CN102405634B - Speakerphone apparatus - Google Patents

Abstract

A speakerphone apparatus includes a microphone and a speaker, a voice switch; an echo canceller, a doubletalk detector, and an echo suppressor. The voice switch includes a transmission side loss insertion unit, a reception side loss insertion unit, and an insertion loss amount control unit, and the insertion loss amount control unit includes a total loss amount calculator and an insertion loss amount allocation processor. Further, the total loss amount calculator has an update mode in which a total sum of the loss amounts inserted in the respective paths is calculated and adaptively updated based on estimations of each feedback gain, and a fixed mode in which a total loss amount is fixed to a initial value, and operates in the fixed mode while the echo canceller does not sufficiently converge after starting communications and in the update mode after the echo canceller fully converges.

Description

Speakerphone apparatus

Technical field

The present invention relates to a kind of speakerphone apparatus for house or office.

Background technology

Known a kind of speakerphone apparatus in prior art, it has reached the hands-free communication without receiver device while communication in the following manner: use loud speaker that voice signal is amplified, and exported to the calling party far away apart from communication terminal; And collect the voice of calling party with microphone, and voice are sent to his/her the other side's communication terminal.Herein, the voice signal that loud speaker sends the communication terminal from the other side amplifies, and microphone is collected the voice that send from calling party.

In this speakerphone apparatus, can be coupled to set up acoustic feedback path by the acoustics between microphone and loud speaker, may hear the echo (acoustic echo) of making us uncomfortable through path thus.In addition, if form closed-loop via described feedback path under a frequency component in communication system, and the gain of described closed-loop exceedes 1, and whistle (howling) may occur under described frequency.Therefore, provide a kind of Echo Canceller (echo canceller) and voice switch (voice switch), to prevent the echo and the whistle that make us uncomfortable.

Described voice switch is constantly assessed communications status (transmission state, accepting state), and by suitably distributing loss based on assessment result, loss is provided to transmitted signal path and receive in signal path.Described Echo Canceller comprises: sef-adapting filter (adaptive filter), for the impulse response in Recognition feedback path, and estimate doubtful (suspected) echo components in feedback path according to the signal that inputs to sef-adapting filter; And subtracter, for deducting estimated echo components from the signal that inputs to subtracter via feedback path.Generally speaking, described sef-adapting filter needs the learning time of some seconds, so that the impulse response in Recognition feedback path.Therefore,, following closely in some seconds that communicate by letter after starting, cannot effectively suppress echo.But, owing to having set up closed-loop in communication system, so wherein may make us uncomfortable echo or whistle.

In this regard, existing people proposes a kind ofly can follow communication closely and start rear inhibition and make us the speakerphone apparatus (for example referring to Japanese Patent Application Publication No.2002-359580) that uncomfortable echo or whistle occur.

In this existing example, although not convergence immediately after communication starts of Echo Canceller, but voice switch operates under fixed mode, the total amount (total losses amount) of the loss wherein providing in signal path is fixed on enough large initial value, suppresses thus the echo or the whistle that make us uncomfortable.On the other hand, after Echo Canceller Complete Convergence, voice switch, at the work of finishing drilling of new model more, wherein needs by occasion and upgrades total losses amount, and therefore can reach full duplex (full duplex) communication.

Simultaneously, such as, if combine digital communication between multiple speakerphone apparatus (Internet Protocol (IP) communication etc.) may postpone because of packet transaction or for the communication buffer processing that suppresses shake (jitter) in speech data sends.Generally speaking, while occurring to postpone during the known speech data when between multiple speakerphone apparatus sends, along with time of delay is elongated, the tolerance variation of speaker to echo when communication, even and if therefore small echo also can make speaker not feel well.

By set the initial value larger than the initial value of existing analog communication under fixed mode, can prevent the echo of making us here uncomfortable.But the amount of degradation of echo tolerance (tolerance) may very shellfish of exceedance, this more cannot covered by the amount of suppression of Echo Canceller under new model, and therefore can only in half-duplex operation, communicate by letter but not communicate by letter in full-duplex communication.In addition, in the time that ambient noise level is higher, may there is following problem: the only side joint that voice switch continues in transmitter side and receiver side leads to, and therefore makes not to be sent to opposite side from the voice of a described side.

On the other hand, for example, when combine digital between speakerphone apparatus and the communicator of use receiver (is communicated by letter, IP communication etc.) time, can make the residual echo decay occurring in the transmitted signal path of speakerphone apparatus by the echo suppressor (echo suppressor) being wherein equipped with, reach thus full-duplex communication.But if there is the noise of high level the signal that is sent to the other side's communicator from speakerphone apparatus, echo suppressor not only makes residual echo decay, but also can make momently noise attentuation.This of short duration decay to noise is postponed, and is discovered by the other side via receiver.Therefore, use the other side of receiver to feel the disconnection sensation of making us uncomfortable in communication period.

In addition, if the other side uses receiver, can be by the attenuation of echo suppressor being set to such an extent that also littlely during with speakerphone apparatus than the other side prevent disconnection sensation mentioned above.But, there is in the case the shortcoming of hearing residual echo via receiver.

Summary of the invention

In view of above, the invention provides a kind of speakerphone apparatus, even if also can reach full-duplex communication when it transmission lag occurs in the voice that are sent to the other side, even and in the time that the other side uses receiver, also can prevent via receiver feel disconnect sensation.

According to an aspect of the present invention, provide a kind of speakerphone apparatus, it comprises: microphone and loud speaker; Voice switch, in order to switch between send and receive under communications status in the following manner: provide loss in receiver side signal path and transmitted signal path, reception signal from the other side's communication terminal is sent to described loud speaker via receiver side signal path, and the transmitted signal of being collected by microphone is sent to the other side's communication terminal via described transmitted signal path; Echo Canceller, the acoustic echo causing for suppressing the acoustics coupling between microphone and loud speaker; And double talk detection device, for detection of dual end communication.

Described voice switch comprises the transmitter side loss plug-in unit for loss is provided in transmitted signal path; For the receiver side loss plug-in unit of loss is provided at receiver side signal path; And for being controlled at the insertion loss amount control unit of the corresponding waste that transmitter side and receiver side provide.

In addition, described insertion loss amount control unit comprises: total losses amount calculator, its lead-out terminal from described receiver side loss plug-in unit in order to estimated path feeds back to the acoustic feedback gain of the input terminal of described transmitter side loss plug-in unit via acoustic echo path, and estimation based on to described acoustic feedback gain and the summation providing to the waste of feedback path is provided; And insertion loss amount distribution processor, it is in order to monitor described transmitted signal and described reception signal to estimate communications status, and the communications status based on estimating and the summation of being calculated by total losses amount calculator, and determine the corresponding insertion loss amount of distributing to transmitter side loss plug-in unit and receiver side loss plug-in unit.

In addition, total losses amount calculator has renewal pattern and fixed mode, in new model more, by based on the corresponding estimation of feedback oscillator the total losses amount providing in feedback path is provided and is upgraded adaptively total losses amount, and in fixed mode, total losses amount is fixed on initial value.Therefore, in the time that Echo Canceller is not fully restrained after communication starts, total losses amount calculator operates with fixed mode, and after Echo Canceller Complete Convergence, total losses amount calculator is with more new model operation.

