned renewal portion.

According to this structure, can make the interference tones signal evaluation characteristic of interference tones signal estimation portion promptly converge to the Optimal Jamming tone signal evaluation characteristic that makes square time average minimum that suppresses the back near end signal, suppress echo and background noise more reliably.

Moreover, communicator of the present invention has following structure, above-mentioned echo suppresses parts and comprises: the first near-end person position detection part, cross-correlation function between above-mentioned second near end signal that above-mentioned first near end signal that goes out according to the above-mentioned first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out, detecting with some in above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component is the above-mentioned near-end person's of benchmark position---the first near-end person position; And the second balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned first near-end person position of the above-mentioned first near-end person position detection part from the above-mentioned second above-mentioned second frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

According to this structure, even near-end person position deviation first reference position can prevent that also speech quality from worsening.

Moreover, communicator of the present invention has following structure, above-mentioned echo suppresses parts and comprises: the second near-end person position detection part, according to the detected above-mentioned currency of above-mentioned currency test section, detecting with above-mentioned remote signaling transformation component is the above-mentioned near-end person's of benchmark position---the second near-end person position; And the 3rd balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned second near-end person position of the above-mentioned second near-end person position detection part from the above-mentioned second above-mentioned second frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

According to this structure, even near-end person position deviation second reference position not only, frequency distribution that also can balanced near-end speech is kept speech quality, and can be made the designs simplification of the structure of the second near-end person position detection part than the first near-end person position detection part.

Moreover communicator of the present invention has following structure, comprising: housing has datum mark; The voice output parts are exported remote signaling as far-end speech; The phonetic entry parts, the input near-end speech; And echo inhibition parts, suppress owing to the said distal ends voice from above-mentioned voice output parts output are imported into the echo that produces the above-mentioned phonetic entry parts; Above-mentioned voice output parts comprise: the remote signaling transformation component, be provided with said reference point predetermined first distance of being separated by, and be the said distal ends voice with said distal ends signal transformation; Above-mentioned phonetic entry parts comprise: the first near-end speech transformation component, be provided with the said reference point predetermined second distance of being separated by, and the above-mentioned near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component, being set at the above-mentioned first near-end speech transformation component is on the preposition of benchmark, and the above-mentioned near-end speech of input is transformed to second near end signal; Above-mentioned echo suppresses parts and comprises: the echo component test section, above-mentioned second near end signal that above-mentioned first near end signal that goes out according to the first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out detects signal---the echo component signal beyond the signal corresponding with real near-end speech that comprises jointly in above-mentioned first near end signal and above-mentioned second near end signal; And the back near end signal efferent of the 3rd inhibition, in some above-mentioned second near end signal that above-mentioned first near end signal that goes out from the first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out, deduct the detected above-mentioned echo component signal of above-mentioned echo component test section and export the 3rd and suppress the back near end signal.

Here so-called echo component signal is meant the signal beyond the signal corresponding with near-end speech, is not limited to the echo in the near end signal, and comprises and the corresponding signal of all noises such as background noise.

According to this structure, by near end signal, deducting the echo component signal, even under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can suppress owing to far-end speech is sneaked into echo and background noise in the near end signal that near-end speech produces, the communicator that can realize carrying out high-quality conversation.

Moreover, communicator of the present invention has following structure, above-mentioned echo component test section comprises: real near-end speech estimation portion, estimate the corresponding real near end signal of real near-end speech that comprises jointly in above-mentioned second near end signal that above-mentioned first near end signal that goes out with the first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out; The 6th near end signal delay portion, the above-mentioned real near-end speech estimation of the some delays in above-mentioned second near end signal that above-mentioned first near end signal that the first near-end speech transformation component conversion is gone out or the above-mentioned second near-end speech transformation component conversion go out portion estimates the above-mentioned required time of real near end signal; And echo component signal efferent, the above-mentioned near end signal that above-mentioned the 6th near end signal delay portion was postponed---the 6th postpones the difference of the above-mentioned real near end signal that near end signal and above-mentioned real near-end speech estimation portion estimate as the signal beyond the above-mentioned real near end signal---first echo component signal output; The above-mentioned the 3rd suppresses back near end signal efferent comprises: echo component estimation portion, according to above-mentioned echo component signal from the output of above-mentioned echo component signal efferent, signal---the second echo component signal beyond the signal of estimating to comprise in some in above-mentioned second near end signal that above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out or the above-mentioned second near-end speech transformation component conversion go out corresponding with above-mentioned near-end speech; The 7th near end signal delay portion, the above-mentioned echo component estimation of the some delays portion in above-mentioned second near end signal that above-mentioned first near end signal that the first near-end speech transformation component conversion is gone out or the above-mentioned second near-end speech transformation component conversion go out estimates the required time of signal component of above-mentioned echo; And the 3rd efferent, the above-mentioned near end signal that postponed from above-mentioned the 7th near end signal delay portion---the 7th postpones to deduct the near end signal the above-mentioned second echo component signal that above-mentioned echo component estimation portion estimates exports the 3rd and suppresses the back near end signal.

According to this structure, by extracting the signal corresponding signal---echo component signal in addition with above-mentioned near-end speech, from near end signal, deduct the echo component signal that extracts, can realize a kind of communicator, even under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can suppress echo and background noise in the near end signal, can carry out high-quality conversation.

Moreover, communicator of the present invention has following structure, comprise: first sef-adapting filter, some adaptive-filterings that carries out in above-mentioned second near end signal that above-mentioned first near end signal that the first near-end speech transformation component conversion is gone out and the above-mentioned second near-end speech transformation component conversion go out are handled, and make from the absolute value minimum of the signal of above-mentioned echo component signal efferent output; Above-mentioned echo component estimation portion has: second sef-adapting filter, and the above-mentioned echo component signal from above-mentioned echo component signal efferent output is carried out adaptive-filtering handle, make from the absolute value minimum of the signal of above-mentioned the 3rd efferent output; Above-mentioned echo suppresses parts and comprises: the currency test section, detect the currency of the parameter of the relative position between some in regulation said distal ends signal transformation portion and above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component; The first initial value determination section, according to the currency of the detected parameter of above-mentioned currency test section, parameter---the initial value of first filter coefficient that the adaptive-filtering that decides above-mentioned first sef-adapting filter to carry out is handled; And the second initial value determination section, parameter---the initial value of second filter coefficient that the adaptive-filtering that decides above-mentioned second sef-adapting filter to carry out according to the currency of the detected parameter of above-mentioned currency test section is handled.

According to this structure, the first initial value determination section and the second initial value determination section are according to the currency of the detected parameter of currency test section, decide the initial value of first filter coefficient and second filter coefficient, so can realize a kind of communicator, even under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can suppress at short notice can carry out high-quality conversation owing to far-end speech is sneaked into echo and background noise in the near end signal that produces in the near end signal.

Moreover, communicator of the present invention has following structure, above-mentioned echo suppresses parts and comprises: adaptive control portion, when the predetermined near end signal in above-mentioned first near end signal or above-mentioned second near end signal is more remarkable than said distal ends signal, indicate above-mentioned first sef-adapting filter to upgrade above-mentioned first filter coefficient; When the said distal ends signal is more remarkable than the predetermined near end signal in above-mentioned first near end signal or above-mentioned second near end signal, indicate above-mentioned second sef-adapting filter to upgrade above-mentioned second filter coefficient.

According to this structure, adaptive control portion is according to remote signaling and near end signal, indicate first sef-adapting filter and second sef-adapting filter to upgrade first filter coefficient and second filter coefficient, so can realize a kind of communicator, can follow the tracks of the variation of communication environment well, even under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can suppress at short notice can carry out high-quality conversation owing to far-end speech is sneaked into echo and background noise in the near end signal that produces in the near end signal.

Description of drawings

By the description that carry out in conjunction with following accompanying drawing the back, the feature of communicator of the present invention and advantage will become clearly.

Fig. 1 is the functional structure chart of the communicator of first embodiment of the invention.

Fig. 2 is the hardware structure diagram of the communicator of first embodiment of the invention.

Fig. 3 is applied to structure chart under the situation of mobile phone with communicator of the present invention.

Fig. 4 is the flow chart of work of the communicator of explanation first embodiment of the invention.

Fig. 5 is the functional structure chart of communicator that comprises first execution mode of the first balanced portion.

Fig. 6 is the flow chart of work of the communicator of explanation first execution mode that comprises the first balanced portion.

Fig. 7 is the flow chart of work of the first balanced portion of the communicator of explanation first execution mode.

Fig. 8 is the curve chart of equalization characteristic of the first balanced portion of the communicator of first execution mode.

Fig. 9 is the functional structure chart of communicator that comprises first execution mode of the second balanced portion.

Figure 10 is the flow chart of work of the communicator of explanation first execution mode that comprises the second balanced portion.

Figure 11 is the flow chart of work of the second balanced portion of the communicator of explanation first execution mode.

Figure 12 is the functional structure chart of communicator of first execution mode of the 3rd balanced portion that comprises of the communicator of first execution mode.

Figure 13 (a)～Figure 13 (c) is the curve chart of equalization characteristic of the 3rd balanced portion of the communicator of first execution mode.

Figure 14 is the functional structure chart of the communicator of second embodiment of the invention.

Figure 15 is the flow chart of work of the communicator of explanation second embodiment of the invention.

Figure 16 is the key diagram that the echo of the communicator of explanation second embodiment of the invention suppresses the work of parts.

Figure 17 is the functional structure chart of communicator that comprises second execution mode of initial characteristic configuration part.

Figure 18 is the flow chart of work of the communicator of explanation second execution mode that comprises the initial characteristic configuration part.

Figure 19 is the functional structure chart of communicator that comprises second execution mode of the second balanced portion.

Figure 20 is the flow chart of work of the communicator of explanation second execution mode that comprises the second balanced portion.

Figure 21 is the functional structure chart of communicator that comprises second execution mode of the 3rd balanced portion.

Figure 22 is the block diagram of the communicator of third embodiment of the invention.

Figure 23 is the flow chart of work of the communicator of explanation third embodiment of the invention.

The figure of Figure 24 (a)～Figure 24 (f) effect that to be explanation obtain with the communicator of third embodiment of the invention.

Figure 25 is the structure chart of existing communicator.

