JP4566078B2 - In-vehicle hands-free communication device - Google Patents

Description

  The present invention relates to an echo cancellation technique in an in-vehicle hands-free communication device.

2. Description of the Related Art Conventionally, there is known a hands-free call device that includes a microphone used for input of transmitted voice and a speaker used for output of received voice and adds a hands-free call function to a telephone.
Also, such a hands-free communication device is generally provided with an echo canceller that cancels the echo component of the received voice output from the speaker included in the sound signal input to the microphone (for example, a patent) Reference 1).
Here, this echo canceller transmits an echo path using the received voice signal output to the speaker as an input and the sound signal output from the microphone as an output based on the received voice signal output to the speaker and the sound signal output from the microphone. It is configured using an adaptive filter that learns functions. Then, using the adaptive filter and the received voice signal output to the speaker, the echo component of the received voice output from the speaker included in the sound signal output from the microphone is estimated, and the estimated echo component is converted to the microphone. Is subtracted from the sound signal output from the sound signal, the echo component of the received voice output from the speaker included in the sound signal input to the microphone is canceled.
JP 5-48547 A

When the above-described hands-free call device is applied to an in-vehicle hands-free call device that provides a user with a hands-free call function to an in-vehicle user, the period during which echo cannot be canceled properly by the echo canceller may be prolonged.
That is, in the echo canceller using the adaptive filter, when the transfer function of the echo path changes, the echo cannot be canceled properly until the learning of the transfer function converges. In an automobile, the transfer function of the echo path often changes temporarily due to a temporary behavior of the automobile or a change in a situation inside the automobile. For example, when the vehicle travels on a curve or when the traveling direction is changed, the driver's posture temporarily changes due to the traveling direction changing operation, so that the transfer function of the echo path in the vehicle temporarily changes. In such a case, if the learning of the transfer function is continued during the temporarily changing period, the period until the learning of the stable transfer function converges may be prolonged.

  Therefore, the present invention has an object to provide an in-vehicle hands-free communication device capable of promptly returning to a state where echo cancellation can be performed properly even when a transfer function of a temporary echo path changes. And

  In order to achieve the above object, the present invention provides an in-vehicle hands-free call device that is mounted on an automobile and processes call voice of a voice communication terminal, a speaker that outputs received voice received by the voice communication terminal, and a voice. A microphone to be input, an echo canceling means for canceling an echo component of a received voice included in a voice input from the microphone, and a transmitted voice transmitted by the voice communication terminal; and temporary movement of a passenger of the vehicle It comprises an estimation means for estimating the generation period and a learning control means. Here, the echo cancellation means learns a transfer function of the echo path of the received voice, uses an learned filter to generate a pseudo echo signal that simulates the echo component from the received voice, and a microphone. Subtracting means for subtracting the pseudo echo signal from the voice input from the voice to obtain the transmitted voice. Further, the learning control means stops learning of the adaptive filter during the generation period of the temporary movement of the vehicle occupant estimated by the estimation means.

  According to such an in-vehicle hands-free communication device, during the period when the transfer function of the echo path in the vehicle temporarily changes due to the temporary movement of the passenger, the learning operation of the transfer function of the echo path of the adaptive filter is performed. In addition to stopping, the learning operation of the transfer function of the echo path of the adaptive filter can be resumed after the temporary movement of the passenger is finished and the transfer function of the echo path in the vehicle is stabilized. Therefore, after the passenger's temporary movement is finished and the transfer function of the echo path in the vehicle is stabilized, the learning of the transfer function of the echo path of the adaptive filter should be quickly converged so that the echo cancellation can be performed properly. Can do.