Described speakerphone apparatus also comprises echo suppressor, echo suppressor in order to by conjunction with voice switch and/or double talk detection device to provide particular decay amount to make residual echo decay in transmitted signal path.

By above configuration, even if there is transmission lag in the voice that send between multiple communicators, echo suppressor also makes the residual echo decay occurring because of transmission lag in transmitted signal path effectively, reaches thus full-duplex communication.

If insertion loss amount distribution processor estimates that voice switch is in accepting state, echo suppressor can insert particular decay amount in transmitted signal path, otherwise does not insert particular decay amount.

Therefore, the inflection (inflection) that can suppress to occur in voice, and after communication, only make the echo attenuation of making us uncomfortable.Therefore, can reach pleasant communication.

In the time that double talk detection device is just detecting dual end communication, echo suppressor can not insert attenuation in transmitted signal path, otherwise inserts attenuation.

Therefore, can make because the inflection that the error fade that is about to the voice that send is occurred in voice is suppressed, and only make the echo of making us uncomfortable be decayed in the time of communication.Therefore, can reach pleasant communication.

If double talk detection device does not detect that dual end communication and insertion loss amount distribution processor estimation voice switch are in accepting state, echo suppressor can insert above-mentioned particular decay amount in transmitted signal path.

By operation like this, can prevent that echo suppressor from not providing mistakenly attenuation and making us uncomfortable residual echo; And therefore only make exactly the echo of making us uncomfortable be decayed in communication period.

Preferably, in the time that Echo Canceller is not fully restrained after communication starts, only, in the situation that insertion loss amount distribution processor is estimated voice switch in accepting state, echo suppressor just inserts above-mentioned particular decay amount in transmitted signal path.

Generally speaking,, in the time that Echo Canceller is not fully restrained, double talk detection device cannot detect dual end communication.Therefore, if insertion loss amount distribution processor estimates that voice switch is in accepting state, how no matter echo suppressor all insert above-mentioned particular decay amount from the testing result of double talk detection device in transmitted signal path, thereby the echo of making us uncomfortable is decayed when Echo Canceller is not fully restrained after communication starts.

In the time that echo suppressor just inserts above-mentioned particular decay amount in transmitted signal path, if insertion loss amount distribution processor estimation voice switch does not detect dual end communication in accepting state or double talk detection device, echo suppressor can stop attenuation being inserted in transmitted signal path immediately.

By operation like this, can prevent from, because of echo suppressor, inflection is occurred in voice the error fade of transmitted signal to.

Echo Canceller also can comprise: sef-adapting filter, for identifying adaptively the characteristic of echo path; Subtracter, for deducting the output of sef-adapting filter near end signal; And double talk detection device, for detection of dual end communication.In addition,, in the time that double talk detection device detects dual end communication, Echo Canceller can not upgrade the coefficient of sef-adapting filter.

By this configuration, can reach the simplification of configuration and the reduction of cost with the double talk detection device providing in Echo Canceller.

Preferably, in transmitted signal path, carry out in order the respective handling in Echo Canceller, voice switch and echo suppressor.

By this configuration, because echo suppressor is placed in after voice switch in transmitted signal path, so even if echo suppressor makes transmitted signal decay mistakenly, also can prevent from, in voice switch, inflection occurs, thereby reach pleasant communication environment.

Echo suppressor can the acoustic feedback based on being estimated by total losses amount calculator gain to adjust attenuation.

By this configuration, even if the voice that send from near-end loud speaker are decayed mistakenly by echo suppressor, also can suppress the generation of voice inflection.

Echo suppressor can be adjusted attenuation by the signal level mean value based on receiving signal.

By this configuration, even if the voice that send from near-end loud speaker are decayed mistakenly by echo suppressor, also can suppress the generation of voice inflection.

In the time receiving the signal level of signal lower than threshold value, echo suppressor can be fixed as standard value by the attenuation of inserting in transmitted signal path, and be threshold value or when higher, echo suppressor is adjusted attenuation in the upper limit from higher than standard value to the scope of standard value when receiving the signal level of signal.

By this configuration, can suppress the generation of voice inflection.

Echo Canceller also can comprise: signal level mean value (signal level average, SLA) arithmetic element, for calculating the signal level mean value that receives signal; Sef-adapting filter, for identifying adaptively the characteristic of echo path; And subtracter, for deduct the output of sef-adapting filter near end signal.In addition, echo suppressor can be adjusted attenuation by the signal level mean value based on being calculated by signal level mean value arithmetic element.

This makes to reach the simplification of configuration and the reduction of cost.

Only in the situation that double talk detection device does not detect that dual end communication and insertion loss amount distribution processor are estimated voice switch in accepting state, echo suppressor can insert particular decay amount in transmitted signal path, and when insertion loss amount distribution processor estimate voice switchs neither in accepting state again not in the time sending state, or estimate that when insertion loss amount distribution processor voice switchs is in the time that accepting state and double talk detection device detect dual end communication, echo suppressor reduces the attenuation of inserting in this transmitted signal path monotonously.

By this configuration, even if there is transmission lag in the voice of communicator that are sent to the other side, echo suppressor also can make the residual echo decay occurring because of transmission lag in transmitted signal path effectively, reaches thus full-duplex communication.In addition,, in the time not sending a transmitted signal from speakerphone apparatus to the communicator that uses receiver, echo suppressor reduces the attenuation of inserting in transmitted signal path monotonously.This speaker who makes to prevent the communicator place that uses receiver feels because of the ambient noise disconnection not feeling well that temporarily do not decayed.

When the communications status of being estimated by insertion loss amount distribution processor in the time that attenuation reduces just monotonously becomes transmission state, echo suppressor can make attenuation be reduced to zero immediately.

By operation like this, can prevent that the voice (transmission voice) that send from near-end loud speaker from being decayed mistakenly by echo suppressor and preventing from, voice that the other side's loud speaker hears, inflection occurs.

In the time that the attenuation of inserting in transmitted signal path reduces just monotonously,, in the time that the communications status that double talk detection device does not detect dual end communication and estimated by insertion loss amount distribution processor is accepting state, echo suppressor can insert particular decay amount again in transmitted signal path.

By this configuration, can prevent that echo suppressor from not providing mistakenly attenuation and making us uncomfortable residual echo; And only make exactly the echo of making us uncomfortable be decayed in communication period.

When the communications status of being estimated by insertion loss amount distribution processor neither accepting state while not being transmission state again, or when communications status be accepting state and double talk detection device when dual end communication detected, the reduction ratio that echo suppressor can be constant reduces the attenuation of inserting in transmitted signal path.

In the case, for example, when with mode progressively and compared with reducing, attenuation was reduced to for zero required time period (changing (transition) time period) and can relatively shortens.

In the time that double talk detection device does not detect that dual end communication and the communications status by the estimation of insertion loss amount distribution processor are accepting state, echo suppressor can insert particular decay amount in transmitted signal path, echo suppressor is estimated the level of the ambient noise superposeing in this transmitted signal, and based on estimate ambient noise level and in the time that ambient noise level uprises, reduce the attenuation of inserting in this transmitted signal path.