Figure 26 is the structure chart of existing communicator.

Embodiment

The following execution mode that communicator of the present invention is described with reference to accompanying drawing.

(first execution mode)

Wherein, in this manual, " datum mark " is the point as benchmark of the relative position between some in regulation remote signaling transformation component and the first near-end speech transformation component and the second near-end speech transformation component, under the situation of the mobile phone that applies the present invention to form, be datum mark in conjunction with the hinge of two partial shell by the housing that two partial shell constitute.

In addition, in this manual, " reference position " is to be the near-end person's of benchmark position with some in remote signaling transformation component or the first near-end speech transformation component and the second near-end speech transformation component, imagines common user mode and determines in the design phase.With some in the first near-end speech transformation component and the second near-end speech transformation component is that the reference position of benchmark is first reference position, is that the reference position of benchmark is second reference position with the remote signaling transformation component.

At first, with reference to Fig. 1, the structure of the communicator of first embodiment of the invention is described.

The communicator of first embodiment of the invention comprises as shown in Figure 1: voice output parts 11, and the output remote signaling is as far-end speech; Phonetic entry parts 12, the input near-end speech; And echo inhibition parts 13, suppress owing to the far-end speech from 11 outputs of voice output parts is imported into the echo that produces the phonetic entry parts 12.

Voice output parts 11 comprise: remote signaling transformation component 112, be provided with datum mark predetermined first distance of being separated by, and remote signaling is transformed to far-end speech.

Phonetic entry parts 12 comprise: the first near-end speech transformation component 121, be provided with the datum mark predetermined second distance of being separated by, and the near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component 122, being set at the first near-end speech transformation component 121 is on the preposition of benchmark, and the near-end speech of input is transformed to second near end signal.

Echo suppresses parts 13 and comprises: currency test section 131, detect the currency of the parameter of the relative position between some in regulation remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122; The first propagation time calculating part 132, calculate first propagation time and second propagation time corresponding according to first function and second function with currency test section 131 detected currencys, this first function and second function will stipulate that the parameter of the relative position between some in remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122 is as independent variable, the far-end speech that this first function goes out 112 conversion of remote signaling transformation component propagates into the propagation time (being called " first propagation time " later on) of the first near-end speech transformation component 121 as the subordinate variable, and the far-end speech that this second function goes out 112 conversion of remote signaling transformation component propagates into the propagation time (being called " second propagation time " later on) of the second near-end speech transformation component 122 as the subordinate variable; And first suppress back near end signal efferent 130, uses first propagation time of calculating according to the first propagation time calculating part 132 and second propagation time and the echo suppression characteristic that changed suppresses echo, first near end signal after the output echo suppresses.

First suppresses back near end signal efferent 130 comprises: the first near end signal delay portion 133, and first propagation time of calculating according to the first propagation time calculating part 132 postpones first near end signal that 121 conversion of the first near-end speech transformation component go out; The second near end signal delay portion 134, second propagation time of calculating according to the first propagation time calculating part 132 postpones second near end signal that 122 conversion of the second near-end speech transformation component go out; And first efferent 135, first difference signal that postpones between the second delay near end signal that near end signal and the second near end signal delay portion 134 postponed that the first near end signal delay portion 133 that exports postponed.

Wherein, Fig. 1 shows the functional structure chart of the structure that comprises the first near end signal delay portion 133 and the second near end signal delay portion 134 in the communicator of first embodiment of the invention, ignoring under the situation in propagation time of the voice of remote signaling transformation component 112 to the first near-end speech transformation components 121 or the second near-end speech transformation component 122, also can omit the corresponding first near end signal delay portion 133 or the second near end signal delay portion 134.

Therefore, in the communicator of first embodiment of the invention, for example omitted under the situation of the second near end signal delay portion 134 second near end signal that first delay near end signal that postponed from first efferent 135 output, the first near end signal delay portion 133 and 122 conversion of the second near-end speech transformation component go out poor.

Wherein, communicator in first embodiment of the invention comprises under the situation of the first near end signal delay portion 133 and the second near end signal delay portion 134, the second delay near end signal that the first delay near end signal that postponed from first efferent 135 output, the first near end signal delay portion 133 and the second near end signal delay portion 134 postponed poor.

Then, with reference to Fig. 2, the hardware configuration of first execution mode of communicator of the present invention is described.

The remote signaling transformation component 112 of voice output parts 11 is made of remote signaling amplifier 211 and loud speaker 212.In addition, the first near-end speech transformation component 121 of phonetic entry parts 12 is made of first microphone 221 and the first near end signal amplifier 223, and the second near-end speech transformation component 122 is made of second microphone 222 and the second near end signal amplifier 224.

Moreover, echo suppresses the currency test section 131 of parts 13 so long as can detect and be transformed to the detector of the signal of telecommunication to the parameter of the relative position between some in regulation remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122 and get final product, and can use rotary encoder, potentiometer etc.In addition, needn't be continuously the parameter of the relative position between some in regulation remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122 be detected, also can use the switch of on off state counter-rotating when arriving a specific relative position at least.

And echo suppresses currency test section 131 work in addition of parts 13 by being realized by program stored in CPU 232 execute stores 231.

Therefore, echo suppresses parts 13 and comprises: analog-to-digital converter (hereinafter to be referred as the A/D converter) 233 will be a digital value from the near end signal and the 131 detected parameter transformations of currency test section of the first near end signal amplifier 223 and 224 outputs of the second near end signal amplifier; And digital-to-analog converter (hereinafter to be referred as the D/A converter) 234, echo is suppressed the back near end signal be transformed to analog signal.Memory 231, CPU 232, A/D converter 233, and D/A converter 234 by bus 235 mutual electric coupling.Wherein, under the situation that currency test section 131 is made of switch, also digital input interface can be set, read in currency through digital input interface.

Wherein, in the execution mode afterwards, hardware configuration is identical, so omit the explanation of the hardware configuration in each execution mode.

Then, with reference to Fig. 3, the concrete example that communicator of the present invention is applied to the following function under the situation of the mobile phone be made up of the housing that two partial shell constitute is described, this function will stipulate that parameter that echo suppresses remote signaling transformation component 112 that the first propagation time calculating parts 132 in the parts 13 use and the relative position between the first near-end speech transformation component 121 or the second near-end speech transformation component 122 as independent variable, will propagate into the propagation time of the first near-end speech transformation component 121 and the second near-end speech transformation component 122 respectively as the subordinate variable from the voice of at least one remote signaling transformation component 112 output.

The mobile phone 30 that the housing that is made of two partial shell is formed has datum mark---the first's housing 32 and the second portion housing 33 of hinge 31 combinations with substantial middle, in first's housing 33, be provided with the loud speaker 212 that comprises in the remote signaling transformation component 112, in second portion housing 33, holding second microphone 222 that comprises in first microphone 221 that comprises in the first near-end speech transformation component 121 and the second near-end speech transformation component 122.

Between the hinge 31 and first microphone 221 and second microphone 222 apart from a and (a+x), the distance b between hinge 31 and the loud speaker 212 is constant.

If establishing second portion housing 33 is θ (with reference to Fig. 3) with first's housing 32 angulations, then angle θ is the parameter of the relative position between some in regulation loud speaker 212 and first microphone 221 and second microphone 222.

Voice propagate from the distance of loud speaker 212 to first microphones 221 required time T 1, and voice propagate from the distance of loud speaker 212 to second microphones 222 required time T 2 respectively as (1) formula reaching shown in (2) formula, be represented as parameter---the function of angle θ of the relative position between some in regulation loud speaker 212 and microphone 221 and the microphone 222.

[formula 1]

${\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="A20048000109400491.png,A20048000109400531.png"\; state="\{\{state\}\}">T</figure-callout>}}_1=\sqrt{a^2+{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="A20048000109400531.png"\; state="\{\{state\}\}">b</figure-callout>}}^2+2\mathrm{ab}\&CenterDot;\cos \mathrm{\&theta;}}/v---\left(1\right)$

[formula 2]

${\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="A20048000109400531.png"\; state="\{\{state\}\}">T</figure-callout>}}_2=\sqrt{a^2+{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="A20048000109400531.png"\; state="\{\{state\}\}">b</figure-callout>}}^2+2\mathrm{ab}\&CenterDot;\cos \mathrm{\&theta;}+2x(b\&CenterDot;\cos \mathrm{\&theta;}+a)+x^2}/v---\left(2\right)$

Wherein, v is the velocity of sound.

Then, with reference to the flow chart of Fig. 4, illustrate that the echo of the communicator of first embodiment of the invention suppresses the work of parts 13.

At first, whether end of conversation conditions such as judgement hook state set up (step S401), and when being judged to be the establishment of end of conversation condition, echo suppresses parts 13 power cut-offs.On the contrary, be judged to be the end of conversation condition when being false, thinking in the conversation, according to the level of near end signal whether predetermined level with first-class judge begin conversation (step S402) whether.

Then, do not begin, then turn back to the judgement (step S401) whether the end of conversation condition is set up if in step S402, be judged to be conversation.On the contrary, in step S402, be judged to be when beginning to converse, read in the currency (step S403) of the parameter of the relative position between some regulation currency test section 131 detected remote signaling transformation components 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122 through A/D converter 233 from currency test section 131.

Then, the currency that substitution is read in through A/D converter 233 from currency test section 131 in the independent variable of first function and second function calculates first propagation time corresponding with currency and second propagation time (step S404); This first function and second function will stipulate that the parameter of the relative position between some in remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122 is as independent variable, the far-end speech that this first function goes out 112 conversion of remote signaling transformation component propagates into first propagation time of the first near-end speech transformation component 121 as the subordinate variable, and the far-end speech that this second function goes out 112 conversion of remote signaling transformation component propagates into second propagation time of the second near-end speech transformation component 122 as the subordinate variable.Wherein, this processing is equivalent to the work of the first propagation time calculating part 132.

Then, read near end signal (step S405) from the first near-end speech transformation component 223 and the second near-end speech transformation component 224 through A/D converter 233, with near end signal delay transit time (step S406).

As the method that postpones near end signal, the method that makes the integral multiple that is similar to the sampling time time of delay, postpones near end signal with shift register is arranged.In addition, in order more critically to postpone near end signal, constitute delay portion with FIR (finite impulse response) filter and get final product.