  In order to achieve the above object, the present invention provides an in-vehicle hands-free call device that is mounted on an automobile and processes call voice of a voice communication terminal, a speaker that outputs received voice received by the voice communication terminal, and A microphone for inputting voice, an echo canceling means for canceling an echo component of the received voice included in the voice inputted from the microphone, and transmitting voice transmitted by the voice communication terminal; and a temporary passenger of the automobile It comprises an estimation means for estimating the motion generation period and a learning control means. Here, the echo cancellation means learns a transfer function of the echo path of the received voice, uses an learned filter to generate a pseudo echo signal that simulates the echo component from the received voice, and a microphone. Subtracting means for subtracting the pseudo echo signal from the voice input from the voice to obtain the transmitted voice. Further, the learning control means sets the transfer function of the adaptive filter to a predetermined transfer function during the generation period of the temporary movement of the vehicle occupant estimated by the estimation means, and the estimation means After the estimated generation period of the temporary movement of the passenger of the vehicle has elapsed, the transfer function of the adaptive filter is restored to the transfer function before the generation of the period. Note that such an in-vehicle hands-free communication device further includes a table in which correspondences between types of movements of the passengers of the automobile and transfer functions are registered, and the estimation means temporarily stores the passengers of the automobile. During the generation period of the temporary movement of the occupant of the vehicle estimated by the estimation means, the transfer function of the adaptive filter is estimated by the learning control means. The transfer function corresponding to the type of motion estimated by the estimation means registered in the table is set, and after the generation period of the temporary movement of the passenger of the vehicle estimated by the estimation means has elapsed, the period You may make it return the transfer function of the said adaptive filter to the transfer function before generation | occurrence | production.

  According to such an in-vehicle hands-free device, the echo of the adaptive filter can be promptly performed so that the echo can be properly canceled after the temporary movement of the passenger is completed and the transfer function of the echo path in the vehicle is stabilized. It is possible to converge the learning of the transfer function of the route and echo better than when the adaptive filter learning is continued or when learning is stopped during the period when the passenger is temporarily moving. Expect to be able to cancel.

  In each of the above-described in-vehicle hands-free communication devices, in the estimation means, the period in which the automobile changes the traveling direction is defined as a period in which a temporary movement of the passenger for the traveling direction changing operation occurs. You may make it estimate. Further, in this case, a current position calculation device for calculating a current position on the map of the vehicle is provided, and the vehicle is turned on a curve or an intersection from the transition of the current position calculated by the current position calculation device in the estimation unit. It is also possible to estimate the period of passing the vehicle and estimate the period of passing the estimated curve or intersection as the period during which the passenger is temporarily moving.

  Alternatively, each of the above-described in-vehicle hands-free communication devices is provided with an operation detection unit that detects an operation of the driving device of the automobile, and the estimation unit is configured according to the operation of the driving device detected by the operation detection unit. You may make it estimate the period when the passenger's temporary movement has generate | occur | produced. Alternatively, each of the above in-vehicle hands-free communication devices is provided with a camera that captures the interior of the vehicle, and the estimation means determines the passenger's image according to the movement of the passenger's image captured by the camera. You may make it estimate the period when the temporary motion has generate | occur | produced. Alternatively, each of the above-described in-vehicle hands-free communication devices is provided with a vehicle speed sensor that detects the vehicle speed of the vehicle, and in the estimation unit, the passenger's temporary communication is performed according to the behavior of the vehicle speed detected by the vehicle speed sensor. You may make it estimate the period when the motion has generate | occur | produced.

  As described above, according to the present invention, when a change occurs in the transfer function of a temporary echo path, an in-vehicle hands-free communication device that can quickly return to a state in which echo cancellation can be properly performed is provided. Can be provided.

Hereinafter, embodiments of the present invention will be described.
FIG. 1 shows a configuration of an in-vehicle hands-free communication device according to the present embodiment.
This in-vehicle hands-free communication device is mounted on an automobile, and includes a voice processing device 1, a speaker 2, a microphone 3, a telephone interface unit 4, and an echo characteristic variation estimation unit 5, as shown in the figure. Further, the in-vehicle hands-free communication device includes various signals for detecting the situation in the vehicle and the interior of the vehicle, such as a winker sensor 61 that detects the operation state of the winker, a navigation device 62, and a rudder angle sensor 63 that detects the amount of rotation of the steering wheel. It has a sensor.

  Next, the speech processing apparatus 1 includes an echo canceller 11, a learning control unit 12, a reception voice amplifier 13, and a transmission voice amplifier 14. The echo canceller 11 includes an adaptive filter 110 including an FIR filter 111 and a coefficient update unit 112, and an adder 113.