By this configuration, even if there is transmission lag in the voice of communicator that are sent to the other side, echo suppressor also can make the residual echo decay occurring because of transmission lag in transmitted signal path effectively, reaches thus full-duplex communication.In addition, owing to uprising with noise level, echo suppressor reduces the attenuation of inserting in transmitted signal path, thus can suppress the instantaneous decay of noise, thereby make to be difficult to feel the sensation of disconnection.In the case, if echo suppressor reduces attenuation, the attenuating of residual echo reduces.But in the time that reality is used, residual echo is hidden by ambient noise, and in communication, can not cause trouble.

Echo suppressor can by the echo amount of estimating according to received signal level with compare according to the attenuation of the ambient noise level superposeing in transmitted signal, and in transmitted signal path, insert less one of numerical value in echo amount and attenuation.

This makes to make fully residual echo decay, and prevents from sensing the sensation of disconnection simultaneously.

In the time that ambient noise level exceeds particular level, echo suppressor can be adjusted into steady state value by attenuation based on noise level.

By this configuration, can in the time that ambient noise level is very high, prevent that residual echo from being decayed.

Brief description of the drawings

By the description of preferred embodiments providing below in conjunction with accompanying drawing, goals and features of the present invention will become apparent, wherein:

Fig. 1 shows the block diagram according to the first embodiment of the present invention;

Fig. 2 has presented the flow chart for the operation of the voice switch shown in key-drawing 1;

Fig. 3 represents the flow chart for the operation of the echo suppressor shown in key-drawing 1;

Fig. 4 shows block diagram according to a second embodiment of the present invention;

Fig. 5 is the flow chart for the echo suppressor shown in key-drawing 4;

Fig. 6 shows the block diagram of a third embodiment in accordance with the invention;

Fig. 7 has presented the flow chart of the operation for explaining echo suppressor illustrated in fig. 6;

Fig. 8 is the flow chart of the operation of the echo suppressor of the amendment example for explaining the 3rd embodiment;

Fig. 9 shows the block diagram of a fourth embodiment in accordance with the invention;

Figure 10 shows the flow chart of the operation of the echo suppressor for explaining the 4th embodiment;

Figure 11 has presented the flow chart of the operation for explaining echo suppressor according to a fifth embodiment of the invention;

Figure 12 shows the attenuation of echo suppressor time to time change in the fifth embodiment of the present invention;

Figure 13 represents the flow chart of the operation for explaining echo suppressor according to a sixth embodiment of the invention;

Figure 14 is the flow chart of the operation of the echo suppressor for explaining the 6th embodiment; And

Figure 15 shows the attenuation of echo suppressor time to time change in the 6th embodiment.

Embodiment

Hereinafter, illustrate in greater detail with reference to the accompanying drawings embodiments of the invention, accompanying drawing forms a part herein.

The first embodiment

With reference to figure 1, the speakerphone apparatus of the present embodiment comprises: microphone 1; Loud speaker 2; Voice switch 10, voice switch 10 by changing in communications status receiving insertion loss in signal path (being illustrated by dotted line A in Fig. 1) and transmitted signal path (being illustrated by dotted line B in Fig. 1) between send and receive, reception signal from the other side's communication terminal is sent to loud speaker 2 via receiving signal path, and the transmitted signal of being collected by microphone 1 is sent to the other side's communication terminal via transmitted signal path; Echo Canceller 20, it suppresses by the acoustic echo causing that is coupled of the acoustics between microphone 1 and loud speaker 2; And amplifier of microphone G1, for amplifying the output signal (transmitted signal) of microphone 1; Speaker amplifier G2, for amplifying the input signal (reception signal) of microphone 2.

In addition, described speakerphone apparatus comprises: amplifier G3, is inserted in the traffic volume between Echo Canceller 20 and voice switch 10 for being adjusted at transmitted signal path; And echo suppressor 30, it by inserting the particular decay amount remaining echo of decaying in transmitted signal path.

Described Echo Canceller 20 has well-known sef-adapting filter 21 and subtracter 22.Echo Canceller 20 is identified by the impulse response of feedback path (acoustic echo path) Hac forming that is coupled of the acoustics between loud speaker 2 and microphone 1 adaptively by sef-adapting filter 21, and suppresses the echo components (acoustic echo) estimated according to reference signal (to the input signal of speaker amplifier G2) by deduct the acoustic echo of estimation from the output signal of amplifier of microphone G1 with subtracter 22.

Voice switch 10 comprises: transmitted signal attenuator 11, at transmitted signal path insertion loss; Receiving signal attenuation device 12, for receiving signal path insertion loss; And insertion loss amount controller 13, for controlling the waste of being inserted by corresponding transmitted signal attenuator 11 and receiving signal attenuation device 12.

Insertion loss amount controller 13 has total losses amount calculator 14, its estimate feed back to from the output point Rout of receiving signal attenuation device 12 via acoustic echo path Hac transmitted signal attenuator 11 input point Tin path (hereinafter, be called acoustic feedback path) the acoustic feedback gain alpha of C, and estimation α ' based on to acoustic feedback gain alpha calculates the summation summation of the insertion loss amount of receiving signal attenuation device 12 (the transmitted signal attenuator 11 with) of the waste being inserted into.

In addition, insertion loss amount controller 13 comprises insertion loss amount (ILA) distribution processor 15, its by monitoring transmitted signal and receive signal estimate communications status, and based on estimate communications status and the total losses amount of calculating determine the distribution for the corresponding insertion loss amount of transmitted signal attenuator 11 and receiving signal attenuation device 12.

In addition, control hardware (for example, digital signal processor (DSP)) be can pass through, for example, Echo Canceller 20 and voice switch 10 realized by the software program of Echo Canceller and voice switch (, for).Therefore, by unshowned A/D converter (analog-to-digital converter) with the particular sample cycle to inputing to voice switch 10 and Echo Canceller 20 and sampling and quantize from the signal of its output (transmitted signal and receive signal).

The corresponding average power of the signal of total losses amount calculator 14 by estimating to input to the signal of transmitted signal attenuator 11 with rectifier, low pass filter etc. and export from receiving signal attenuation device 12 short cycle.Total losses amount calculator 14 also obtains the average power of the output signal of attenuator 12 estimation to received signal under the maximum delay time of expecting in acoustic echo path Hac.Subsequently, total losses amount calculator 14 will be set a value for to the estimation α ' of acoustic feedback gain, and described value obtains in the following manner: the average power estimation with the average power of the output signal of attenuator 12 estimation to received signal under maximum delay time divided by the input signal of transmitted signal attenuator 11.

In addition, total losses amount calculator 14 is by using estimation α ' to acoustic feedback gain alpha and required gain margin (gain margin) MG to calculate total losses amount Lt, and exports total losses amount Lt to insertion loss amount distribution processor 15.

Insertion loss amount distribution processor 15 is monitored the corresponding input/output signal of transmitted signal attenuator 11 and receiving signal attenuation device 12, and communications status (accepting state, transmission state) is determined in magnitude relationship in power level based on these input/output signals and the existence of voice signal.In addition, insertion loss amount distribution processor 15 is adjusted the insertion loss amount of respective attenuation device 11 and 12, to can total losses amount Lt be dispensed to transmitted signal attenuator 11 and receiving signal attenuation device 12 pro rata based on determined communications status.