Wherein, this processing is equivalent to the work of the first near end signal delay portion 133 and the second near end signal delay portion 134.

Then, first postponed the second delay near end signal that near end signal and the second near end signal delay portion 134 postponed and subtracted each other, suppressed echo (step S407) by what the first near end signal delay portion 133 was postponed.This processing is equivalent to the work of first efferent 135.

In addition, the processing from step S406 to step S408 is equivalent to the work of the first inhibition back near end signal efferent 130.

Then, the back near end signal be will suppress with D/A converter 234 and analog signal and output (step S408) will be transformed to.

At last, whether the level according near end signal judges with inferior whether conversation finishes (step S409) at predetermined level, more than predetermined level, when conversation does not finish, turn back to read in (the step S405) of near end signal at the level that is judged to be near end signal.On the contrary,, below predetermined level, when conversation has finished, turn back at the level that is judged to be near end signal according to whether being the judgement (step S401) that hook state waits the end of conversation condition of carrying out whether to set up.

Wherein, in above-mentioned, first propagation time that the first propagation time calculating part 132 is calculated is set at the time of delay of the first near end signal delay portion 133, and second propagation time that the first propagation time calculating part 132 is calculated is set at the time of delay of the second near end signal delay portion 134.

But, in fact, the propagation time of the voice of (for example first near-end speech transformation component 121) that preferably will be from remote signaling transformation component 112 to the first near-end speech transformation components 121 or the second near-end speech transformation component 122 is set at fixed delay time, will be from remote signaling transformation component 112 to another near-end speech transformation component the propagation time of the voice of (for example second near-end speech transformation component 122) and the difference of fixed delay time be set at the variable delay time.

This be because, not only under the situation that can imagine remote signaling transformation component 112 in the conversation and one relative position in the first near-end speech transformation component 121 or the second near-end speech transformation component 122 in advance, can be made as the predetermined set time time of delay, and the position relation between voice output parts 11 and the phonetic entry parts 12 tackled has easily been changed, the variable delay time is negative situation.

As mentioned above, the communicator of first execution mode of the present invention suppresses to be provided with currency test section 131, the first propagation time calculating part 132, the first near end signal delay portion 133 and the second near end signal delay portion 134 and first efferent 135 in the parts 13 at echo, even so under the situation that relative position between some in current remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122 has been changed, also can suppress echo reliably.

In addition, the communicator of first embodiment of the invention also can be as shown in Figure 5, echo suppresses parts 13 and comprises in the first back level that suppresses back near end signal efferent 130: the first balanced portion 136 will be and the corresponding equalization characteristic of currency test section 131 detected currencys from the first first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent 130 outputs.

Fig. 6 is the flow chart that echo that explanation comprises the first balanced portion 136 suppresses the work of parts 13, the flow chart of Fig. 4 has been added the work (step S410) of the first balanced portion 136 of the balanced frequency distribution that suppresses the back near end signal.

Wherein, in the case, the work of the first balanced portion 136 realizes by program stored in the execute store 231.

Therefore, the processing from step S401 to step S407, and the later processing of step S408 identical with Fig. 4, so omit its explanation.

Here, with reference to the flow chart of Fig. 7, the work of the first balanced portion 136 is described.

At first, behind the echo (step S407) that has suppressed to comprise in the near end signal, use the equalization characteristic that is stored in advance in the memory 231 to calculate the equalization characteristic (step S411) corresponding, the frequency distribution of balanced near end signal (step S412) with currency test section 131 detected currencys.More than handle work corresponding to the first balanced portion 136.

Then, with reference to Fig. 8, the work that communicator of the present invention is applied to have the first balanced portion 136 under the situation of mobile phone of the housing that is made of two partial shell is described.

In Fig. 8, the level of near end signal is with respect to the variation of frequency after the inhibition that the person that shows the near-end is positioned at remote signaling transformation component 112 when being second reference position of benchmark.Wherein, solid line, chain-dotted line, and dotted line represent that respectively first's housing 32 and second portion housing 33 angulation θ are respectively the situations of 0 degree, 30 degree and 60 degree, the data in advance of this curve chart is stored in the memory 231.

For example, if it is 30 degree with second portion housing 33 angulation θ that currency test section 131 detects first's housing 32, then the equalization characteristic shown in the choice point line uses this equalization characteristic to come the balanced frequency distribution that suppresses the back near end signal from first of first efferent, 135 outputs.

Wherein, for example be 45 degree, be not to be stored in advance under the situation of the angle in the memory, according to calculating equalization characteristic with waiting with interpolation method near the corresponding equalization characteristic of the angle of storing of this angle at currency test section 131 detected angle θ.

As mentioned above, above-mentioned communicator is provided with the first balanced portion 136 that equalization characteristic changes according to currency test section 131 detected currencys in the first back level that suppresses back near end signal efferent 130, even, can prevent that also speech quality from worsening so the relative position of voice output parts 11 and phonetic entry parts 12 changes and the frequency distribution that suppresses the back near end signal is worsened.

Moreover, the communicator of first embodiment of the invention also can be as shown in Figure 9, echo suppresses parts 13 and comprises: the first near-end person position detection part 137, according to the person's that detects the near-end from first near end signal of the first near-end speech transformation component 121 and 122 outputs of the second near-end speech transformation component and the cross-correlation function between second near end signal position; And the second balanced portion 138, will be with currency test section 131 detected currencys, reach the corresponding equalization characteristic in the first near-end person position detection part, the 137 detected first near-end person positions from first first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent 130 outputs.

Figure 10 is that explanation comprises that the echo of the second balanced portion 138 suppresses the flow chart of the work of parts 13, the flow chart of Fig. 4 has been added second equilibrium treatment (step S420) of the frequency distribution of balanced inhibition back near end signal.Wherein, the processing from step S401 to step S407, and the later processing of step S408 identical with Fig. 4, so omit its explanation.

Then, with reference to Figure 11, second equilibrium treatment is described.

At first, between the dead space that detection far-end person does not speak.Whether this detection generally is to judge with following at fixed threshold according to the level of remote signaling, but under the situation of noise that superposeed on the remote signaling, threshold value also can be used the mean value of the remote signaling of official hour.

After detecting between the dead space, judge whether the near-end person speaks.This judgement can according to the level of near end signal whether the mean value of the official hour of fixed threshold or near end signal with on carry out, but under background noise environment, according to the maximum of whitening cross-correlation function whether fiducial value (0.5～0.7) with on judge very favourable.

Here, the so-called white cross-correlation function is the amassing normalization and carry out the function of inverse Fourier transform of power of the power of first near end signal that goes out with 121 conversion of the first near-end speech transformation component of the cross-spectrum between second near end signal that goes out of first near end signal that 121 conversion of the first near-end speech transformation component are gone out and 122 conversion of the second near-end speech transformation component and second near end signal that 122 conversion of the second near-end speech transformation component go out.

Then, if the near-end person speaks in being judged to be between the dead space, the whitening cross-correlation function between second near end signal that reads in according to first near end signal that has read in from the first near-end speech transformation component 121 with from the second near-end speech transformation component 122 then, calculating with some in the first near-end speech transformation component 121 or the second near-end speech transformation component 122 is the near-end person's of benchmark position---the first near-end person position (step S421).This processing is equivalent to the work of the first near-end person position detection part 137.

Then, use the equalization characteristic that is stored in advance in the memory 231 to calculate and the first near-end person position detection part, 137 detected first near-end person positions and the corresponding equalization characteristic (step S422) of currency test section 131 detected currencys, come the balanced frequency distribution (step S423) that suppresses the back near end signal with this equalization characteristic.More than handle the work of the second balanced portion 138 that is equivalent to.

As mentioned above, above-mentioned communicator suppresses to be provided with the first near-end person position detection part, the 137 and second balanced portion 138 in the parts 13 at echo, even so under the situation of the first near-end person position deviation, first reference position, also can prevent that speech quality from worsening by using and departing from the balanced frequency distribution that suppresses the back near end signal of corresponding equalization characteristic.

Moreover, the communicator of first embodiment of the invention also can be as shown in figure 12, comprise: the second near-end person position detection part 139, according to currency test section 131 detected currencys, detecting with remote signaling transformation component 112 is the near-end person's of benchmark current position---the second near-end person position; And the 3rd balanced portion 140, will be and the corresponding equalization characteristic of currency test section 131 detected currencys and the second near-end person position detection part, the 139 detected second near-end person positions from the first first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent 130 outputs.

Therefore, in above-mentioned communicator, the person that supposes the near-end be positioned at remote signaling transformation component 112 over against be second reference position of benchmark with remote signaling transformation component 112, calculate equalization characteristic.

Under the situation of the mobile phone that above-mentioned communicator is applied to have the housing that is made of two partial shell, the near-end person is positioned at the position with second portion housing 33 angled α (with reference to Fig. 3 and Figure 12).But, since supposed the near-end person be positioned at remote signaling transformation component 112 over against be second reference position of benchmark with remote signaling transformation component 112, be first's housing 32 and the function of second portion housing 33 angulation θ so angle [alpha] and near-end speech propagate into the propagation time of the first near-end speech transformation component 121 and the second near-end speech transformation component 122.Therefore, can calculate the propagation time that angle [alpha] and near-end speech propagate into the first near-end speech transformation component 121 and the second near-end speech transformation component 122 according to currency test section 131 detected currencys.

Here, with reference to Figure 13, the work that above-mentioned communicator is applied to have the 3rd balanced portion 140 under the situation of mobile phone of the housing that is made of two partial shell is described.

The (a) and (b) of Figure 13 and (c) show the frequency distribution of near end signal after the inhibition under the situations that first's housing 32 and second portion housing 33 angulation θ are respectively 0 degree, 30 degree and 60 degree.The data in advance corresponding with these three curve charts is stored in the memory 231.

Wherein, the chain-dotted line of each curve chart, dotted line and solid line person that represents the near-end and second portion housing 33 angulation α are respectively the frequency distribution of near end signal after the inhibition under the situations of 90 degree, 120 degree and 150 degree.

