KR0138642B1 - Echo canceller - Google Patents

Abstract

This eliminates the need for training, reduces the user's operational burden, and eliminates the need for training or backup power when the acoustics of the room change or power failure. Acoustic echo generated by setting the optimum state at all times, thereby improving the call quality and miniaturizing the circuit, and the sound output from the speaker 22 for hand-piper call is input to the microphone 23. Sound echo canceller 10 and its training control means 31 for canceling the sound output from the speaker 22 by the means 31 for holding the operation state during connection to a telephone line. Voice communication apparatus configured to train acoustic characteristics of the acoustic echo canceller according to the present invention.

Description

Voice communication device with echo canceller

1 is a circuit block diagram showing the configuration of a radiotelephone apparatus according to an embodiment of the present invention.

2 is a circuit block diagram showing the main components of the apparatus shown in FIG.

3 is a flowchart showing the reception standby control procedure and control contents of the control circuit in the apparatus shown in FIG.

4 is a flowchart showing incoming control reception and control contents of a control circuit in the apparatus shown in FIG.

FIG. 5 is a flowchart showing the transmission control procedure and the control contents of the control circuit in the apparatus shown in FIG.

6 is a circuit block diagram showing the configuration of a conventional echo canceller.

The present invention relates to a voice communication device including an echo canceller for removing echoes generated during hand free calls, such as, for example, an automobile wireless telephone device.

Conventionally, for example, some vehicle radiotelephone apparatuses, such as an automobile radiotelephone apparatus, are provided with a handfree telephone mode separately from the handset telephone mode. The hand-free call mode is provided with a speaker and a microphone for a handset separately from a handset, such as a communicator or an instrument, so that the speaker and the microphone can be used as handsets to make a call. This hand-free call mode allows the caller to make a call without lifting the handset. Therefore, the caller can call without having to drive with one hand while driving.

However, when using such a hand-free call mode, a handset sound output from a speaker may be sent back to the other party through a microphone to generate an echo during a call.

For this reason, this type of telephone apparatus is generally equipped with an echo canceller for canceling the echo. This echo canceller is indispensable, especially in a radiotelephone apparatus that digitalizes and transmits a voice signal and thus has a large signal delay due to a codec.

6 shows an example of the configuration of this kind of echo canceller. This echo canceller 10 'has an adaptive filter 11, an adder 12 and a RAM 13 for echo cancellation information storage. When a part of the sign sound output from the speaker 22 becomes an echo (EC) and is input to the microphone 23, the echo signal ES is converted into a digital signal by the A / D converter 21S and then the echo canceller 10 '), The echo signal ES becomes the residual deviation signal TS through the adder 12 and is input to the adaptive filter 11. In the adaptive filter 11, an adaptation operation for making the residual deviation signal TS zero is performed, and the adaptation operation estimates the acoustic characteristics of the communication space, that is, the echo path, and the estimation result and sign language. The pseudo echo signal EES is generated based on the signal RS. In the adder 12, arithmetic processing for subtracting the pseudo echo signal EES from the echo signal ES is performed, thereby eliminating the echo signal ES.

However, in order to operate the echo canceller 10 'effectively, the echo canceller 10' is trained prior to the call to estimate the acoustic characteristics of the call space, that is, the echo path, and create echo cancellation information according to the estimation result. The RAM 13 needs to be stored in advance. In the prior art, for example, the echo canceller 10 'was trained as follows.

That is, for example, in the case of an automobile radiotelephone apparatus, the user manipulates the learning key when the acoustic characteristics of the vehicle are changed, such as when a passenger rides more while riding.

Then, the switching switch 14 is switched by the control of the control circuit of the radiotelephone apparatus so that the learning signal generator 16 is connected to the speaker 22.

In this state, a learning signal is generated from the learning signal generator 16 and output to the speaker 22. In this state, in the echo canceller 10 ', an adaptive operation for estimating the echo path of the call space is performed by the learning sound. Then, based on the estimation result, information for removing the acoustic characteristics in the vehicle is generated, and the removal information is stored in the RAM 13.