In the present embodiment, total losses amount calculator 14 has more new model and fixed mode, and in new model more, the estimation α ' based on to acoustic feedback gain alpha calculates and upgrade adaptively the summation of the waste being inserted into, in fixed mode, total losses amount is fixed on an initial value.After communication starts, before Echo Canceller 20 Complete Convergences, total losses amount calculator 14 operates in fixed mode, and subsequently, after Echo Canceller Complete Convergence, total losses amount calculator 14 is more operating in new model.

Also be, once in the special time period after communication starts (for example, hundreds of millisecond or more of a specified duration) to remain on threshold epsilon following (for example for the value of estimation α ' of acoustic feedback gain alpha, the value of the little 10dB to 15dB of estimation α ' while beginning than communication), total losses amount calculator 14 just thinks that Echo Canceller 20 fully restrains, and switch to after this more operating in new model, and until now, it operates always in fixed mode.In addition,, in fixed mode, initial value is set to the value that is fully greater than the total losses amount of upgrading by occasion needs in new model more.

Therefore, although Echo Canceller 20 is not fully convergence immediately after communication starts, but total losses amount calculator 14 operates in fixed mode, and total losses amount is set for to the initial value with fully large value, and therefore can inhibition make us uncomfortable echo (acoustic echo) or the generation of whistle, and reach stable half-duplex operation.

In addition, when Echo Canceller 20 from the communication time started through after the reasonable time when Complete Convergence, the operator scheme of total losses amount calculator 14 changes over more new model from fixed mode, and the total losses amount being inserted into is reduced to the fully value lower than initial value, and therefore can reach full-duplex communication.

Hereinafter, explain the concrete operations of total losses amount calculator 14 in new model more with reference to Fig. 2.

Total losses amount calculator 14 is carried out the estimation processing to acoustic feedback gain alpha within the particular sample time period changing over from fixed mode the time point of new model more, and calculates the estimation α ' (step 1) to acoustic feedback gain alpha.According to estimating α ' and gain margin MG, by calculate required total losses amount Lr (n) (step 2) with following formula, to can maintain the gain margin MG[dB of closed-loop]:

Lr(n)＝20log|α′(n)|+MG[dB]

Herein, α ' (n) and Lr (n) be respectively estimation α ' to acoustic feedback gain alpha and required total losses amount, it is to change over by fixed mode the n time sampling that time of new model more lights to calculate.In addition, determine and when the actual previous total losses amount Lt (n-1) inserting, total losses amount calculator 14 is by fractional increments Δ i[dB when required total losses amount Lr (n) is greater than in (n-1) inferior sampling place] be added into previous total losses amount Lt (n-1) (step 3 and 4).Also, total losses amount calculator 14 is set total losses amount Lt (n) for Lt (n-1)+Δ i.

In addition, in the time that desired total losses amount Lr (n) is less than previous total losses amount Lt (n-1), total losses amount calculator 14 deducts small decrement Δ d[dB from previous total losses amount Lt (n-1)], and total losses amount is set for to the result (step 5 and 6) of described subtraction.Also, Lt (n)=Lt (n-1)-Δ d.

As mentioned above, increase and reduce small value Δ i and Δ d because total losses amount calculator 14 makes respectively total losses amount, even (just as just having started communicate by letter afterwards with the other side's communication terminal) makes us uncomfortable sensation in also can eliminating the process of listening when therefore upgrading actively a coefficient (filter coefficient) in acoustic feedback gain alpha because of Echo Canceller 20 and fluctuating widely to restrain.

,,, because of packet transaction or for suppressing the communication buffer processing of shake, may in voice data transmission, postpone if carry out the digital communication such as IP communication between multiple speakerphone apparatus about as described in existing example as above.Generally speaking, known in the time occurring to postpone in the voice data transmission between multiple speakerphone apparatus, along with the quantitative change of time of delay is long, the tolerance variation of speaker to echo when communication, and therefore speaker perceives even a kind of small echo, thereby does not feel well.By set the initial value larger than the initial value of existing example for fixed mode, can prevent the echo of making us here uncomfortable.

But the amount of degradation of echo tolerance may reach tens of decibels sometimes, this more cannot covered by the amount of suppression of Echo Canceller under new model completely.In the case, in half-duplex but not executive communication under full duplex.In addition, in the time that noise level is higher, may there is following problem: wherein only a side joint is logical in transmitter side and receiver side for the continuation of voice switch, and therefore, is not sent to opposite side from the voice of a side.

In this regard, in the present embodiment, in transmitted signal path, provide extraly echo suppressor 30 below at Echo Canceller 20 and voice switch 10.Echo suppressor 30 makes acoustic echo (hereinafter, being called " the residual echo ") decay that should have been suppressed by Echo Canceller 20.Echo suppressor 30 need to make effectively sending voice signal (transmitted signal) residual echo decay in not adding decay.

Echo suppressor 30 is inserted into an attenuation in transmitted signal path in conjunction with voice switch 10, and more specifically, diagramatic way in flow chart as shown in Figure 3 and operating.Also, echo suppressor 30 continues the state (estimated result (accepting state or transmission state) of insertion loss amount distribution processor 15 to communications status) of monitoring (step 11) voice switch 10.In the time that voice switch 10 is in accepting state, echo suppressor 30 thinks and is sent to transmitted signal path without any voice signal, and by the signal of inputting from voice switch 10 make its decay and exported (step 12) with particular decay multiplication.

In addition, in the time that voice switch 10 is in transmission state, echo suppressor 30 think without any residual echo wait to eliminate or any voice signal (transmitted signal) to be sent, and the signal of inputting from voice switch 10 is not added damply in statu quo output (step 13).

By according to the first embodiment of the present invention, even if there is transmission lag in the voice that send between multiple communicators, echo suppressor also makes the residual echo decay occurring because of transmission lag in transmitted signal path effectively, reaches thus full-duplex communication.

This point, when echo suppressor 30 is in the time that voice switch 10 is for example in transmission state but not inserts an attenuation when accepting state in transmitted signal path, the voice that send from speaker's (in the present embodiment for using the speaker of speakerphone apparatus communication) of proximal lateral are decayed, this is undesirable, and may be heard by the speaker at the other side's communicator place and as voice inflection, wherein the volume step-down of voice or uprise.But, in the present embodiment, in the time that voice switch 10 is in accepting state, echo suppressor 30 inserts an attenuation in transmitted signal path, and in the time that voice switch 10 is in transmission state, echo suppressor 30 does not insert attenuation in transmitted signal path.Therefore, can only make the echo (residual echo) of making us uncomfortable decay, and voice inflection can not occur.

In addition, if echo suppressor 30 is placed in transmitted signal path at voice switch 10 above and make mistakenly transmitted signal decay, because of transmitted signal is decayed mistakenly by echo suppressor 30 may be in the inflection of the interior generation voice of voice switch 10.But, in the present embodiment, because echo suppressor 30 is to be configured in after Echo Canceller 20 and voice switch 10 in transmitted signal path, and corresponding processing is carried out in order, even if so echo suppressor 30 makes transmitted signal decay mistakenly, also can suppress the generation of voice inflection in voice switch 10.Therefore, can be communication pleasant environment is provided.