If currency test section 131 detected currencys---first's housing 32 is 30 degree with second portion housing 33 angulation θ, the second near-end person position detection part, 139 detected near-end person positions and second portion housing 33 angulation α are 120 degree, then the dotted line with curve chart (B) is chosen as equalization characteristic, uses this equalization characteristic to come the balanced frequency distribution that suppresses the back near end signal.Wherein, in the present embodiment, also can under not be stored in situation in the memory 231 in advance, equalization characteristic use the equalization characteristic of storing to use interpolation method to wait and calculate equalization characteristic.

As mentioned above, above-mentioned communicator suppresses to be provided with the second near-end person position detection part the 139 and the 3rd balanced portion 140 in the parts 13 at echo, even so near-end person position deviation second reference position not only, can prevent that also speech quality from worsening, and can simplify the structure of the second near-end person position detection part 139.

(second execution mode)

Then, with reference to Figure 14, the structure of the communicator of second embodiment of the invention is described.

The communicator of second embodiment of the invention comprises: voice output parts 11, and the output remote signaling is as far-end speech; Phonetic entry parts 12, the input near-end speech; And echo inhibition parts 13, suppress owing to the far-end speech from 11 outputs of voice output parts is imported into the echo that produces the phonetic entry parts 12.

Voice output parts 11 comprise: remote signaling transformation component 112, be provided with datum mark predetermined first distance of being separated by, and remote signaling is transformed to far-end speech.

Phonetic entry parts 12 comprise: the first near-end speech transformation component 121, be provided with the datum mark predetermined second distance of being separated by, and the near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component 122, being set at the first near-end speech transformation component 121 is on the preposition of benchmark, and the near-end speech of input is transformed to second near end signal.

Echo suppresses parts 13 and comprises: currency test section 131, detect the currency of the parameter of the relative position between some in regulation remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122; The second propagation time calculating part 141, calculate three propagation time and four propagation time corresponding according to the 3rd function and the 4th function with currency test section 131 detected currencys, the 3rd function and the 4th function with parameter as independent variable, the 3rd propagation time that the 3rd function propagates into the first near-end speech transformation component 121 with near-end speech, the 4th function propagated into the 4th propagation time of the second near-end speech transformation component 122 as the subordinate variable with near-end speech as the subordinate variable; Residual voice signal extraction unit 150 is used the 3rd propagation time of calculating according to the second propagation time calculating part 141 and the 4th propagation time and the residual voice signal extraction property that changed extracts residual voice signal; And second suppress back near end signal efferent 151, deducts residual voice signal that residual voice signal extraction unit 150 extracts in some second near end signal that first near end signal that goes out from 121 conversion of the first near-end speech transformation component and 122 conversion of the second near-end speech transformation component go out and export and suppressed second of residual voice and suppress the back near end signal.

Residual voice signal extraction unit 150 comprises: the 3rd near end signal delay portion 142, and the 3rd propagation time of calculating according to the second propagation time calculating part 141 postpones first near end signal that 121 conversion of the first near-end speech transformation component go out; Nearby place's end signal delay portion 143, the 4th propagation time of calculating according to the second propagation time calculating part 141 postpones second near end signal that 122 conversion of the second near-end speech transformation component go out; And residual voice signal efferent 144, the difference between the 4th delay near end signal that the 3rd delay near end signal that the 3rd near end signal delay portion 142 that exports postponed and nearby place's end signal delay portion 143 postponed is as residual voice signal.

Moreover, second suppresses back near end signal efferent 151 comprises: interference tones signal estimation portion 145, according to residual voice signal from 144 outputs of residual voice signal efferent, signal---the interference tones signal beyond the signal of estimating to comprise in some in second near end signal that first near end signal that 121 conversion of the first near-end speech transformation component go out or 122 conversion of the second near-end speech transformation component go out corresponding with near-end speech; The 5th near end signal delay portion 146, the some delay interference tones signal estimations portion 145 in second near end signal that first near end signal that 121 conversion of the first near-end speech transformation component are gone out or 122 conversion of the second near-end speech transformation component go out estimates the interference tones required times of signal; Second efferent 147, the near end signal that postponed from the 5th near end signal delay portion 146---the 5th postpones the near end signal, deducts interference tones signal that interference tones signal estimation portion 145 estimates and exports second and suppress the back near end signal; And renewal portion 148, the interference tones signal evaluation characteristic of interference tones signal estimation portion 145 is updated to one by one makes the second Optimal Jamming tone signal evaluation characteristic that suppresses square time average minimum of back near end signal.

Then, with reference to the flow chart of Figure 15, illustrate that the echo of the communicator of second embodiment of the invention suppresses the work of parts 13; The processing of step S401 to 403 is identical with first execution mode, so omit its explanation.

According to the currency that reads in, the person's that calculates the near-end voice propagate into the first near-end speech transformation component 121 and required propagation time---second propagation time (step S430) of the second near-end speech transformation component 122 in step S403.This be because, the person is positioned at position---second reference position of the predetermined predetermined distance of being separated by in the front of remote signaling transformation component 112 if suppose the near-end, and the propagation time that then near-end speech can be propagated into the first near-end speech transformation component 121 and the second near-end speech transformation component 122 is expressed as the function of the some angulation θ in voice output parts 11 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122.This processing is equivalent to the work of the second propagation time calculating part 141.

Wherein, the whitening cross-correlation function that the propagation time that near-end speech propagates into the first near-end speech transformation component 121 and the second near-end speech transformation component 122 also can illustrate according to the work as the first near-end person position detection part 137 of first execution mode calculates, but after the near-end person loquiturs, underestimating of propagation time is so speech quality might worsen.

On the other hand, the person is positioned at second reference position and after calculating the propagation time, departs under the situation of second reference position speech quality deterioration the near-end person to suppose the near-end.

Therefore, also can be when the near-end person speak, the person that supposes the near-end is positioned at second reference position and calculates the propagation time; In other cases, calculate the propagation time according to the whitening cross-correlation function.

Then, read near end signal (step S405) through A/D converter 233, suppress echo and background noise (step S440) from the first near end signal amplifier 223 and the second near end signal amplifier 224.

Wherein, the later processing of step S408 is identical with the work of the echo suppressing portion of first execution mode, so omit its explanation.

Then, with reference to Figure 16, the processing of the step S440 of second execution mode is described.

At first, propagate into the first near-end speech transformation component 121 and required propagation time of the second near-end speech transformation component 122, postpone near end signal (step S441) according to near-end person's voice.Wherein, this processing is equivalent to the work of the 3rd near end signal delay portion 142 and nearby place's end signal delay portion 143.

Then, by getting the near end signal that the 3rd near end signal delay portion 142 and nearby place's end signal delay portion 143 postponed---first postpones poor between near end signal, come the near-end person's that comprises in the cancellation near end signal voice, export residual voice signal (step S442).Wherein, this processing is equivalent to the work of residual voice signal efferent 144.

Then, use known sef-adapting filter, according to this residual voice signal, estimate with first near end signal or second near end signal in the corresponding signal of interference tones---the interference tones signal (step S443) that mixes of the far-end speech that comprises and background noise.Wherein, this processing is equivalent to the work of interference tones signal estimation portion 145.

And then, with one in first near end signal and second near end signal, for example first near end signal that goes out of the first near-end speech transformation component, 121 conversion postpones to estimate the interference tones required suitable times (step S444) in processing time of signal with interference tones signal estimation portion 145.Wherein, this processing is equivalent to the work of the 5th near end signal delay portion 146.

Then, from the near end signal that this postponed, deduct the interference tones signal, output has suppressed near end signal (step S445) after the inhibition of echo and background noise.Wherein, this processing is equivalent to the work of second efferent 147.

And then, upgrade the interference tones signal evaluation characteristic of interference tones signal estimation portion 145 one by one, make square time average minimum (step S446) of this inhibition back near end signal, the interference tones when promptly not having near-end speech.This processing is equivalent to the work of interference tones signal evaluation characteristic renewal portion 148.

Wherein, the renewal of interference tones signal evaluation characteristic can be used known study method of identification, FRLS (Fast Recursive Least Squares, recurrence least square fast) method etc.

As mentioned above, the communicator of second embodiment of the invention suppresses to be provided with currency test section 131 in the parts 13 at echo, the second propagation time calculating part 141, the 3rd near end signal delay portion 142 and nearby place's end signal delay portion 143, residual voice signal efferent 144, interference tones signal estimation portion 145, the 5th near end signal delay portion 146, second efferent 147, and renewal portion 148, so can be suppressed at echo and the background noise that mixes far-end speech in the near-end speech that comprises in the near end signal and produce reliably according to the relative position between current remote signaling transformation component 112 and the first near-end speech transformation component 121 and the second near-end speech transformation component 122.

In addition, the communicator of second embodiment of the invention also can be as shown in figure 17, echo suppresses parts 13 and comprises: initial characteristic configuration part 149, according to currency test section 131 detected currencys, set the initial interference tone signal evaluation characteristic of renewal portion 148.

With reference to Figure 18, illustrate that the echo that comprises initial characteristic configuration part 149 suppresses the work of parts 13; Processing from step S401 to step S430 is identical with the echo of second execution mode inhibition parts 13, so omit its explanation.

After calculating second propagation time (step S430), according to the initial interference tone signal evaluation characteristic that is stored in advance in the memory 231, calculate the initial interference tone signal evaluation characteristic corresponding, be set at the initial value (step S431) of the interference tones signal evaluation characteristic of renewal portion 148 with currency test section 131 detected currencys.Wherein, this processing is equivalent to the work of initial characteristic configuration part 149.Under the initial interference tone signal evaluation characteristic corresponding with currency test section 131 detected currencys is not stored in situation in the memory 231, according to the initial interference tone signal evaluation characteristic that is stored in advance in the memory 231, wait with interpolation method and to calculate initial interference tone signal evaluation characteristic.

The later processing of step S405 is identical with the flow chart of Figure 15, so omit its explanation.

As mentioned above, in above-mentioned communicator, be provided with initial characteristic configuration part 149, suppress the Optimal Jamming tone signal evaluation characteristic of square time average minimum of near end signal afterwards so the interference tones evaluation characteristic of interference tones signal estimation portion 145 is promptly converged to make, can suppress echo and background noise more reliably.