In this way, if the learning is performed before the call, the echo canceler 10 'immediately removes the optimal echo signal ES from the beginning of the call according to the removal information stored in the RAM 13, thereby providing high quality. The call becomes possible.

However, the apparatus adopting such a conventional learning method has a problem to be solved as follows.

(1) The user must manually learn the echo canceller 10 'in advance by operating the learning key in advance, which is cumbersome to operate.

(2) Even after learning has been performed, if the number of people in the vehicle is changed later, for example, the acoustic characteristics of the passing space are changed, so that the echo signal ES is not removed. Therefore, the user must relearn the echo canceller 10 'by manipulating the learning key whenever the number of people in the vehicle is changed. For this reason, the operation is very troublesome and often forgets to learn. If you forget learning, even if not removed during the period from the start of the call to the echo canceller 10 ', the echo signal (ES) is sent to the other party as a call signal, which causes a decrease in the call quality, which is very undesirable. not.

(3) In order to prevent the removal information stored in the RAM 13 from being erased due to the power failure of the apparatus accompanying the engine stop of the vehicle, it is necessary to install a backup power source 15 in the RAM 13. . For this reason, the circuit structure of the echo canceller 10 'is complicated and large. If the backup power supply 15 is not provided, the operation is further troublesome since the user must manually learn the echo canceller 10 'in advance at the time of starting the engine of the vehicle by manually operating the learning key.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and its object is to eliminate the need for manual operation for learning at all, and to reduce the operational burden on the user, and to learn even when the acoustic characteristics of the call space are changed or the power is cut off. The present invention provides a voice communication apparatus having an echo canceller that can set the operating state of the echo canceller to the optimum state without requiring a backup power supply, thereby improving call quality and miniaturizing the circuit.

In order to achieve the above object, the present invention provides a speaker for amplifying and outputting a spoken voice signal separately from a handset call mode, and a microphone for collecting a call tone and outputting a call signal, and the sound output from the speaker is input to the microphone. And a hand-free call mode configured as an echo canceller for canceling the generated echoes, and performing a hand-free call using the speaker and the microphone, wherein the echo canceler is amplified and output to the call space through the speaker. In the voice communication apparatus for estimating the acoustic characteristics of the space in advance by using a sound echo canceller, the sound echo canceller is provided, and the first speech signal is supplied to the speaker in the first state, and the predetermined communication in the second state. Switching means for supplying the sound to the speaker, and incoming or outgoing voice to the voice communication device And a control means for controlling the switching sound switch to be supplied to the switching switch means by switching the switching means from the first state to the second state when the setting of the call path is completed with the transmission from the apparatus. It provides a voice communication device characterized in that.

In another aspect, the present invention provides a speaker for amplifying and outputting a spoken voice signal separately from a handset talk mode, a microphone for collecting a call tone and outputting a call signal, and to remove echoes caused by the sound output from the speaker being input to the microphone. And a hand-free call mode configured as an echo canceller, and conducting a hand-free call using the speaker and the microphone, wherein the echo canceler uses the notification sound amplified and output to the call space through the speaker. A voice communication apparatus for learning characteristics in advance, which is provided in the echo canceller, switches to supply the sign language audio signal to the speaker in a first state, and to supply a predetermined notification sound to the speaker in a second state. A switch means and the handset call mode is selected or a hand free call mode Judging means for judging whether it is selected, and if it is determined by the judging means that the handset call mode is selected, the call mode is changed to the hand-free call mode only for at least the learning period, and the switch switch means And a control means for switching from the state of 1 to the second state and controlling the notification sound to be supplied to the selection means.

In addition, the present invention is characterized by learning a sound characteristic of an echo canceller by configuring a notification sound such as an incoming call or a dial confirmation sound including a frequency component necessary for learning, and outputting the notification sound from a speaker.

As a result, according to the present invention, for example, the echo canceller is automatically learned by using a notification sound normally used, such as a ringing tone or a dial confirmation tone, when the call is connected with an incoming or outgoing call. As a result, the operator does not need any key operation for learning, thereby reducing the operator's operational burden.