Meanwhile, the inventor has proposed a kind of speakerphone apparatus, wherein in transmitted signal path, after voice switch, is provided for the attenuator (seeing Japanese Patent Application Publication No.2003-324371) of insertion loss.But, in this file, only in the time that the temporary transient power of transmitted signal is estimated to remain on below a threshold value in special time period, the loss that attenuator just inserts specified quantitative in transmitted signal path.Therefore, if echo has the large volume higher than described threshold value, not insertion loss of attenuator, and the echo of making us uncomfortable may be sent to the other side.

In addition, even if the power level of the voice that send from the speaker of proximal lateral lower than described threshold value, attenuator also may unexpectedly make voice signal decay because of insertion loss.But according in the speakerphone apparatus of embodiment, echo suppressor 30 operates to insert an attenuation in conjunction with voice switch 10, prevents that thus above situation from occurring.

The second embodiment

Fig. 4 schematically shows the block diagram of speakerphone apparatus according to a second embodiment of the present invention.The speakerphone apparatus of the second embodiment is different from the speakerphone apparatus of the first embodiment, because echo suppressor 30 provides an attenuation in conjunction with Echo Canceller 20 in transmitted signal path, and other side is identical with the aspect in the first embodiment.Therefore, same components is represented by same numeral, and will omit its description.

In this embodiment, except sef-adapting filter 21 and subtracter 22, Echo Canceller 20 comprises double talk detection device 23.The signal of double talk detection device 23 based on whether comprise a certain level of sef-adapting filter 21 from the signal of amplifier of microphone G1 output disturbs convergence, detect dual end communication,, near-end speaker and far-end speaker are in fact simultaneously to the state of speaking each other.In the time that double talk detection device 23 detects dual end communication, sef-adapting filter 21 does not upgrade filter coefficient, and keeps previous value.

Next, explain the operation of echo suppressor 30 with reference to the flow chart shown in Fig. 5.Echo suppressor 30 continues the double talk detection device 23 (step 21) of monitoring Echo Canceller 20.In the time that double talk detection device 23 does not detect dual end communication, echo suppressor 30 determines and exists in residual echo to be eliminated or transmitted signal path without voice signal to be sent, and by making input signal and particular decay multiplication make its decay and be exported (step 22).

In addition, in the time that double talk detection device 23 detects dual end communication, echo suppressor 30 is definite there is voice signal to be sent, and by not making input signal and particular decay multiplication, input signal is not added to output (step 23) damply.

By the second embodiment, even if there is transmission lag in the voice that send between multiple communicators, echo suppressor also can make the residual echo decay occurring because of transmission lag in transmitted signal path effectively, reaches thus full-duplex communication.

In this embodiment, in the time that double talk detection device 23 is detecting dual end communication, echo suppressor 30 does not insert attenuation in transmitted signal path, and in the time that double talk detection device 23 does not detect dual end communication, echo suppressor 30 inserts attenuation in transmitted signal path.Therefore, can prevent that the voice that send from the speaker of proximal lateral from being decayed mistakenly, and prevent from occurring in voice inflection.In addition,, due to the echo attenuation that can make to make us uncomfortable, provide pleasant environment so can be communication in the time communicating by letter.

In addition, due to the double talk detection module (double talk detection device 23) providing in Echo Canceller 20 can be provided in this embodiment, for example, so (can reach the simplification of configuration, lower specification in minimizing, the DSP of the procedure quantity of carrying out in DSP, for minimizing of stored program ROM capacity etc.), and the minimizing of reaching cost.

The 3rd embodiment

Fig. 6 shows the block diagram of the speakerphone apparatus of a third embodiment in accordance with the invention.The basic configuration of the present embodiment is identical with the first embodiment and the second embodiment, and difference is, echo suppressor 30 in conjunction with voice switch 10 and Echo Canceller 20 both and in transmitted signal path, insert an attenuation.Therefore, same components is represented by same numeral, and will omit its description.

Hereinafter, explain the operation of echo suppressor 30 with reference to Fig. 7.Echo suppressor 30 is monitored the state (step 31) of voice switch 10 always, and in the time that voice switch 10 is in accepting state, echo suppressor 30 determines whether double talk detection device 23 detects dual end communication (step 32).In the time that double talk detection device does not detect dual end communication, echo suppressor 30 thinks and has residual echo to be eliminated or there is no voice signal to be sent, and by making input signal and particular decay multiplication make its decay and be exported (step 33).

In the time that voice switch 10 is not in accepting state, or in the time that voice switch 10 is in accepting state in the time that double talk detection device 23 detects dual end communication, echo suppressor 30 is determined without residual echo or to be sent without voice signal, and input signal is not added damply in statu quo output (step 34).

By this embodiment, can prevent that the voice that send from the speaker of proximal lateral from being decayed mistakenly, and prevent from occurring in voice inflection.In addition, echo suppressor 30 can only make the echo attenuation of making us uncomfortable, and does not make us uncomfortable residual echo (this is because echo suppressor 30 does not unexpectedly insert attenuation), is provided for thus the pleasant environment of communication.

Herein, echo suppressor 30 can be adjusted the attenuation of inserting in this transmitted signal path by the estimation α ' based on acoustic feedback gain alpha, and the estimation α ' of acoustic feedback gain alpha is estimated by total losses amount calculator 14.Next, with reference to Fig. 8, to echo suppressor 30, the operation in the time that echo suppressor 30 is adjusted the attenuation of inserting in these transmitted signal paths describes.

Echo suppressor 30 is monitored the state (step 41) of voice switch 10 always.In the time that voice switch 10 is in accepting state, echo suppressor 30 determines whether double talk detection device 23 detects dual end communication (step 42).In the time that voice switch 10 is not in accepting state, or in the time that voice switch 10 is in accepting state in the time that double talk detection device detects dual end communication, echo suppressor 30 is thought without residual echo or to be sent without voice signal, and input signal is not added damply in statu quo output (step 47).

On the other hand, in the time that voice switch 10 is in accepting state and double talk detection device 23 and dual end communication do not detected, echo suppressor 30 is set an attenuation coefficient, and described attenuation coefficient and input signal are multiplied by mutually and reach standard value SUP_MIN (step 43).After this, what estimate when voice switch 10 is threshold value TH or when larger (step 44), the upper limit SUP_MAX (step 45) that is greater than this standard value changed into attenuation coefficient by echo suppressor 30 from standard value SUP_MIN to the estimation α ' of acoustic feedback gain alpha and the summation of total losses amount Lt (n).

In addition, when voice switch 10 estimate the estimation α ' of acoustic feedback gain alpha is less than to threshold value TH (step 44) with the summation of total losses amount Lt (n) time, echo suppressor 30 maintains standard value SUP_MIN without change using as attenuation coefficient.Finally, by making input signal and determined attenuation coefficient (standard value SUP_MIN or upper limit SUP_MAX), multiply each other makes attenuated input signal to echo suppressor 30, and Output rusults (step 46).

By this configuration, even if echo suppressor 30 makes the voice decay of sending from the speaker of proximal lateral mistakenly, also can prevent voice inflection.