Moreover, the communicator of second embodiment of the invention also can be as shown in figure 19, echo suppresses parts 13 and comprises: the first near-end person position detection part 137, cross-correlation function between second near end signal that first near end signal that 121 conversion go out according to the first near-end speech transformation component and 122 conversion of the second near-end speech transformation component go out, detecting with some in the first near-end speech transformation component 121 or the second near-end speech transformation component 122 is the near-end person's of benchmark position---the first near-end person position; And the second balanced portion 138, will be and the corresponding equalization characteristic of currency test section 131 detected currencys and the first near-end person position detection part, the 137 detected first near-end person positions from the second second frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent 151 outputs.

Here, comprise that with reference to explanation the echo of the communicator of the second balanced portion 138 suppresses the flow chart of Figure 20 of the work of parts 13, carries out second equilibrium treatment (step S420) between step S440 and step S408.Wherein, the processing from step S401 to step S440, and the later processing of step S408 identical with the work of the echo inhibition parts 13 of second execution mode, so omit the explanation of its work.

The work of second equilibrium treatment (step S420) is described with reference to the flow chart of Figure 11.

At first, according to the peak intervals of the whitening cross-correlation function between the near end signal that reads in from the first near-end speech transformation component 121 and the second near-end speech transformation component 122, calculating with phonetic entry parts 12 is the near-end person's of benchmark position---the first near-end person position (step S421).Wherein, this processing is equivalent to the work of the first near-end person position detection part 137.

Then, use the equalization characteristic that is stored in advance in the memory 231 to calculate and the first near-end person position detection part, 137 detected first near-end person positions and the corresponding equalization characteristic (step S422) of currency test section 131 detected currencys, carry out the equilibrium treatment (step S423) that suppresses the back near end signal with this equalization characteristic.Above-mentioned two processing are equivalent to the work of the second balanced portion 138.

Near-end person position deviation first reference position as mentioned above,, suppresses to be provided with the first near-end person position detection part, the 137 and second balanced portion 138 in the parts 13, so even can prevent that also speech quality from worsening at echo according to above-mentioned communicator.

Moreover, the communicator of second embodiment of the invention also can be as shown in figure 21, echo suppresses parts 13 and comprises: the second near-end person position detection part 139, and detecting with remote signaling transformation component 112 according to currency test section 131 detected currencys is the near-end person's of benchmark position---the second near-end person position; And the 3rd balanced portion 140, will be and the corresponding equalization characteristic of currency test section 131 detected currencys and the second near-end person position detection part, the 139 detected second near-end person positions from the second second frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent 151 outputs.

In the case, the person that supposes the near-end be positioned at remote signaling transformation component 112 over against second reference position, decide equalization characteristic according to currency test section 131 detected currencys.

Then, the second near-end person position detection part 139 was as illustrating in the first embodiment, the person that supposes the near-end be positioned at voice output parts 11 over against be that calculate second reference position of benchmark with voice output parts 11.

As mentioned above, according to above-mentioned communicator, suppress to be provided with the second near-end person position detection part the 139 and the 3rd balanced portion 140 in the parts 13 at echo, even so near-end person position deviation second reference position not only, the frequency distribution of the near-end speech the when person that also frequency distribution of near-end speech can be maintained the near-end is positioned at second reference position---reference frequency distributes, and can make the designs simplification of the structure of the second near-end person position detection part 139 than the first near-end person position detection part 137.

Wherein, in the explanation of first and second execution modes, illustrated according to the geometry site between some in remote signaling transformation component and the first near-end speech transformation component and the second near-end speech transformation component and calculated the variable delay time of setting in the first propagation time calculating part or the second propagation time calculating part, but also can be according to the time have calculated the variable delay time from the cross-correlation function of the impulse response of the first near-end speech transformation component and the output of the second near-end speech transformation component from remote signaling transformation component output impact signal.

In addition, in the above description, the first balanced portion, the second balanced portion, and the explanation of the concrete structure of the 3rd balanced portion have been omitted; Can use known FIR (finite impulse response) filter or IIR (infinite impulse response) filter.

If use the FIR filter, then can critically set equalization characteristic; If the use iir filter then can reduce operand.

With known bank of filters, also can be each frequency band decision equalization characteristic.

Moreover obviously the processing of the echo of above-mentioned first and second execution mode inhibition parts both can be carried out in time domain, also can carry out in frequency domain.But, under the situation about in frequency domain, handling, can reduce operand though compare with the processing in the time domain, carry out the frame transform of voice signal time span is in accordance with regulations handled to the frame of frequency domain, so produce the time delay suitable with frame length.

Therefore, preferably suitably select in time domain, to handle, or in frequency domain, handle according to purposes of the present invention.

Wherein, in the above description, supposed the remote signaling converter---loud speaker is one, as long as but obviously can regard a remote signaling converter as of equal valuely, even then used many loud speakers, also can use the present invention.

(the 3rd execution mode)

With reference to Figure 22, the structure of the communicator of third embodiment of the invention is described.To in each formation portion of the communicator that constitutes third embodiment of the invention, attached with the same member of the member of the communicator of the communicator of first embodiment of the invention or second execution mode with same label, omit its detailed description.The communicator of third embodiment of the invention comprises: voice output parts 11, and the output remote signaling is as far-end speech; Phonetic entry parts 12, the input near-end speech; And echo inhibition parts 13, suppress owing to the far-end speech from 11 outputs of voice output parts is imported into the echo that produces the phonetic entry parts 12.

Voice output parts 11 comprise: remote signaling transformation component 112, be provided with datum mark predetermined first distance of being separated by, and remote signaling is transformed to far-end speech.

Phonetic entry parts 12 comprise: the first near-end speech transformation component 121, be provided with the datum mark predetermined second distance of being separated by, and the near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component 122, being set at the first near-end speech transformation component 121 is on the preposition of benchmark, and the near-end speech of input is transformed to second near end signal.

Echo suppresses parts 13 and comprises: echo component test section 160, second near end signal that first near end signal that 121 conversion go out according to the first near-end speech transformation component and 122 conversion of the second near-end speech transformation component go out detects the signal corresponding with real near-end speech signal---the echo component signal in addition that comprises jointly in first near end signal and second near end signal; And the back near end signal efferent 170 of the 3rd inhibition, in some second near end signal that first near end signal that goes out from 121 conversion of the first near-end speech transformation component and 122 conversion of the second near-end speech transformation component go out, deduct echo component test section 160 detected echo component signals and export the 3rd and suppress the back near end signal.Here so-called echo component signal is meant the signal beyond the signal corresponding with near-end speech, is not limited to the echo that comprises in the near end signal of narrow sense, and comprises and the corresponding signal of all noises such as background noise.

Be applied to have at communicator under the situation of mobile phone of the housing that constitutes by two partial shell present embodiment, communicator as shown in Figure 3, housing has hinge 31 and with 2 partial shell 32,33 of hinge 31 combinations, can change 2 partial shell, 32,33 angulations---subtended angle θ by enough hinges 31.In addition, as shown in figure 22, also display 341 can be set on housing.Here, the direction indication shown in the label 352 is from the words person's of communicator direction.

Echo component test section 160 comprises: real near-end speech estimation portion, estimate the real near end signal corresponding with real near-end speech that comprises jointly in second near end signal that first near end signal that 121 conversion of the first near-end speech transformation component go out and 122 conversion of the second near-end speech transformation component go out; The 6th near end signal delay portion 166, the real near-end speech estimation of the some delays portion in second near end signal that first near end signal that 121 conversion of the first near-end speech transformation component are gone out or 122 conversion of the second near-end speech transformation component go out estimates the time that real near end signal is required; And echo component signal efferent 165,---the 6th postpones the difference of the real near end signal that near end signal and real near-end speech estimation portion estimate as the signal beyond the real near end signal---the first echo component signal of exporting the near end signal that the 6th near end signal delay portion 166 postponed.Real near-end speech estimation portion is made of following part: first sef-adapting filter 163, and the some adaptive-filterings that carries out in second near end signal that first near end signal that 121 conversion go out to the first near-end speech transformation component and 122 conversion of the second near-end speech transformation component go out are handled; And the first coefficient storage portion 164, store parameter---first filter coefficient that adaptive-filtering that first sef-adapting filter 163 carries out is handled; First sef-adapting filter 163 carries out adaptive-filtering to be handled, and makes from the absolute value minimum of the differential signal of echo component signal efferent 165 outputs.

The 3rd suppresses back near end signal efferent 170 comprises: echo component estimation portion, according to echo component signal from echo component signal efferent 165 output, signal---the second echo component signal beyond the signal of estimating to comprise in some in second near end signal that first near end signal that 121 conversion of the first near-end speech transformation component go out or 122 conversion of the second near-end speech transformation component go out corresponding with near-end speech; The 7th near end signal delay portion 176, the some delayed echo component estimations portion in second near end signal that first near end signal that 121 conversion of the first near-end speech transformation component are gone out or 122 conversion of the second near-end speech transformation component go out estimates the required time of signal component of echo; And the 3rd efferent 175, the near end signal that postponed from the 7th near end signal delay portion 176---the 7th postpones to deduct the near end signal the second echo component signal that echo component estimation portion estimates exports the 3rd and suppresses the back near end signal.

Echo component estimation portion is made of following part: second sef-adapting filter 173, carry out adaptive-filtering to the echo component signal from 165 outputs of echo component signal efferent and handle; And the second coefficient storage portion 174, store parameter---second filter coefficient that adaptive-filtering that second sef-adapting filter 173 carries out is handled; Second sef-adapting filter 173 carries out adaptive-filtering to be handled, and makes from the absolute value minimum of the differential signal of the 3rd efferent 175 outputs.

Echo suppresses parts 13 and also comprises: currency test section 131, detect the currency of the parameter of the relative position between some in regulation remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122; The first initial value determination section 161, according to the currency of currency test section 131 detected parameters, parameter---the initial value of first filter coefficient that the adaptive-filtering that decides first sef-adapting filter 163 to carry out is handled; And the second initial value determination section 171, parameter---the initial value of second filter coefficient that the adaptive-filtering that decides second sef-adapting filter 173 to carry out according to the currency of currency test section 131 detected parameters is handled.

Echo suppresses parts 13 and also comprises: adaptive control portion 181, when the predetermined near end signal in first near end signal or second near end signal is more remarkable than remote signaling, indicate first sef-adapting filter 163 to upgrade first filter coefficient; And when remote signaling is more remarkable than the predetermined near end signal in first near end signal or second near end signal, indicate second sef-adapting filter, 173 changes, second filter coefficient.