In addition, since the learning is automatically performed for each call, the new removal information is set to the echo canceler at each time. For example, even if the configuration of the passengers is changed and the acoustic characteristics of the used space are changed, the optimum removal at the time of the call is removed. The echo canceller can be operated according to the information. For this reason, the echo is not transmitted to the call counterpart at the start of the call, thereby making it possible to always make a high quality call.

In addition, even if the cancellation information set by the power failure is erased, the echo canceller can be operated at the start of the call according to the optimum removal information. This eliminates the need for a backup power supply for the memory device for storing the removal information, and eliminates the need for expensive memory devices such as EEPROMs, making the circuit configuration simple and inexpensive.

Further, according to the present invention, even when the handset call mode is selected, the call mode is temporarily changed to the hand free call mode, and the echo canceller is learned in this state.

Therefore, for example, even when the call is switched from the handset call mode to the hand-free call mode during a call, the echo canceller can perform the echo cancellation operation according to the latest removal information, so that a high quality handfree call can be made.

In addition, instead of using the existing ring tone or dial confirmation sound as it is for learning, it is possible to obtain the optimum removal information by substituting or converting the sound into a sound including a frequency component necessary for learning.

EXAMPLE

1 is a circuit block diagram showing the configuration of an automobile radiotelephone apparatus according to an embodiment of the present invention.

In the figure, the carrier signal is received by the predetermined radio call channel at the antenna 1 and then input to the receiving circuit RX 3 via a duplexer (DUP) 2. The reception circuit 3 converts the received carrier signal into a local frequency signal synthesized with the local oscillation signal output from the frequency synthesizer (SYN) 4. This receiving intermediate frequency signal is input to a digital demodulation circuit (DEM) 6R. In the digital demodulation circuit 6R, the reception intermediate frequency signal is digitally demodulated and converted into an analog baseband signal. This baseband signal is converted into a digital signal by the A / D converter 7R, and then error correction processing is performed in the error correction code decoding circuit (CH-COD) 8, followed by the speech code decoding circuit (SP-COD) ( In 9). The digital dehydration signal outputted from the audio code decoding circuit 9 is converted into an analog signal by the D / A converter 21R after passing through the echo canceller (EC-CAN) 10 and then amplified by the speaker 22. Is output.

On the other hand, the echo sound signal collected by the microphone 23 is converted into a digital signal by the A / D converter 21S, and then the echo canceling process is performed by the echo canceller 10. FIG. The talk signal subjected to this echo cancellation processing is input to the audio code decoding circuit 9. In this speech code decoding circuit 9, the coding process of the digital transmitted signal is performed, and the signal is then subjected to error correction coding in the error correction code decoding circuit 8. The coded digital transmission signal is converted into an analog signal by the D / A converter 7S and then input to the digital modulation circuit (MOD) 6S. In this digital modulation circuit 6S, the digital modulation of the carrier is performed using the digital transmission signal as a modulation signal. As the digital modulation method, for example, the QPSK method is used. The transmission signal output from this digital modulation circuit 6S is synthesized by the local oscillation signal output from the frequency synthesizer 4 and transmitted from the transmission circuit TX 5 and converted into a signal of a radio frequency band, and then passed through the duplexer 2. It is transmitted from the antenna 1.

24 denotes a console unit (CU), which includes a switch group such as an outgoing switch and a dial key and a display unit such as a light emitting diode or a liquid crystal display for displaying the operation state of the device. 26 is a power supply circuit. For example, the power supply voltage Vcc is generated and supplied to each circuit in accordance with the output voltage of the battery 25 of the vehicle.

By the way, the echo canceller 10 is configured as follows. 2 is a circuit block diagram showing the configuration together with the pickup circuit. In addition, the same code | symbol is attached | subjected to the same part as said FIG. 6 in the same figure. The echo canceller 10 includes an adaptive filter 11, an adder 12, a RAM 13 for storing echo cancellation information, and a switching switch 14 for switching a sign signal RS.

The adaptive filter 11 takes in the residual deviation signal TS output from the adder 12 and performs an adaptive operation to make this residual deviation signal TS zero, thereby acoustic characteristics of the used space, that is, echo Estimate the path.