Meanwhile, if speaker is in the environment that ambient noise is generally high level, the sef-adapting filter 21 of Echo Canceller 20 does not upgrade filter coefficient in longer time period after communication starts.For described reason, Echo Canceller 20 is not restrained, and double talk detection device 23 cannot correctly detect dual end communication.In the case, even if voice switch 10 is in accepting state and double talk detection device 23 does not detect dual end communication, echo suppressor 30 can not add input signal damply in statu quo output yet, and this may cause the residual echo that makes us uncomfortable.

In this regard, although Echo Canceller 20 is starting not fully convergence after communication via the other side's communication terminal, in the time that insertion loss amount distribution processor 15 estimates that voice switch 10 is in accepting state, echo suppressor 30 inserts particular decay amount in transmitted signal path.Also, echo suppressor 30 makes the residual echo decay of making us uncomfortable, and the pleasant environment for speaker-phone communication is reached in Echo Canceller 20 also insufficient convergence after communication starts thus.

On the other hand, if in the time that total losses amount calculator 14 is just operating in fixed mode (after just having started in communication) simultaneously from proximal lateral and distal side input different phonetic, voice switch 10 replaces between accepting state and transmission state.For described reason, echo suppressor 30 maloperations, and therefore in sent voice, may there is inflection, the level fluctuation of the voice that wherein sent.

In this regard, in the time that insertion loss amount distribution processor 15 estimates that voice switch 10 is not in accepting state or double talk detection device 23 detects dual end communication when just inserting particular decay amount in transmitted signal path time, echo suppressor 30 stops particular decay amount to be inserted in transmitted signal path immediately.This makes to prevent that the error fade because of transmitted signal from inflection occurring, thus for speaker-phone communication provides pleasant condition.

The 4th embodiment

Fig. 9 shows the block diagram of the speakerphone apparatus of a fourth embodiment in accordance with the invention.In this embodiment, be characterised in that the signal level mean value that echo suppressor 30 receives signal by use adjusts attenuation, and other basic configuration of the present embodiment and the 3rd embodiment's is identical.Therefore, same components is represented by same numeral, and will omit its description.

In the present embodiment, sef-adapting filter 21 is identified the impulse response of acoustic echo path Hac adaptively, and according to carrying out estimated echo component (acoustic echo) g (n) from the signal y (n) (, inputing to the reception signal of speaker amplifier G2) of distal side input.For estimated echo component g (n), by upgrade filter coefficient hp (n) with following formula:

Hp(n+1)＝hp(n)+K·F(Ey(n))

Herein, p is tap (tap) number, and n is the sampling period, and K is step-length gain, and F (n) is coefficient update function.In addition, Ey (n) is the signal level mean value of input signal (reception signal) y (n), and it is by using the following formula in signal level mean value (SLA) arithmetic element 24 providing in this sef-adapting filter 21 to calculate.Herein, SPAN for increase this input signal y (n) amplitude abs[y (n)] time period.

$\mathrm{Ey}\left(n\right)=\underset{i=0}{\overset{\mathrm{SPAN}-1}{\mathrm{\Σ}}}\mathrm{abs}\left[y(n-i)\right]$

In addition, echo components (estimation) g (n) can be by calculating with following formula:

$g\left(n\right)=\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}\mathrm{hp}\left(n\right)y(n-k)$

In the present embodiment, echo suppressor 30 is by the SLA arithmetic element picked up signal level mean value Ey (n) providing in sef-adapting filter 21.

Hereinafter, explain the operation of echo suppressor 30 with reference to the flow chart shown in Figure 10.

Echo suppressor 30 is monitored the state (step 51) of voice switch 10 always.In the time that voice switch 10 is in accepting state, echo suppressor 30 determines whether the double talk detection device 23 of Echo Canceller 20 detects dual end communication (step 52).In the time that voice switch 10 is not in accepting state, or in the time that double talk detection device 23 detects that dual end communication and voice switch 10 are in accepting state, echo suppressor 30 is thought without residual echo or to be sent without voice signal, and input signal is not added damply in statu quo output (step 58).

On the other hand, in the time that voice switch 10 is in accepting state and double talk detection device 23 and dual end communication do not detected, depended on the signal level mean value RS_AVE (n) (=Ey (n)) being calculated by SLA arithmetic element 24 by echo suppressor 30 attenuation coefficient multiplying each other with input signal that carry out.More specifically, in the time of RS_AVE (n) > TH, can determine attenuation coefficient according to following formula (1), and can determine attenuation coefficient according to following formula (2) in the time of RS_AVE (n) < TH:

Herein, SUP_MAX and SUP_MIN are respectively and corresponding to loud (loud voice) (for example suppress, OOdB acoustic pressure or larger) signal level and corresponding to typical problem (for example, OO～OOdB acoustic pressure) the needed attenuation coefficient of signal level, and RS_AVEmax is loud signal level mean value RS_AVE (n).About formula (2), in the time of each sampling period RS_AVE (n)≤TH, attenuation coefficient is multiplied by special transition coefficient (transition coefficient) and is decreased to gradually SUP_MIN from SUP_MAX × RS_AVE (n)/RS_AVEmax, and wherein SUP_MIN is set as the lower limit of attenuation coefficient.

More specifically, first echo suppressor 30 determines whether attenuation coefficient is set as SUP_MIN (step 53).In the time that attenuation coefficient is not set as SUP_MIN, echo suppressor 30 makes it and changes multiplication, sets thus new attenuation coefficient (step 54).Otherwise, SUP_MIN being used as to attenuation coefficient, this is because it is the lower limit of attenuation coefficient.

Next, if RS_AVE (n) > is TH, also meet the condition (step 55) of formula (1), attenuation coefficient is set as SUP_MAX × RS_AVE (n)/RS_AVEmax (step 56) by echo suppressor 30, otherwise, use the attenuation coefficient of setting in previous steps and do not change attenuation coefficient.Finally, echo suppressor 30 makes attenuated input signal so that output (step 57) by the attenuation coefficient of input signal and setting is multiplied each other.

By the present embodiment, even if echo suppressor 30 unexpectedly makes the voice decay of sending from the speaker of proximal lateral, also can suppress the generation of voice inflection.In addition, because the signal level mean value arithmetic element 24 by with providing in Echo Canceller 20 is calculated signal level mean value RS_AVE (n), for example, so (can reach the simplification of configuration, lower specification in minimizing, the DSP of the procedure quantity of carrying out in DSP, for minimizing of stored program ROM capacity etc.), and cost reduction.

The 5th embodiment

Next, explain the fifth embodiment of the present invention with reference to accompanying drawing.In the present embodiment, echo suppressor 30 inserts attenuation in conjunction with signal level mean value (SLA) arithmetic element 24 providing in the double talk detection device 23 of voice switch 10, Echo Canceller 20 and the sef-adapting filter 21 of Echo Canceller 20 in transmitted signal path.Other configuration of the present embodiment is identical with the 4th embodiment.Therefore, same components is represented by same numeral, and will omit its description.

Explain in detail the operation of echo suppressor 30 with reference to the flow chart shown in Figure 11.Step 61 in Figure 11 is to 67 identical to step 57 with the step 51 about the 4th embodiment in Figure 10.Therefore, will omit its description.

In the time that voice switch 10 is not in accepting state, or in the time that voice switch 10 is in accepting state in the time that double talk detection device 23 detects dual end communication, echo suppressor 30 further determines whether voice switch 10 is in transmission state (step 68).In the time that voice switch 10 is in transmission state, the definite residual echo or to be sent without voice signal that do not exist of echo suppressor 30, and input signal is not added damply in statu quo output (step 611).