Below, the flow process of the signal in the communicator of third embodiment of the invention is described.At first, the output signal from the second near-end speech transformation component 122 is imported in first sef-adapting filter 163.Here, on first sef-adapting filter 163, connecting the first coefficient storage portion 164.Be imported in the echo component signal efferent 165 from the output signal of the 6th near end signal delay portion 166 and from the output signal of first sef-adapting filter 163.

Output signal from echo component signal efferent 165 is imported in first sef-adapting filter 163 and second sef-adapting filter 173.Here, on second sef-adapting filter 173, connecting the second coefficient storage portion 174, from wherein reading filter coefficient.Be imported in the 3rd efferent 175 from the output signal of the 7th near end signal delay portion 176 and from the output signal of second sef-adapting filter 173.

Output signal from the 3rd efferent 175 is imported in second sef-adapting filter 173, and is sent to the distal side and suppresses signal as echo.Hinge 31 is connected on the currency test section 131 as the state-detection parts, and currency test section 131 detects the isoparametric currency of subtended angle θ of hinges 131.The output signal of currency test section 131 is imported in the first initial value determination section 161 and the second initial value determination section 171.

The output signal of the first initial value determination section 161 is imported in first sef-adapting filter 163, and the output signal of the second initial value determination section 171 is imported in second sef-adapting filter 173.From remote signaling transformation component 112 remote signaling is input to the adaptive control portion 181.In addition, also import second near end signal from the second near-end speech transformation component 122 in adaptive control portion 181, the output signal of adaptive control portion 181 is output to respectively in first sef-adapting filter 163 and second sef-adapting filter 173.

The work of the communicator that as above constitutes is described with Figure 23.At first, echo suppresses parts 13 and judges whether the end of conversation condition of power cut-off sets up (S501) owing to cut off the operation such as power supply of communicator, is being judged as under the situation of establishment power cut-off; Being judged as under the invalid situation, work on, handle proceeding to step S502.

In step S501, be judged as under the invalid situation of end of conversation condition, echo suppress parts 13 according to the level of near end signal whether at predetermined level with first-class, judge whether conversation begins (S502).In S502, be judged as conversation when having begun, handle proceeding to S503; Being judged as conversation when not beginning, turn back to step S501.

Be judged as conversation when having begun in S502, currency test section 131 detects the shape (S503) of current housing.In present embodiment 3, the detection of the shape of current housing is undertaken by the subtended angle θ that detects and calculate housing with the currency test section 131 that connects on the hinge 31.Here, as long as currency test section 131 can be the signal of telecommunication with the size conversion of subtended angle θ, for example can wait and realize with rotary encoder, variable resistor.In addition, currency test section 131 is the parameter of the relative position between some in computational rules remote signaling transformation component 112 and the first near-end speech transformation component 121 or the second near-end speech transformation component 122 continuously, also can calculate angle discretely.Moreover, also can wait only 2 states such as state that housing opens and closing state that detect with switch.

Then, the first initial value determination section 161 and the second initial value determination section 171 are according to currency test section 131 detected parameters, decide the initial value of first filter coefficient and the initial value of second filter coefficient respectively, set and give first sef-adapting filter 163 and second sef-adapting filter 173 (S504).

Specifically, the first initial value determination section 161 will with the corresponding signal of the subtended angle of housing (hereinafter referred to as the subtended angle signal.) output in first sef-adapting filter 163, in the filter coefficient that first sef-adapting filter 163 is stored from the first coefficient storage portion 164, read and the corresponding filter coefficient of above-mentioned subtended angle signal.Here, the first coefficient storage portion 164 stores the corresponding filter coefficient of subtended angle with specified number or amount in advance, the subtended angle that first sef-adapting filter 163 also can identify according to the subtended angle signal with input waits the filter coefficient of reading nearest angle with interpolation method.

Equally, the second initial value determination section 171 will output in first sef-adapting filter 173 with the corresponding signal of the subtended angle of housing, in the filter coefficient that second sef-adapting filter 173 is stored from the second coefficient storage portion 174, read and the corresponding filter coefficient of subtended angle signal.Here, the second coefficient storage portion 174 stores the corresponding filter coefficient of subtended angle with specified number or amount in advance, the subtended angle that second sef-adapting filter 173 also can identify according to the subtended angle signal with input waits the filter coefficient of reading nearest angle with interpolation method.

Determined the initial value of filter coefficient in step S504 after, the first near-end speech transformation component 121 and the second near-end speech transformation component 122 are collected near-end speech, generate first near end signal and second near end signal (S505).The later processing of step S506 also can be an analog, but generally carries out with Digital Signal Processing.Therefore, below, suppose that first near end signal and second near end signal are transformed to digital signal by not shown A/D transform component from analog signal.

In addition, carrying out with Digital Signal Processing under the situation of the later processing of step S506, generally adopting the structure that in communicator, comprises microcomputer or DSP elements such as (Digital Signal Processor, digital signal processors).In this case, suppose that the later processing of step S506 described by the program of carrying among microcomputer or the DSP.

Be transformed to first near end signal of digital signal and second near end signal still in time domain or be transformed frequency domain earlier and carry out later processing.Under situation about still in time domain, handling, carry out sampling processing by the sampling period in the AD transform component.And transforming under the situation of frequency domain, the input signal in the frame with regular hour width is concentrated carried out the frame processing.Under the situation that frame is handled, have than little this advantage of sampling processing operand, produce delay this shortcoming suitable but also have with frame length.

Therefore, select according to purposes wherein some.

After in step S505, having generated first near end signal and second near end signal of digital signal, first near end signal is by the delay of 166 additional stipulated times of the 6th near end signal delay portion, and second near end signal carries out adaptive-filtering by first sef-adapting filter 163 and handles (S506).Here, retardations that the 6th near end signal delay portion 166 is additional are set more suitablely with the coefficient appearance of first sef-adapting filter 163 etc.Specifically, preferably be made as about half of tap length of first sef-adapting filter 163.

The integral multiple in the sampling period the when method of additional delay preferably will be rounded among the step S506 and be transformed to digital signal with transform component time of delay comes additional delay with simple buffer.In addition, above-mentioned adaptive-filtering is handled by the method to the coefficient of output signal convolution first filter of the second near-end speech transformation component 122 and is carried out Filtering Processing.

In step S506, first near end signal has been added delay, second near end signal has been carried out after adaptive-filtering handles, signal after signal after the additional delay and adaptive-filtering are handled is imported in the echo component signal efferent 165, is calculated the difference (S507) of these signals by echo component signal efferent 165.The difference that echo component signal efferent 165 is calculated is fed back in first sef-adapting filter 163, carries out adaptive-filtering by 163 pairs second near end signals of first sef-adapting filter and handles, and makes the absolute value minimum of differential signal.

Then, first near end signal is carried out adaptive-filtering from the output signal of echo component signal efferent 165 by second sef-adapting filter 173 and is handled (S508) by the delay of 176 additional stipulated times of the 7th near end signal delay portion.Here, set the 7th near end signal delay portion 176 additional retardations proper with the coefficient equal in length of second sef-adapting filter 173.Specifically, preferably be made as about half of tap length of second sef-adapting filter 173.

The integral multiple in the sampling period the when method of additional delay preferably will be rounded among the step S506 and be transformed to digital signal with the AD transform component time of delay comes additional delay with simple buffer.In addition, above-mentioned adaptive-filtering is handled by the method from the coefficient of output signal convolution second filter of echo component signal efferent 165 is carried out Filtering Processing.

In step S508 to first near end signal added delay, to having carried out from the output signal of echo component signal efferent 165 after adaptive-filtering handles, signal after signal after the additional delay and adaptive-filtering are handled is imported in the 3rd efferent 175, is calculated the difference (S509) of these signals by echo component signal efferent 165.The difference that the 3rd efferent 175 is calculated is fed back in second sef-adapting filter 173, carries out adaptive-filtering by 173 pairs of output signals from echo component signal efferent 165 of second sef-adapting filter and handles, and makes the absolute value minimum of differential signal.

The signal that generates among the step S509 suppresses signal as echo and is output to distal side (S510).

Then, upgrade first filter coefficient and second filter coefficient (S511).Adaptive control portion 181 judges whether to upgrade the filter coefficient of each sef-adapting filter 163,173, upgrades filter coefficient will upgrade the time being judged as.Adaptive control portion 181 is being judged as near end signal than the filter coefficient that upgrades first sef-adapting filter 163 under the situation of the remote signaling brilliance of distal side input, makes the power output minimum of echo component signal efferent 165.In addition, adaptive control portion 181 upgrades the filter coefficient of second sef-adapting filter 173 under than the situation of near end signal brilliance being judged as remote signaling from distal side input, makes the power output minimum of the 3rd efferent 175.

The preferably following method of the method that adaptive control portion 181 is judged: briefly, surpass at the power of near end signal under the relevant high situation between the output signal of the first threshold of regulation and the remote signaling and the second near-end speech transformation component 122, be judged to be the remote signaling brilliance; Surpass at the power that the power of remote signaling surpasses the output signal of the predetermined second threshold value and the second near-end speech transformation component 122 under the situation of the 3rd threshold value of regulation, be judged to be near-end words person voice brilliance.Between the output signal of the remote signaling and the second near-end speech transformation component 122 relevant can according to the maximum of the whitening cross-correlation function between the output signal of the remote signaling and the second near-end speech transformation component 122 whether the value of regulation with on judge.

Perhaps, adaptive control portion 181 also can surpass the first threshold of regulation at the power of the output signal of the second near-end speech transformation component 122 and the power of remote signaling surpasses under the situation of predetermined second threshold value, is judged to be remote signaling than near end signal brilliance; Surpass at the power that the power of remote signaling surpasses the output signal of the 3rd threshold value of regulation and the second near-end speech transformation component 122 under the situation of the 4th threshold value of regulation, be judged to be near end signal than remote signaling brilliance.

By controlling like this, the filter coefficient of first sef-adapting filter 163 is updated under the situation of near-end words person brilliance.Consequently, though echo component signal efferent 165 output comprise with remote signaling by the far-end speech of 112 conversion of remote signaling transformation component and output around to the corresponding signal of the voice of the second near-end speech transformation component 122, suppressed the signal with the corresponding signal of near-end words person voice.Therefore, from the strong signal of echo component signal efferent 165 output echo component.