Then, the echo cancellation information is generated based on the estimation result of the echo path, the pseudo echo signal EES is generated based on the echo cancellation information and the sign signal RS, and the pseudo echo signal EES is added to the adder ( 12). The removal information is written to the RAM 13. The adder 12 performs digital operation to extract the pseudo echo signal EES from the echo signal ES output from the A / D converter 21S, and uses the residual deviation signal TS as a speech signal to decode the speech code. Output to the circuit (9). The switching switch 14 is constituted of, for example, a semiconductor switch, and is switched in accordance with the switching control signal SS output from the control circuit 30 to perform the voice code decoding circuit 9 for the D / A converter 21R. And control circuit 30 are selectively connected.

The control circuit (CONT) 30 is composed of a CPU, for example, having a microcomputer as a main control unit, and a normal control function such as a setting control means or a call control means for a wireless call channel associated with transmission, reception, and the like. In addition, as described above, the apparatus further includes a learning control means 31 of the echo canceller 10.

The learning control means 31 controls the switching switch 14 of the echo canceller 10 in the output period of the ringing tone and the output period of the dial confirmation tone during the connection control of the call accompanying incoming and outgoing calls. In this state, the ringing tone data and the dial confirmation tone data HD are read out from the ROM (not shown) provided in the control circuit 30, and these data HD are transferred to the D through the switch 14. It is supplied to the / A converter 21R to output ringing tones and dial confirmation tones from the speaker 22. Here, the ring tone data and the dial confirmation tone data are set to be the same as the conventional ring tone data and the dial confirmation tone data with respect to the intermittent period or the output time, but the bandwidth of the ringer effectively sets the frequency required for the learning of the echo canceller 10. It is set to be an included sound.

Next, the operation of the apparatus configured as described above will be explained by the control procedure of the control circuit 30.

In the reception standby state, the control circuit 30 of the radiotelephone apparatus repeats the arrival monitoring of the incoming data and the operation monitoring of the originating switch in steps 3a and 3b, as shown in FIG.

In this state, it is assumed that incoming data has arrived from a radiotelephone base station (not shown). At this time, the control circuit 30 returns the incoming response data in step 4a, as shown in Fig. 4, when the incoming data is for the own station and the station can be received. Then wait for the arrival of radio channel specific data from the base station in step 4b. When radio channel designation data is received, control for setting a radio call channel designated by this designation data is executed in step 4c.

Thus, when the setting of the radio call channel is completed, the control circuit 30 starts learning control of the echo canceller 10.

That is, the control circuit 30 outputs the switching control signal SS to the switching switch 14 of the echo canceller 10 in step 4d to cut the path of the sign signal RS, and instead the control circuit 30 ) (See Fig. 2). In step 4e, the ring tone data HD is read out from the ROM in the control circuit 30, and the ring tone data is supplied to the D / A converter 21R via the switching switch 14. Therefore, the ringing tone is output from the speaker 22, thereby notifying that the incoming call has occurred to the user.

At this time, part of the ringing tone is echoed to the microphone 23 and the echo signal ES is digitalized by the A / D converter 21S and then input to the echo canceller 10.

In the echo canceller 10, learning is performed in accordance with the ring tone, and the removal information of the echo path estimated by this learning is written in the RAM 13. When the optimum removal information corresponding to the echo path of the call space is stored in the RAM 13 as described above, the control circuit 30 determines that the learning of the echo canceller 10 has been completed, and then proceeds from step 4f to step 4g. Monitor off-hook operation. Next, when the user responds to the off-hook operation in this state, the control circuit 30 returns the switching switch 14 to the path of the sign signal RS in step 4h, and then transfers to call control.

Therefore, in the call state, the echo canceller 10 starts the operation for removing the echo signal EC based on the optimum echo canceller information stored in the RAM 13 by learning using the ringing tone from the control circuit 30. . Therefore, the echo generated during the call is certainly removed at the beginning of the call and is not transmitted to the call counterpart. Therefore, the caller can make a good call from the beginning of the call.

The echo canceller 10 also removes the echo signal ES while the call space continuously follows the change in the echo path and updates the echo canceller information during the call. Therefore, even if the echo path is changed by, for example, a passenger opening a car window or moving a seat during a call, it is possible to follow the change of the echo path and then perform the optimal echo canceller operation at that time.