  In such a configuration, the telephone interface unit 4 is detachably or wirelessly connected to a mobile telephone 10 that is generally called a mobile phone or the like, and outputs a received voice signal input from the mobile telephone 10 to the voice processing device 1. The transmitted voice signal input from the voice processing device 1 is output to the mobile telephone 10. In addition, the telephone interface unit 4 includes an input device such as a switch, relays an on-hook / off-hook command to the mobile phone 10 according to a user operation of the input device, and causes the mobile phone 10 to perform an on-hook / off-hook operation. Also do. The mobile telephone 10 outputs the received voice signal received from the other party through the mobile telephone network to the telephone interface unit 4 and transmits the transmitted voice signal input from the telephone interface unit 4 to the other party through the mobile telephone network. To do.

In the voice processing device 1, the reception voice amplifier 13 amplifies the reception voice signal input from the mobile telephone 10 via the telephone interface unit 4 to drive the speaker 2 and outputs the reception voice to the interior space.
On the other hand, the transmission voice amplifier 14 sends the voice of the interior space picked up by the microphone 3 to the echo canceller 11 as a transmission voice signal. The echo canceller 11 cancels the echo component of the received voice included in the transmitted voice signal and outputs it to the mobile telephone 10 via the telephone interface unit 4.

  Next, in the echo canceller unit, the adder 113 subtracts the pseudo echo signal calculated by the adaptive filter 110 from the transmission voice signal input from the transmission voice amplifier 14 by the adder 113 and then the telephone interface unit 4. Send to. The coefficient updating unit 112 of the adaptive filter 110 uses an LIR (mean square error) algorithm, an NLMS (normalized mean square error) algorithm, or the like so that the power of the transmitted voice signal output from the adder 113 is minimized. The update amount of the tap coefficient of 111 is calculated, and the process of updating the tap coefficient of the FIR filter 111 is repeated. Here, the impulse response of the FIR filter 111 in which the tap coefficient is set in this way simulates the transfer function of the echo path of the received voice signal. Therefore, the FIR filter 111 applies an impulse response determined by the set tap coefficient to the reception voice signal output from the reception voice amplifier 13 and receives the reception voice included in the transmission voice signal output from the transmission voice amplifier 14. A pseudo echo signal that simulates the echo component of the voice is generated and output to the adder 113.

As described above, in the adaptive filter 110, in order to make the impulse response of the FIR filter 111 follow the change in the transfer function of the echo path, the tap coefficient updating operation of the FIR filter 111 by the coefficient updating unit 112 is continuously performed. ing.
Next, an echo characteristic variation estimation process performed by the echo characteristic variation estimation unit 5 will be described.
FIG. 2a shows the procedure of this echo characteristic variation estimation process.
As shown in the figure, in this process, first, it is checked whether or not the occurrence of a temporary movement of a passenger in the car is estimated (step 202). Here, the estimation of the occurrence of the temporary movement of the passenger is obtained, for example, from the navigation device 62 by acquiring the current position of the own vehicle, the moving speed of the own vehicle, the moving direction of the own vehicle, and map information around the current position, Judgment is made whether the vehicle is entering a curve or an intersection, and if it is entering, it is estimated that a temporary movement has occurred due to the driver's steering operation, etc. for changing the traveling direction of the vehicle To do. Alternatively, when the turn signal sensor 61 detects a change in the turn-on state of the turn signal, it is estimated that a temporary movement of the passenger due to the driver's steering operation for changing the traveling direction of the own vehicle has occurred. To do. Or, for example, when the rudder angle sensor 63 detects that the rudder angle of the steering wheel is equal to or greater than a predetermined rotation angle, a temporary movement due to a driver's steering wheel operation or the like for changing the traveling direction of the host vehicle is performed. This is done by estimating that it has occurred.