Moreover the filter coefficient of second sef-adapting filter 173 is updated under the situation of remote signaling brilliance.After second filter coefficient is updated, in the output signal of the 7th near end signal delay portion 176, comprise from the far-end speech of remote signaling transformation component 112 output around to the echo of the second near-end speech transformation component 122 and signal---the near end signal of near-end words person voice, but the echo component that comprises in the input signal of second sef-adapting filter 173 has been offseted.Consequently, in the output of the 3rd efferent 175, near-end words person voice are enhanced output.The method of upgrading the filter coefficient of first sef-adapting filter 163 and second sef-adapting filter 173 adopts known method based on Adaptive Signal Processing proper, for example can adopt study method of identification or FRLS method etc.

After in step S511, upgrading the finishing dealing with of filter coefficient, wait to judge whether voice call finishes (S512),, handle proceeding to step S513 being judged as under the situation that voice call finished according to the level of near end signal; Be judged as under the unclosed situation, turning back to step S503, repeating above-mentioned processing.Here, when reprocessing, also can omit the processing among step S503, the S504.

After voice call finishes in step S512, the filter coefficient of first sef-adapting filter 163 is associated with currency test section 131 detected subtended angles and stores in the first coefficient storage portion 164, the filter coefficient of second sef-adapting filter 173 is associated with currency test section 131 detected subtended angles and stores in the second coefficient storage portion 174, turn back to step S501, repeat above-mentioned processing.

Below, the effect of the communicator of third embodiment of the invention is described with accompanying drawing.Figure 24 is the exemplary plot that comprises in the remote signaling under the situation of voice signal from the output waveform of the voice signal of remote signaling transformation component 112 outputs.Here, Figure 24 (a)～(c) illustrates from the voice signal of remote signaling transformation component 112 output does not have distortion, is the example that remote signaling is transformed to the output waveform under the situation of far-end speech linearly.

Figure 24 (a) is the figure of signal waveform that does not carry out outputing under the situation that echo suppress to handle the signal of distal side.Under the situation of not carrying out echo inhibition processing, from the voice of remote signaling transformation component 112 output around to the first near-end speech transformation component 121 and the second near-end speech transformation component 122, so shown in Figure 24 (a), around to the voice signal that causes of voice be superimposed on the signal that outputs to the distal side as echo.

Shown in Figure 24 (a), can not ignore, disturb under the situation of conversation at echo, expectation suppresses echo.Figure 24 (b) is the figure that has adopted the signal waveform of the signal that outputs to the distal side under the existing situation based on the echo eliminator of learning method of identification.As can be known, even under the situation that adopts existing echo eliminator, also can fully suppress echo.

Figure 24 (c) is the figure that has adopted the signal waveform of the signal that outputs to the distal side under the situation of communicator of third embodiment of the invention.As can be known, under the situation of the communicator that has adopted third embodiment of the invention, also can fully suppress echo.

Then, Figure 24 (d)～(f) illustrates from the voice signal of remote signaling transformation component 112 output and loses under the genuine situation, is the example that remote signaling non-linearly is transformed to the output waveform under the situation of far-end speech.Figure 24 (d) is the figure of signal waveform that does not carry out outputing under the situation that echo suppress to handle the signal of distal side.As can be known, under the situation of not carrying out echo inhibition processing, the echo of big amplitude is arranged.

Figure 24 (e) is the figure that has adopted the signal waveform of the signal that outputs to the distal side under the situation of existing echo eliminator.As can be known, under the situation that has adopted existing echo eliminator, though can suppress a part of echo, leakage takes place disappear.Existing echo eliminator carries out linear echo to be suppressed to handle, so the output signal of remote signaling transformation component 112 is had nonlinear systems such as distortion, can not suitably predict echo on principle.The echo that dopes like this is non-linear, and is residual so depart from the component of genuine echo not to be suppressed, so under the situation that adopts existing echo eliminator, echogenic leakage disappears.

Figure 24 (f) is the figure that has adopted the signal waveform of the signal that outputs to the distal side under the situation of communicator of third embodiment of the invention.As can be known, under the situation of the communicator that has adopted third embodiment of the invention, though fail complete cancellation echo, the leakage of echo disappears littler than existing echo eliminator.

In recent years, the miniaturization day by day of terminals such as mobile phone constitutes the also miniaturization day by day thereupon of loud speaker of voice output parts 11.Along with the loud speaker miniaturization, loudspeaker distortions can not be avoided, so produce with the indeterminable problem of existing echo eliminator.In third embodiment of the invention, illustrated, and can suppress the echo under the situation that the distortion of this loud speaker exists more effectively.

As mentioned above, the communicator of third embodiment of the invention is handled by carry out adaptive-filtering with first sef-adapting filter, make from the absolute value minimum of the differential signal of echo component signal efferent output, detect and current remote signaling transformation component and corresponding first echo component of the relative position between the near-end speech transformation component, second sef-adapting filter generates second echo component that is included in the near end signal, the 3rd efferent is removed echo component, even so under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can keep the echo rejection, carry out high-quality conversation.

In addition, the first initial value determination section and the second initial value determination section decide the initial value of first filter coefficient and second filter coefficient according to the currency of the detected parameter of currency test section, even so under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can in the conversation beginning back short time, keep the echo rejection, carry out high-quality conversation.

Moreover, adaptive control portion is according to remote signaling and near end signal, indicate first sef-adapting filter and second sef-adapting filter to upgrade first filter coefficient and second filter coefficient, so can follow the tracks of the variation of communication environment well, even under the situation that current remote signaling transformation component and the relative position between the near-end speech transformation component have been changed, also can keep the echo rejection at short notice, carry out high-quality conversation.

Wherein, in the above-described embodiment, all adopted the first near-end speech transformation component 121 to be configured in, but the invention is not restricted to this than the structure of the second near-end speech transformation component 122 from the nearer position of remote signaling transformation component.For example, obviously also can adopt the second near-end speech transformation component 122 to be configured in than the structure of the first near-end speech transformation component 121 from the nearer position of remote signaling transformation component.

Utilizability on the industry of the present invention is, even communicator of the present invention can be applied to the purposes such as communicator that relative position, the echo rejection of change loud speaker and microphone do not worsen yet, can carry out the high-quality conversation.

Claims (14)

1, a kind of communicator is characterized in that,

Comprise: housing has datum mark; The voice output parts are exported remote signaling as far-end speech; The phonetic entry parts, the input near-end speech; And echo inhibition parts, suppress owing to the said distal ends voice from above-mentioned voice output parts output are imported into the echo that produces the above-mentioned phonetic entry parts;

Above-mentioned voice output parts comprise: the remote signaling transformation component, be provided with said reference point predetermined first distance of being separated by, and be the said distal ends voice with said distal ends signal transformation;

Above-mentioned phonetic entry parts comprise: the first near-end speech transformation component, be provided with the said reference point predetermined second distance of being separated by, and the above-mentioned near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component, being set at the above-mentioned first near-end speech transformation component is on the preposition of benchmark, and the above-mentioned near-end speech of input is transformed to second near end signal;

Above-mentioned echo suppresses parts and comprises: the currency test section, detect the currency of the parameter of the relative position between some in regulation said distal ends signal transformation portion and above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component; The first propagation time calculating part, calculate first propagation time and second propagation time corresponding according to first function and second function with the detected above-mentioned currency of above-mentioned currency test section, this first function and second function with above-mentioned parameter as independent variable, the propagation time that the said distal ends voice that this first function goes out the conversion of said distal ends signal transformation portion propagate into the above-mentioned first near-end speech transformation component, the said distal ends voice that this second function goes out the conversion of said distal ends signal transformation portion propagated into the propagation time of the above-mentioned second near-end speech transformation component as the subordinate variable as the subordinate variable; And first suppress back near end signal efferent, and above-mentioned first propagation time of calculating according to the above-mentioned first propagation time calculating part and above-mentioned second propagation time are used in output and the echo suppression characteristic that changed has suppressed first of echo and suppresses the back near end signal.

2, communicator as claimed in claim 1, it is characterized in that, above-mentioned first suppresses back near end signal efferent comprises: the first near end signal delay portion, and above-mentioned first propagation time of calculating according to the above-mentioned first propagation time calculating part postpones above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out; The second near end signal delay portion, above-mentioned second propagation time of calculating according to the above-mentioned first propagation time calculating part postpones above-mentioned second near end signal that the above-mentioned second near-end speech transformation component conversion goes out; And first efferent, that exports that the above-mentioned first near end signal delay portion postponed first postpones the poor of the second delay near end signal that near end signal and the above-mentioned second near end signal delay portion postponed.