If the user off-hooks before the end of the learning, the control circuit 30 immediately proceeds to step 4h at that point and returns the changeover switch 14 to transfer to call control. In this case, therefore, the call is started with insufficient echo canceller 10. However, although the echo canceller 10 is in an inadequate state, since the estimation of the echo path is carried out to some extent by the above learning, the learning is started by initiating the echo canceller operation for the call based on the removal information obtained by the estimation. The call can be started relatively smoothly without generating a large echo compared with the case where it is not done at all. In addition, the echo canceller 10 makes the optimum state quickly by the following operation during a call.

On the other hand, in the reception standby state, for example, if the user has operated the originating switch to call, the control circuit 30 first generates the originating data in step 5a and transmits it to the radiotelephone base station as shown in FIG. In step 5b, the transmission of outgoing response data from the base station is monitored, and when the outgoing response data is received, control for setting up a radio call channel is performed in step 5c based on the radio channel designation information included in this data.

Thus, when the setting of the radio call channel is completed, the control circuit 30 continues to start learning control of the echo canceller 10. That is, in step 5d, the switching control signal SS is output to the switching switch 14 of the echo canceller 10, whereby the sign signal path is cut off and connected to the control circuit 30 instead. In step 5e, the input of the dial key is monitored, and when the dial key is operated, the dial data corresponding to the key is generated and sent to the base station in step 5f. At this time, the control circuit 30 reads the dial confirmation tone data from the ROM installed in the control circuit 30 in step 5g, and supplies this data to the D / A converter 21R via the changeover switch 14. do. Therefore, the dial confirmation sound is output from the speaker 22, whereby the user can confirm that the dial operation is performed reliably.

At this time, a part of the dial confirmation sound is echoed to the microphone 23 and the echo signal ES is digitalized by the A / D converter 21S and then input to the echo canceller 10. . The echo canceller 10 then learns in accordance with the dial confirmation tone. The removal information of the echo path estimated by this learning is written into the RAM 13. When the optimum removal information corresponding to the echo path of the call space is stored in the RAM 13, the control circuit 30 determines that the learning is completed, and shifts from step 5h to step 5i, where the golden or switch 14 is moved. Return to the sign language signal. In this state, when the other party's caller responds by off-hook, the device enters a call operation state.

Therefore, in the call state, the echo canceller 10 initiates an operation for canceling the echo signal in accordance with the optimum removal information stored in the RAM 13 by the learning using the dial confirmation tone. For this reason, the echo (EC) generated during the call is reliably removed from the start of the call, and is not transmitted to the call counterpart. For this reason, the caller can make a call with good quality from the start of the call.

As described above, according to this embodiment, the learning of the echo canceller 10 is automatically performed by using the ringing tone and the dial confirmation tone, respectively. For this reason, the user does not need to perform any manual operation for learning at all, thus reducing the operational burden.

In addition, since the learning is automatically performed at the time of incoming and outgoing calls, the echo removal operation is always performed in accordance with the optimum removal information at that time. Therefore, during the reception waiting, for example, the number of people in the vehicle is changed, so that even if the acoustic characteristics of the call space are changed, the call can be always performed with the best quality. Also in this case, the user does not need to perform any troublesome manual operation for learning, thereby reducing the burden on the user.

In addition, even if the removal information stored in the RAM 13 is erased due to a power failure, the learning is automatically performed at the time of the call, and the optimal removal information is written into the RAM 13 at that time, so the storage information of the RAM 13 is backed up. Since there is no need to protect by using a power supply, the circuit configuration can be simplified and downsized, and the circuit can be reduced because there is no need to use expensive memory devices such as EEPROM.

In addition, according to the present embodiment, since the learning is performed using the generation period of the ringing tone and the generation period of the dial confirmation tone, there is no fear that the time required for connection to the call is unnecessarily long.