If speakerphone apparatus according to the present invention with use receiver communicator (hereinafter, be called " receiver communicator ") between executive communication, and the level of ambient noise is quite high the transmitted signal that is sent to receiver communicator from speakerphone apparatus, when the echo suppressor 30 in speakerphone apparatus makes residual echo decay, ambient noise temporarily decays.This decay of ambient noise heard by receiver through a delay, and therefore makes the speaker at receiver communicator place feel the uncomfortable disconnection sensation of making us in voice in communication period.

In this regard, in the present embodiment, in the time that in step 68, voice switch 10 is not in transmission state (for example, insertion loss amount distribution processor 15 is inserted same loss amount (hereinafter in transmitted signal attenuator 11 and receiving signal attenuation device 12, this state is called to " intermediateness ")), echo suppressor 30 makes the attenuation decay of inserting in transmitted signal path monotonously.

More specifically, in the time that in step 68, voice switch 10 is not in transmission state, echo suppressor 30 further determines whether attenuation coefficient is zero (step 69).If attenuation coefficient is zero, processes and be back to step 61, and do not change attenuation coefficient.Otherwise echo suppressor 30 is by making attenuation coefficient and transformation multiplication set new attenuation coefficient (step 610).Also, echo suppressor 30 makes attenuation coefficient (attenuation) reduce gradually (as shown in the line A in Figure 12, reducing monotonously) by the processing of repeating step 61 or 62-> step 68-> step 69-> step 610-> step 61.

In the present embodiment, because making the attenuation of inserting in transmitted signal path, echo suppressor 30 reduces monotonously, so ambient noise can temporarily not decay.This makes to be difficult to make the speaker at receiver communicator place to feel the disconnection sensation of making us uncomfortable.

Simultaneously, when in the time that attenuation coefficient (attenuation) reduces just monotonously by the processing of repeating step 61 or 62-> step 68-> step 69-> step 610-> step 61, the state changes of voice switch 10 is transmission state, echo suppressor 30 makes attenuation coefficient (attenuation) be reduced to zero (for example, referring to, the dotted line B in Figure 12) immediately.Therefore, in the voice that can prevent from hearing the speaker at the other side's communicator place, inflection occurs, this is because the voice (transmission voice) that send from the speaker of speakerphone apparatus are unexpectedly decayed by echo suppressor 30.

In addition, in the time that attenuation coefficient (attenuation) reduces just monotonously by the processing of repeating step 61 or 62-> step 68-> step 69-> step 610-> step 61, the state changes of voice switch 10 is that transmission state and double talk detection device 23 be not when dual end communication detected, processing by execution step 63 to 67 of echo suppressor 30 (referring to, for example dotted line C in Figure 12) and in transmitted signal path, insert a particular decay amount.This makes to prevent that echo suppressor 30 from unexpectedly not inserting an attenuation and preventing the residual echo that makes us uncomfortable, and only makes in communication period the echo attenuation of making us uncomfortable exactly.

In the present embodiment, to make time ratio (reducing ratio) that attenuation reduces be monotonously constant (for example, referring to, the line A in Figure 12) to echo suppressor 30.In the time of attenuation linear attenuation described above, compared with being decreased to for zero required time period (fringe time section) and progressively making the situation of attenuation decay, can shorten.

The 6th embodiment

Next, explain according to a sixth embodiment of the invention with reference to Figure 13 to Figure 15.As shown in figure 13, the difference of the 6th embodiment and the 5th embodiment is, between step 77 and 78, provides the step 712 of controlling attenuation based on ambient noise level.Step 71 to 79 and step 710 and 711 with the step 61 illustrating in the 5th embodiment to 69 and step 610 and 611 identical, and will omit its description.

In the present embodiment, echo suppressor 30 is controlled attenuation further combined with ambient noise level.In the flow chart shown in Figure 14, illustrate that combining environmental noise level controls the processing of attenuation.

Also,, along with the ambient noise level of speakerphone apparatus one side uprises, the attenuation that echo suppressor 30 inserts in transmitted signal path reduces.Therefore, can suppress the instantaneous decay of ambient noise, make to be thus difficult to feel the sensation disconnecting.In the case, if echo suppressor reduces attenuation, the attenuating of residual echo reduces.But in the time that reality is used, residual echo is hidden by ambient noise, and in communication, can not cause trouble.

Hereinafter, with reference to the flow chart shown in Figure 14, description taken in conjunction ambient noise level is controlled the processing of attenuation.

Echo suppressor detects (estimation) level (ambient noise level) LEVEL_NOISE of the ambient noise of continued presence in transmitted signal by using by the long-term average of the transmitted signal of the input point Tin input of transmitted signal attenuator 11, and by estimated ambient noise level LEVEL_NOISE and threshold value Nth comparison (step 821).When ambient noise level LEVEL_NOISE is threshold value Nth or more hour, echo suppressor 30 is by calculating standard attenuation coefficient ATT (step 822) with following formula:

ATT＝α×LEVEL_NOISE+β，α＜0

In addition,, in the time that ambient noise level LEVEL_NOISE exceeds threshold value Nth, how echo suppressor 30 ambient noise level LEVEL_NOISE are set as steady state value ATTmin (step 823) by standard attenuation coefficient ATT.

Subsequently, echo suppressor 30 compares (step 824) by the attenuation coefficient calculating to 76 processing by step 73 (corresponding to the attenuation of the echo amount of estimating according to received signal level) with the standard attenuation coefficient ATT calculating by step 821 to 823 processing.When attenuation coefficient is standard attenuation coefficient ATT or more hour, echo suppressor 30 upgrades attenuation coefficient (step 825) with standard attenuation coefficient ATT, thereby advances to step 77.In addition,, in the time that attenuation coefficient is greater than standard attenuation coefficient ATT, echo suppressor 30 advances to step 77, and does not upgrade this attenuation coefficient.

Also be, in step 824 and 825, echo suppressor 30 by the attenuation of the echo amount corresponding to estimating according to receiving the speech signal level in signal path (referring to, the for example curve D in Figure 15) come with based on ambient noise level and definite attenuation (referring to, the for example curve E in Figure 15) compare, and in transmitted signal path, insert subsequently the attenuation more by a small margin that has in attenuation.In addition,, in the time that ambient noise level LEVEL_NOISE exceeds threshold value Nth, echo suppressor 30 is set as steady state value ATT based on ambient noise level LEVEL_NOISE by attenuation (attenuation coefficient).This way can prevent that in the time that ambient noise level is very high remaining echo from being decayed.

Although illustrated and described the present invention with respect to embodiment, it will be apparent to those skilled in the art that in the case of not departing from the category of the present invention defining in following claim, can make various changes and amendment.