3, communicator as claimed in claim 1, it is characterized in that, above-mentioned echo suppresses parts and comprises: the first balanced portion will be and the corresponding equalization characteristic of the detected above-mentioned currency of above-mentioned currency test section from the above-mentioned first above-mentioned first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

4, communicator as claimed in claim 1, it is characterized in that, above-mentioned echo suppresses parts and comprises: the first near-end person position detection part, cross-correlation function between above-mentioned second near end signal that above-mentioned first near end signal that goes out according to the above-mentioned first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out, detecting with some in above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component is the near-end person's of benchmark position---the first near-end person position; And the second balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned first near-end person position of the above-mentioned first near-end person position detection part from the above-mentioned first above-mentioned first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

5, communicator as claimed in claim 1, it is characterized in that, above-mentioned echo suppresses parts and comprises: the second near-end person position detection part, according to the detected above-mentioned currency of above-mentioned currency test section, detecting with above-mentioned remote signaling transformation component is the above-mentioned near-end person's of benchmark current position---the second near-end person position; And the 3rd balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned second near-end person position of the above-mentioned second near-end person position detection part from the above-mentioned first above-mentioned first frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

6, a kind of communicator is characterized in that,

Comprise: housing has datum mark; The voice output parts are exported remote signaling as far-end speech; The phonetic entry parts, the input near-end speech; And echo inhibition parts, suppress owing to the said distal ends voice from above-mentioned voice output parts output are imported into the echo that produces the above-mentioned phonetic entry parts;

Above-mentioned voice output parts comprise: the remote signaling transformation component, be provided with said reference point predetermined first distance of being separated by, and be the said distal ends voice with said distal ends signal transformation;

Above-mentioned phonetic entry parts comprise: the first near-end speech transformation component, be provided with the said reference point predetermined second distance of being separated by, and the above-mentioned near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component, being set at the above-mentioned first near-end speech transformation component is on the preposition of benchmark, and the above-mentioned near-end speech of input is transformed to second near end signal;

Above-mentioned echo suppresses parts and comprises: the currency test section, detect the currency of the parameter of the relative position between some in regulation said distal ends signal transformation portion and above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component; The second propagation time calculating part, calculate above-mentioned three propagation time corresponding and above-mentioned the 4th propagation time according to the 3rd function and the 4th function with the detected above-mentioned currency of above-mentioned currency test section, the 3rd function and the 4th function with above-mentioned parameter as independent variable, the 3rd propagation time that the 3rd function propagates into the above-mentioned first near-end speech transformation component with above-mentioned near-end speech, the 4th function propagated into the 4th propagation time of the above-mentioned second near-end speech transformation component as the subordinate variable with above-mentioned near-end speech as the subordinate variable; Residual voice signal extraction unit is used above-mentioned the 3rd propagation time of calculating according to the above-mentioned second propagation time calculating part and above-mentioned the 4th propagation time and the residual voice signal extraction property that changed extracts residual voice signal; And second suppress back near end signal efferent, deducts above-mentioned residual voice signal that above-mentioned residual voice signal extraction unit extracts in some above-mentioned second near end signal that above-mentioned first near end signal that goes out from the above-mentioned first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out and export second and suppress the back near end signal.

7, communicator as claimed in claim 6 is characterized in that,

Above-mentioned residual voice signal extraction unit comprises: the 3rd near end signal delay portion, and above-mentioned the 3rd propagation time of calculating according to the above-mentioned second propagation time calculating part postpones above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out; Nearby place's end signal delay portion, above-mentioned the 4th propagation time of calculating according to the above-mentioned second propagation time calculating part postpones above-mentioned second near end signal that the above-mentioned second near-end speech transformation component conversion goes out; And residual voice signal efferent, that exports that above-mentioned the 3rd near end signal delay portion postponed the 3rd postpones difference between the 4th delay near end signal that near end signal and above-mentioned nearby place's end signal delay portion postponed as residual voice signal;

Above-mentioned second suppresses back near end signal efferent comprises: interference tones signal estimation portion, according to above-mentioned residual voice signal from above-mentioned residual voice signal efferent output, signal---the interference tones signal beyond the signal of estimating to comprise in some in above-mentioned second near end signal that above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out or the above-mentioned second near-end speech transformation component conversion go out corresponding with above-mentioned near-end speech; The 5th near end signal delay portion, the above-mentioned interference tones signal of the some delays estimation portion in above-mentioned second near end signal that above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion is gone out or the above-mentioned second near-end speech transformation component conversion go out estimates the interference tones required time of signal; Second efferent, the above-mentioned near end signal that postponed from above-mentioned the 5th near end signal delay portion---the 5th postpones the near end signal, deducts above-mentioned interference tones signal that above-mentioned interference tones signal estimation portion estimates and exports second and suppress the back near end signal; And renewal portion, the interference tones signal evaluation characteristic of above-mentioned interference tones signal estimation portion is updated to one by one makes the above-mentioned second Optimal Jamming tone signal evaluation characteristic that suppresses square time average minimum of back near end signal.

8, communicator as claimed in claim 7 is characterized in that, above-mentioned echo suppresses parts and comprises: the initial characteristic configuration part, according to the detected above-mentioned currency of above-mentioned currency test section, set the initial interference tone signal evaluation characteristic of above-mentioned renewal portion.

9, as claim 6 or 7 described communicators, it is characterized in that, above-mentioned echo suppresses parts and comprises: the first near-end person position detection part, cross-correlation function between above-mentioned second near end signal that above-mentioned first near end signal that goes out according to the above-mentioned first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out, detecting with some in above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component is the above-mentioned near-end person's of benchmark position---the first near-end person position; And the second balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned first near-end person position of the above-mentioned first near-end person position detection part from the above-mentioned second above-mentioned second frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

10, as claim 6 or 7 described communicators, it is characterized in that, above-mentioned echo suppresses parts and comprises: the second near-end person position detection part, according to the detected above-mentioned currency of above-mentioned currency test section, detecting with above-mentioned remote signaling transformation component is the above-mentioned near-end person's of benchmark position---the second near-end person position; And the 3rd balanced portion, will be and the corresponding equalization characteristic of detected above-mentioned currency of above-mentioned currency test section and the detected above-mentioned second near-end person position of the above-mentioned second near-end person position detection part from the above-mentioned second above-mentioned second frequency distribution equilibrium that suppresses the back near end signal that suppresses back near end signal efferent output.

11, a kind of communicator is characterized in that,

Comprise: housing has datum mark; The voice output parts are exported remote signaling as far-end speech; The phonetic entry parts, the input near-end speech; And echo inhibition parts, suppress owing to the said distal ends voice from above-mentioned voice output parts output are imported into the echo that produces the above-mentioned phonetic entry parts;

Above-mentioned voice output parts comprise: the remote signaling transformation component, be provided with said reference point predetermined first distance of being separated by, and be the said distal ends voice with said distal ends signal transformation;

Above-mentioned phonetic entry parts comprise: the first near-end speech transformation component, be provided with the said reference point predetermined second distance of being separated by, and the above-mentioned near-end speech of input is transformed to first near end signal; And the second near-end speech transformation component, being set at the above-mentioned first near-end speech transformation component is on the preposition of benchmark, and the above-mentioned near-end speech of input is transformed to second near end signal;

Above-mentioned echo suppresses parts and comprises: the echo component test section, above-mentioned second near end signal that above-mentioned first near end signal that goes out according to the first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out detects signal---the echo component signal beyond the signal corresponding with real near-end speech that comprises jointly in above-mentioned first near end signal and above-mentioned second near end signal; And the back near end signal efferent of the 3rd inhibition, in some above-mentioned second near end signal that above-mentioned first near end signal that goes out from the first near-end speech transformation component conversion and the above-mentioned second near-end speech transformation component conversion go out, deduct the detected above-mentioned echo component signal of above-mentioned echo component test section and export the 3rd and suppress the back near end signal.

12, communicator as claimed in claim 11 is characterized in that,

Above-mentioned echo component test section comprises: real near-end speech estimation portion, estimate the pairing real near end signal of real near-end speech that comprises jointly in above-mentioned second near end signal that above-mentioned first near end signal that the first near-end speech transformation component conversion goes out and the above-mentioned second near-end speech transformation component conversion go out; The 6th near end signal delay portion, the above-mentioned real near-end speech estimation of the some delays in above-mentioned second near end signal that above-mentioned first near end signal that the first near-end speech transformation component conversion is gone out or the above-mentioned second near-end speech transformation component conversion go out portion estimates the above-mentioned required time of real near end signal; And echo component signal efferent, the above-mentioned near end signal that above-mentioned the 6th near end signal delay portion was postponed---the 6th postpones the difference of the above-mentioned real near end signal that near end signal and above-mentioned real near-end speech estimation portion estimate as the signal beyond the above-mentioned real near end signal---first echo component signal output;

The above-mentioned the 3rd suppresses back near end signal efferent comprises: echo component estimation portion, according to above-mentioned echo component signal from the output of above-mentioned echo component signal efferent, signal---the second echo component signal beyond the signal of estimating to comprise in some in above-mentioned second near end signal that above-mentioned first near end signal that the above-mentioned first near-end speech transformation component conversion goes out or the above-mentioned second near-end speech transformation component conversion go out corresponding with above-mentioned near-end speech; The 7th near end signal delay portion, the above-mentioned echo component estimation of the some delays portion in above-mentioned second near end signal that above-mentioned first near end signal that the first near-end speech transformation component conversion is gone out or the above-mentioned second near-end speech transformation component conversion go out estimates the required time of signal component of above-mentioned echo; And the 3rd efferent, the above-mentioned near end signal that postponed from above-mentioned the 7th near end signal delay portion---the 7th postpones to deduct the near end signal the above-mentioned second echo component signal that above-mentioned echo component estimation portion estimates exports the 3rd and suppresses the back near end signal.

13, communicator as claimed in claim 12 is characterized in that,

Above-mentioned real near-end speech estimation portion has: first sef-adapting filter, some adaptive-filterings that carries out in above-mentioned second near end signal that above-mentioned first near end signal that the first near-end speech transformation component conversion is gone out and the above-mentioned second near-end speech transformation component conversion go out are handled, and make from the absolute value minimum of the signal of above-mentioned echo component signal efferent output; Above-mentioned echo component estimation portion has: second sef-adapting filter, and the above-mentioned echo component signal from above-mentioned echo component signal efferent output is carried out adaptive-filtering handle, make from the absolute value minimum of the signal of above-mentioned the 3rd efferent output;

Above-mentioned echo suppresses parts and comprises: the currency test section, detect the currency of the parameter of the relative position between some in regulation said distal ends signal transformation portion and above-mentioned first near-end speech transformation component or the above-mentioned second near-end speech transformation component; The first initial value determination section, according to the currency of the detected parameter of above-mentioned currency test section, parameter---the initial value of first filter coefficient that the adaptive-filtering that decides above-mentioned first sef-adapting filter to carry out is handled; And the second initial value determination section, parameter---the initial value of second filter coefficient that the adaptive-filtering that decides above-mentioned second sef-adapting filter to carry out according to the currency of the detected parameter of above-mentioned currency test section is handled.

14, communicator as claimed in claim 13, it is characterized in that, above-mentioned echo suppresses parts and comprises: adaptive control portion, when the predetermined near end signal in above-mentioned first near end signal or above-mentioned second near end signal is more remarkable than said distal ends signal, indicate above-mentioned first sef-adapting filter to upgrade above-mentioned first filter coefficient; When the said distal ends signal is more remarkable than the predetermined near end signal in above-mentioned first near end signal or above-mentioned second near end signal, indicate above-mentioned second sef-adapting filter to upgrade above-mentioned second filter coefficient.

Images