In addition, this invention is not limited to the said Example. For example, in a wireless telephone apparatus having a handfree call mode and a handset call mode, it is determined whether the handset call mode or the handfree call mode is selected before the call is started, and the call mode by the handset is selected. In this case, the call mode may be changed to the hands-free call mode for at least the learning period, and a notification sound may be output from the speaker so as to learn the acoustic characteristics of the echo canceller based on the notification sound.

This produces the following effects. That is, in general, when making a call in the handset call mode, the echo canceller is not used and thus no learning is required.

However, if the call start is the handset call mode and the learning of the echo canceller is omitted, when the call is switched from the handset call mode to the hand-free call mode during the call, the echo immediately after the switch is not removed and is sent out as it is.

However, even in the handset call mode as described above, if the learner of the echo canceler is temporarily switched to the hand-free call mode, for example, even if the mode is switched during a call, a call signal that does not include an echo can be sent. The call quality can be kept high.

In the above embodiment, although the dial confirmation tone is used as the learning signal at the time of transmission, a ring tone may be used. In addition, you may study using a dial confirmation tone and ring tone, respectively. This ensures a sufficiently long learning period.

Alternatively, a voice synthesis circuit may be used to output a voice message for that purpose instead of ringing tones or ring tones. In this way, accurate learning can be performed as compared with the case of using the ring tone or ring tone.

In the above embodiment, the case where the present invention is applied to a digital radiotelephone apparatus has been described as an example. However, the present invention may be applied to a so-called dual mode device that communicates by selectively using a digital mode and an analog mode, or may be applied to an analog mode dedicated device. good. The present invention may be applied not only to a wireless telephone apparatus but also to a wired telephone apparatus having a hand-free call mode.

In addition, the circuit arrangement of the echo canceller, the control procedure of the learning control means, the control contents, and the like can also be modified in various ways without departing from the spirit of the present invention.

As described above, according to the present invention, an echo canceller and a learning control means for removing echoes generated by the sound input from the speaker for hand-free calling are input into a microphone, and the learning control means provides the speaker. By learning the acoustic characteristics of the echo canceller according to the sound output from the user, the manual operation for learning is unnecessary, thereby reducing the user's operational burden, and the learning operation even when the acoustic characteristics of the used space are changed or disconnected. Voice communication device with echo canceller and echo canceller with echo canceller to improve call quality and simplify circuits can be set at all times by operating condition of echo canceller without needing backup power. It may provide a method.

Claims (3)

A speaker for amplifying and outputting a spoken voice signal separately from a handset call mode and a microphone for collecting a tone and outputting a call signal, and an echo canceller for removing echoes caused by the sound output from the speaker being input to the microphone. A hand-free call mode configured to include a hand-free call using the speaker and a microphone, wherein the echo canceler uses the notification sound that is amplified and output to the call space through the speaker to adjust the acoustic characteristics of the space. A voice communication apparatus for estimating in advance, which is provided in the acoustic echo canceller, switching switch means for supplying the sign language audio signal to the speaker in a first state and supplying a predetermined notification sound to the speaker in a second state. And the setting of the call path with the reception to or transmission from the voice communication device And a control means for controlling the switching switch means to switch from the first state to the second state and to supply the notification sound to the switching switch means at the completion point. 2. The voice communication apparatus as claimed in claim 1, wherein said notification sound of said control means includes a frequency component necessary for learning as a notification sound for notifying the operation state during a call connection. It is composed of a speaker that amplifies and outputs a spoken voice signal separately from the handset call mode, a microphone that picks up a tone and outputs a call signal, and an echo canceler that removes echoes caused by the sound output from the speaker being input to the microphone. And a hand-free call mode, wherein a hand-free call is made using the speaker and the microphone, and the echo canceler learns in advance the acoustic characteristics of the call space by using the notification sound amplified and output to the call space through the speaker. A voice communication apparatus, comprising: switching switch means provided in the echo canceller, for supplying the speech signal to the speaker in a first state, and for supplying a predetermined notification sound to the speaker in a second state; The handset call mode is selected or the hand free call mode is selected. A judging means for judging whether or not the handset call mode is selected by the judging means, and the call switch mode is changed to the hand-free call mode only at least during the learning period, and the switching switch means And control means for switching from the state to the second state and controlling the notification sound to be supplied to the selection means.