And when it is estimated that the passenger's temporary movement has occurred, the echo characteristic fluctuation occurrence is estimated (step 204), and the process proceeds to step 206, where the passenger's temporary movement has occurred. If not, return to Step 202.
In step 206, it is monitored whether or not the passenger's temporary movement has ended. The detection of the end of the temporary movement of the passenger is performed by, for example, acquiring the current position of the own vehicle, the moving speed of the own vehicle, the moving direction of the own vehicle, or map information around the current position from the navigation device 62. It is determined by determining whether or not the vehicle has passed a curve or an intersection, and when it has passed, by detecting the end of the temporary movement of the passenger by the steering wheel operation for changing the course. Alternatively, it is performed by detecting the end of the temporary movement of the passenger when the turn signal sensor 61 detects a change of the turn signal to the off state. Alternatively, when the steering angle sensor 63 detects that the steering angle of the steering wheel has returned to the straight traveling position, it detects the end of the temporary movement of the passenger by the steering wheel operation for changing the course.

When the passenger's temporary movement is completed, the end of the echo characteristic variation is estimated (step 208), and the process returns to step 202.
Next, the learning control process performed by the learning control unit 12 will be described.
FIG. 2b shows the procedure of this learning control process.
As shown in the figure, in this process, first, the occurrence of an estimation of the occurrence of echo characteristic variation (step 204 in FIG. 2a) by the echo characteristic variation estimation unit 5 is monitored (step 252). If the estimation of the occurrence of the echo characteristic variation occurs, the coefficient updating unit 112 stops the tap coefficient updating operation of the FIR filter 111 (step 254).

  Then, the generation of the echo characteristic fluctuation end estimation (step 208 in FIG. 2a) by the echo characteristic fluctuation estimation unit 5 is monitored (step 256). If the end of the echo characteristic variation is estimated, the coefficient updating unit 112 restarts the tap coefficient updating operation of the FIR filter 111 (step 258), and returns to step 252.

The embodiment of the present invention has been described above.
As described above, according to the present embodiment, the tap coefficient of the FIR filter 111 of the coefficient updating unit 112 is updated during a period in which the transfer function of the echo path in the vehicle temporarily changes due to the temporary movement of the passenger. The operation is stopped, and after the temporary movement of the passenger is finished and the transfer function of the echo path in the vehicle is stabilized, the updating operation of the tap coefficient of the FIR filter 111 of the coefficient updating unit 112 is resumed.

  Therefore, after the occupant's temporary movement is completed and the transfer function of the echo path in the vehicle is stabilized, the tap coefficient of the FIR filter 111 is promptly set, and the impulse response of the FIR filter 111 is the echo path of the received voice signal. It is possible to follow the transfer function and promptly shift to a state where echo cancellation can be performed appropriately.

By the way, in the above embodiment, in an echo characteristic fluctuation | variation estimation process, you may make it detect a passenger's temporary motion as follows.
That is, for example, the in-vehicle hands-free communication device is provided with an in-vehicle camera that captures the interior of the vehicle, monitors the interior of the vehicle captured by the in-vehicle camera, and detects the movement of the driver and other passengers by image analysis. When the driver or other passenger moves greatly, the occurrence of echo characteristic fluctuation is estimated, and the learning control process stops the coefficient updating operation of the FIR filter 111 of the coefficient updating unit 112, and the driver or other boarding When the person's movement becomes stable, the end of the echo characteristic variation may be estimated, and the coefficient update operation of the FIR filter 111 of the coefficient update unit 112 may be resumed in the learning control process. Or, when a vehicle speed sensor that detects the vehicle speed of the vehicle is provided in the in-vehicle hands-free call device, the occurrence of the gear change operation of the user is predicted from the vehicle speed detected by the vehicle speed sensor, and the occurrence of the gear change operation is predicted, The occurrence of echo characteristic fluctuation is estimated, and the learning control process stops the coefficient updating operation of the FIR filter 111 of the coefficient updating unit 112, and when the vehicle speed is stabilized, the end of the echo characteristic fluctuation is estimated and the learning control process is performed. The coefficient update operation of the FIR filter 111 of the coefficient update unit 112 may be restarted. Alternatively, the in-vehicle hands-free communication device is provided with a gear change sensor for detecting the gear shift state of the transmission of the vehicle or the driver's gear shift operation, and when the gear change occurs, the occurrence of the echo characteristic fluctuation is estimated and the learning control processing is performed. Then, the coefficient updating operation of the FIR filter 111 of the coefficient updating unit 112 is stopped, and after a certain period of time has elapsed after the gear change occurs, the end of the co-characteristic variation is estimated, and the FIR filter 111 of the coefficient updating unit 112 is used for the learning control process. The coefficient update operation may be resumed.