Claims (19)

1. a speakerphone apparatus, comprising:

Microphone;

Loud speaker;

Voice switch, in order to by providing loss to switch between send and receive under communications status receiving in signal path and transmitted signal path, wherein, reception signal from the other side's communication terminal is sent to described loud speaker via described reception signal path, and the transmitted signal of being collected by described microphone is sent to described the other side's described communication terminal via described transmitted signal path;

Echo Canceller, for suppressing acoustics between described microphone and the described loud speaker acoustic echo causing that is coupled;

Double talk detection device, for detection of dual end communication; And

Echo suppressor, in order to by making residual echo decay in conjunction with described voice switch and/or described double talk detection device to insert particular decay amount in described transmitted signal path,

Wherein, described voice switch comprises: for the transmitter side loss plug-in unit at described transmitted signal path insertion loss; For the receiver side loss plug-in unit at described reception signal path insertion loss; And for controlling the insertion loss amount control unit of respective amount of the loss of being inserted by described transmitter side loss plug-in unit and described receiver side loss plug-in unit,

Described insertion loss amount control unit comprises: total losses amount calculator, feed back to the acoustic feedback gain of the input terminal of described transmitter side loss plug-in unit via acoustic echo path in order to the lead-out terminal from described receiver side loss plug-in unit of estimated path, and described estimation based on to described acoustic feedback gain and calculate the summation of the described waste of inserting in described path; And insertion loss amount distribution processor, in order to monitor described transmitted signal and described reception signal and to estimate communications status, and the result based on described estimation and the value of being calculated by described total losses amount calculator, and the distribution of definite insertion loss amount to described transmitter side loss plug-in unit and described receiver side loss plug-in unit, and

Described total losses amount calculator has renewal pattern and fixed mode, in described more new model, corresponding estimation based on to described feedback oscillator and calculate and be updated in adaptively the summation of the described waste of inserting in described path, in described fixed mode, total losses amount is fixed on initial value, and described in the time that described Echo Canceller is not fully restrained after communication starts, total losses amount calculator operates under described fixed mode, and described total losses amount calculator operates under described more new model after described Echo Canceller Complete Convergence.

2. speakerphone apparatus as claimed in claim 1, wherein, if described insertion loss amount distribution processor estimates that described voice switch is in accepting state, described echo suppressor inserts described particular decay amount in described transmitted signal path, otherwise does not insert described particular decay amount.

3. speakerphone apparatus as claimed in claim 1, wherein, in the time that described double talk detection device is just detecting dual end communication, described echo suppressor does not insert described particular decay amount in described transmitted signal path, otherwise inserts described particular decay amount.

4. speakerphone apparatus as claimed in claim 3, wherein, if described double talk detection device do not detect dual end communication and described insertion loss amount distribution processor and estimate that described voice switch is in described accepting state, described echo suppressor inserts described particular decay amount in described transmitted signal path.

5. speakerphone apparatus as claimed in claim 4, wherein, in the time that described Echo Canceller is not fully restrained after communication starts, only, in the situation that described insertion loss amount distribution processor is estimated described voice switch in described accepting state, described echo suppressor just inserts described particular decay amount in described transmitted signal path.

6. speakerphone apparatus as claimed in claim 1, wherein, insert described particular decay amount in described transmitted signal path during, if described insertion loss amount distribution processor estimates described voice switch and dual end communication do not detected in described accepting state or described double talk detection device, described echo suppressor stops described attenuation to be inserted in described transmitted signal path immediately.

7. speakerphone apparatus as claimed in claim 1, wherein, described Echo Canceller comprises: sef-adapting filter, for identifying adaptively the characteristic of echo path; Subtracter, for deducting the output of described sef-adapting filter from described transmitted signal; And described double talk detection device, for detection of dual end communication, and in the time that described double talk detection device detects dual end communication, described Echo Canceller does not upgrade the coefficient of described sef-adapting filter.

8. speakerphone apparatus as claimed in claim 1 wherein, provides in order described Echo Canceller, described voice switch and described echo suppressor, and carries out it and process accordingly in described transmitted signal path.

9. speakerphone apparatus as claimed in claim 1, wherein, the described acoustic feedback of described echo suppressor based on being estimated by described total losses amount calculator gains and adjusts described attenuation.

10. speakerphone apparatus as claimed in claim 1, wherein, the signal level mean value of described echo suppressor based on described reception signal and adjust described attenuation.

11. speakerphone apparatus as claimed in claim 10, wherein, when the described signal level of described reception signal is during lower than threshold value, the described attenuation of inserting in described transmitted signal path is set as specified standard value by described echo suppressor, and when described signal level is described threshold value or when higher, described echo suppressor is adjusted described attenuation in the upper limit from higher than described specified standard value to the scope of described specified standard value.

12. speakerphone apparatus as claimed in claim 10, wherein, described Echo Canceller comprises: signal level mean value (SLA) arithmetic element, for calculating the signal level mean value of described reception signal; Sef-adapting filter, for identifying adaptively the characteristic of described echo path; And subtracter, for deduct the output of described sef-adapting filter from described transmitted signal, and described echo suppressor is by using the described signal level mean value being calculated by described signal level mean value arithmetic element to adjust described attenuation.

13. speakerphone apparatus as claimed in claim 1, wherein, only in the situation that described double talk detection device does not detect that dual end communication and described insertion loss amount distribution processor are estimated described voice switch in described accepting state, described echo suppressor just inserts particular decay amount in described transmitted signal path, and when described insertion loss amount distribution processor estimates that described voice switch is neither in described accepting state during again not in described transmission state, or estimate that described voice switch is in the time that described accepting state and described double talk detection device detect dual end communication when described insertion loss amount distribution processor, described echo suppressor is reduced in the described attenuation of inserting in described transmitted signal path monotonously.

14. speakerphone apparatus as claimed in claim 13, wherein, in the time that the described communications status of being estimated by described insertion loss amount distribution processor in the time that described attenuation reduces just monotonously becomes described transmission state, described echo suppressor is reduced to zero by described attenuation immediately.

15. speakerphone apparatus as claimed in claim 13, wherein, in the time that the described attenuation of inserting in described transmitted signal path reduces just monotonously, in the time that described double talk detection device does not detect that dual end communication and the described communications status by described insertion loss amount distribution processor estimation are described accepting state, described echo suppressor inserts described particular decay amount again in described transmitted signal path.

16. speakerphone apparatus as described in any one in claim 13-15, wherein, when the described communications status of being estimated by described insertion loss amount distribution processor is neither in described accepting state during again not in described transmission state, or when described communications status be described accepting state and described double talk detection device when dual end communication detected, described echo suppressor reduces with constant reduction ratio the described attenuation of inserting in described transmitted signal path.

17. speakerphone apparatus as claimed in claim 1, wherein, only in the time that the described communications status that described double talk detection device does not detect dual end communication and estimated by described insertion loss amount distribution processor is described accepting state, described echo suppressor just inserts described particular decay amount in described transmitted signal path, described echo suppressor is estimated the ambient noise level superposeing in described transmitted signal, and the estimated result based on to described ambient noise level and reduce the described attenuation of inserting in described transmitted signal path in the time that described ambient noise level uprises.

18. speakerphone apparatus as claimed in claim 17, wherein, described echo suppressor will according to the estimated echo of the described signal level of described reception signal amount with depend on that the attenuation of the ambient noise level superposeing in described transmitted signal compares, and in described transmitted signal path, insert less one of numerical value in described echo amount and described attenuation.

19. speakerphone apparatus as claimed in claim 18, wherein, in the time that described noise level exceeds particular level, described echo suppressor is adjusted into steady state value based on described ambient noise level by described attenuation.