  In the above embodiment, the tap coefficient update operation of the FIR filter 111 of the coefficient update unit 112 is stopped during a period in which the transfer function of the echo path in the vehicle temporarily changes due to the temporary movement of the passenger. In addition, the coefficient updating unit 112 sets a predetermined tap coefficient in the FIR filter 111, and after the temporary movement of the passenger is finished and the transfer function of the echo path in the vehicle is stabilized, the coefficient updating unit 112 The tap coefficient of the FIR filter 111 may be returned to the original tap coefficient before setting the predetermined tap coefficient.

  That is, in this case, as shown in FIG. 3, a tap coefficient table 15 and a save memory 16 are further provided in the in-vehicle hands-free device. The tap coefficient table 15 is a predetermined value that is expected to realize the impulse response of the FIR filter 111 that simulates the type of movement of the vehicle occupant and the transfer function of the echo path when the movement occurs. Register the correspondence with the tap coefficient.

  The echo characteristic fluctuation estimation unit 5 estimates the occurrence of the echo characteristic fluctuation in step 204 when the occurrence of the temporary movement of the vehicle occupant is estimated in step 202 of the echo characteristic fluctuation estimation process. Is notified to the learning control unit 12. That is, for example, when the occurrence of temporary movement due to the driver's steering operation for changing the traveling direction of the host vehicle is detected as described above, the steering control is notified to the learning control unit 12 as the type of movement. To do. Further, when the occurrence of a temporary movement due to the gear change operation is predicted as described above, the learning control unit 12 is notified of the gear change operation as the type of movement.

  On the other hand, the learning control unit 12 causes the coefficient update unit 112 to store the current tap coefficient of the FIR filter 111 in the save memory 16 when the echo characteristic variation estimation is estimated by the echo characteristic variation estimation unit 5. After that, the coefficient update unit 112 causes the FIR filter 111 to set the tap coefficient registered in the tap coefficient table 15 corresponding to the type of motion notified from the echo characteristic variation estimation unit 5.

  If the echo characteristic variation estimator 5 estimates the end of the echo characteristic variation, the coefficient updating unit 112 causes the FIR filter 111 to set the tap coefficient stored in the save memory 16. And it waits for generation | occurrence | production of the estimation of echo characteristic fluctuation | variation generation | occurrence | production by the next echo characteristic fluctuation | variation estimation part 5.

  In this manner, when the coefficient updating unit 112 continues to update the tap coefficient of the FIR filter 111 according to the output of the adder 113 during the period when the passenger is temporarily moving, the FIR filter It can be expected that echo cancellation can be performed better than when the tap coefficient update of 111 is stopped.

It is a block diagram which shows the structure of the vehicle-mounted hands-free telephone apparatus which concerns on embodiment of this invention. It is a flowchart which shows the echo characteristic fluctuation | variation estimation process and learning control process which concern on embodiment of this invention. It is a block diagram which shows the other structural example of the vehicle-mounted hands-free telephone apparatus which concerns on embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Voice processing apparatus, 2 ... Speaker, 3 ... Microphone, 4 ... Telephone interface part, 5 ... Echo characteristic variation estimation part, 10 ... Mobile telephone, 11 ... Echo canceller, 12 ... Learning control part, 13 ... Received voice amplifier, DESCRIPTION OF SYMBOLS 14 ... Voice amplifier, 15 ... Tap coefficient table, 16 ... Saving memory, 61 ... Winker sensor, 62 ... Navigation apparatus, 63 ... Steering angle sensor, 110 ... Adaptive filter, 111 ... FIR filter, 112 ... Coefficient update part, 113: Adder.

Claims (8)

An in-vehicle hands-free call device that is mounted on a car and processes call voice of a voice communication terminal,
A speaker for outputting the received voice received by the voice communication terminal;
A microphone for voice input,
Echo canceling means for canceling the echo component of the received voice included in the voice input from the microphone and making the voice to be transmitted by the voice communication terminal;
Estimating means for estimating a period of occurrence of temporary movement of a passenger of the vehicle;
Learning control means,
The echo cancellation means learns the transfer function of the echo path of the received voice, uses a learned transfer coefficient to generate a pseudo echo signal that simulates the echo component from the received voice, and inputs from a microphone Subtracting the pseudo-echo signal from the voice to have the transmission voice,
The in-vehicle hands-free call device, wherein the learning control means stops learning of the adaptive filter during a period of occurrence of temporary movement of the vehicle occupant estimated by the estimating means. An in-vehicle hands-free call device that is mounted on a car and processes call voice of a voice communication terminal,
A speaker for outputting the received voice received by the voice communication terminal;
A microphone for voice input,
Echo canceling means for canceling the echo component of the received voice included in the voice input from the microphone and making the voice to be transmitted by the voice communication terminal;
Estimating means for estimating a period of occurrence of temporary movement of a passenger of the vehicle;
Learning control means,
The echo cancellation means learns the transfer function of the echo path of the received voice, uses a learned transfer coefficient to generate a pseudo echo signal that simulates the echo component from the received voice, and inputs from a microphone Subtracting the pseudo-echo signal from the voice to have the transmission voice,
The learning control means sets the transfer function of the adaptive filter to a predetermined transfer function during the generation period of temporary movement of the occupant of the vehicle estimated by the estimation means, and the estimation means estimates An in-vehicle hands-free communication device, wherein the transfer function of the adaptive filter is restored to the transfer function before the occurrence of the temporary movement of the vehicle occupant after the passage of the generation period. An in-vehicle hands-free call device according to claim 2,
A table in which the correspondence between the type of movement of the passenger of the car and the transfer function is registered;
The estimation means estimates the generation period of the temporary movement of the passenger of the automobile and the type of the movement,
The learning control means, during the generation period of the temporary movement of the occupant of the vehicle estimated by the estimation means, the transfer function of the adaptive filter is the motion estimated by the estimation means registered in the table. The transfer function corresponding to the type is set, and after the generation period of the temporary movement of the vehicle occupant estimated by the estimation means, the transfer function of the adaptive filter is restored to the transfer function before the generation of the period. An in-vehicle hands-free communication device. The in-vehicle hands-free call device according to claim 1, 2, or 3,
The in-vehicle hands-free call device characterized in that the estimation means estimates a period during which the vehicle changes its traveling direction as a period during which a temporary movement of a passenger for a traveling direction changing operation occurs. . The in-vehicle hands-free call device according to claim 1, 2, or 3,
Operation detecting means for detecting operation of the driving device of the automobile;
The in-vehicle hands-free call device characterized in that the estimation means estimates a period during which the occupant's temporary movement is occurring according to the operation of the driving device detected by the operation detection means. The in-vehicle hands-free call device according to claim 1, 2, or 3,
A camera for photographing the interior of the car;
The in-vehicle hands-free call device characterized in that the estimating means estimates a period during which the temporary movement of the occupant is generated according to a movement of an image of the occupant in the vehicle photographed by the camera. . The in-vehicle hands-free call device according to claim 1, 2, or 3,
A vehicle speed sensor for detecting the vehicle speed of the automobile;
The in-vehicle hands-free call device characterized in that the estimating means estimates a period in which the occupant's temporary movement occurs according to a behavior of a vehicle speed detected by the vehicle speed sensor. An in-vehicle hands-free call device according to claim 4,
A current position calculating device for calculating a current position on the map of the car;
The estimation means estimates a period during which the vehicle passes a curve or an intersection from the transition of the current position calculated by the current position calculation device, and determines a period during which the vehicle passes the estimated curve or intersection as a temporary An in-vehicle hands-free communication device characterized by estimating a period during which a large movement is occurring.