JP2012195801A - Conversation support device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a conversation support device which can suppress howling when controlling a conversation voice level according to noise in a device supporting a conversation within a vehicle by a microphone and a speaker.SOLUTION: A conversation support device comprises: multipliers 107 and 108 which change gain when outputting a voice recorded by a microphone from a speaker; a noise level analyzer 109 which analyzes a noise level within a vehicle; a control amount calculation device 111 which calculates coefficients of the multipliers 107 and 108; and a howling margin setter 112 which sets the maximum gain amount that does not cause howling when changing the coefficients of the multipliers 107 and 108. The control amount calculation device 111 takes the howling margin set in the howling margin setter 112 into account to calculate the coefficients of the multipliers 107 and 108 according to the result of the noise level analysis by the noise level analyzer 109.

Description

  The present invention relates to a conversation support apparatus that records conversation voice with a microphone and reproduces it with a speaker in a vehicle cabin.

  As a conventional conversation support device, the voice input from the hands-free microphone of the in-vehicle phone installed in the front seat is from the rear seat speaker, and the voice input from the microphone installed in the rear seat is from the front seat speaker. As a result, the front and rear seats can talk with the front and rear microphones and speakers, and when the speaker's voice is output from the front and rear speakers, By reducing the output other than the voice and mixing it with the voice, the speaker's voice is output from the front seat speaker and the rear seat speaker, making it easier to hear the speaker's voice. Is known (see, for example, Patent Document 1).

  In addition, as another conventional conversation support device, a microphone is installed for each seat, the conversational sound is separated from the sound collected by each microphone, and the conversational sound is recorded according to the sound pressure level of the audio sound and surrounding noise. There is known one that performs gain correction and outputs a conversational voice together with an audio sound from a speaker (see, for example, Patent Document 2).

  Further, as another conventional conversation support device, an in-car conversation assistance device that assists a person's conversation in an automobile using an acoustic device including a microphone and a speaker, the detection means for detecting the running state of the automobile, Noise estimation means for estimating a noise level indicating the magnitude of acoustic noise in the vehicle based on the detected driving state, and acoustic control means for setting the acoustic characteristics of the acoustic device according to the estimated noise level. The thing characterized by this is known (for example, refer patent document 3).

Japanese Patent No. 3921988 JP 2002-51392 A JP 2008-153743 A

  However, in the conventional conversation support device, it is easy to hear by reducing the music signal other than the voice, but there is a problem that the size of the music is reduced.

  In another conventional conversation support apparatus, the voice level is changed according to the music and noise levels. If there are a plurality of microphones and speakers in the same space, if the conversation voice level is too high, There was a problem that it became easy to do howling.

  The present invention has been made to solve the conventional problems, and it is an object of the present invention to provide a conversation support apparatus that can talk without lowering the music level and is difficult to perform howling.

A conversation support device according to the present invention is a device that supports a conversation in a vehicle interior using a microphone and a speaker, and a multiplier that changes a gain when audio recorded by the microphone is output from the speaker, and a noise level in the vehicle interior. A configuration including a noise level analyzer to analyze, a control amount calculator for calculating a coefficient of the multiplier, and a howling margin setting unit for setting a maximum gain amount that does not howling when the coefficient of the multiplier is changed. Have.

  The present invention can provide a conversation support device that has an effect of being difficult to howling in order to consider a howling margin when the level of conversation speech that is difficult to hear due to noise is increased according to the noise level. It is.

The block diagram of the vehicle-mounted audio apparatus in the 1st Embodiment of this invention The block diagram of the vehicle-mounted audio apparatus in the 2nd Embodiment of this invention The block diagram of the vehicle-mounted audio apparatus in the 3rd Embodiment of this invention The block diagram of the vehicle-mounted audio apparatus in the 4th Embodiment of this invention The block diagram which shows an example of the music canceller in the conversation assistance apparatus of this invention The block diagram which shows an example of the echo canceller in the conversation assistance apparatus of this invention The block diagram which shows an example of the noise suppressor in the conversation assistance apparatus of this invention

(Embodiment 1)
Hereinafter, a conversation support apparatus according to a first embodiment of the present invention will be described with reference to the drawings.
A conversation support apparatus according to a first embodiment of the present invention is shown in FIG.

  In FIG. 1, the conversation support apparatus 100 includes a front seat microphone 118 and a front seat speaker 120 installed in the front seat in the car 117, and a rear seat microphone 119 and a rear seat speaker 121 installed in the rear seat in the car 117. And connected.

  The conversation support apparatus 100 is also connected to an audio signal generator such as a CD player 114. An amplifier for amplifying a signal may be provided between the CD player 114, the front seat microphone 118, the rear seat microphone 119, the front seat speaker 120, the rear seat speaker 121, and the conversation support apparatus 100.

  In addition, an A / D converter that converts an analog signal into a digital signal is provided between the CD player 114, the front seat microphone 118, and the rear seat microphone 119 and the conversation support device 100, and the front seat speaker 120 and the rear seat speaker 121. A D / A converter that converts a digital signal into an analog signal may be provided between each of them and the conversation support apparatus 100.

The conversation support apparatus 100 removes the echo components input to the front seat microphone 118 and the rear seat microphone 119 as well as the music cancellers 103 and 104 that remove the music signals input to the front seat microphone 118 and the rear seat microphone 119. Similarly, the echo cancellers 101 and 102, the noise suppressors 105 and 106 for suppressing the noise signals input to the front seat microphone 118 and the rear seat microphone 119, and the gains of the signals input from the front seat microphone 118 and the rear seat microphone 119 Multipliers 107 and 108, and adders 115 and 116 for adding music signals from the CD player 114 to signals input to the front seat microphone 118 and the rear seat microphone 119, front seat microphone 118, and rear seat microphone 119, respectively. A noise level analyzer 109 for analyzing the level of the noise signal input to the CD player 114, and a music signal from the CD player 114. Music signal level analyzer 110 for analyzing the level of the sound signal, howling margin setting device 112 for setting the coefficient that does not howling of multipliers 107 and 108, noise level analyzer 109, music signal level analyzer 110, and howling margin A control amount calculator 111 that calculates coefficients to be set in the multipliers 107 and 108 from the output signal of the setting unit 112; and a coefficient updater 113 that sets the coefficients calculated by the control amount calculator 111 in the multipliers 107 and 108; It consists of

  The conversation support apparatus 100 may be realized by a digital signal processor (DSP) capable of performing digital signal processing, a microcomputer, or the like.

The operation of the conversation support apparatus configured as described above will be described with reference to FIG.
The voice of the passenger in the front seat in the vehicle 117 is recorded by the front seat microphone 118 installed in the front seat and output from the rear seat speaker 121 installed in the rear seat.

  The voice of the passenger in the rear seat is recorded by the rear seat microphone 119 installed in the rear seat and output from the front seat speaker 120 installed in the front seat.

  As described above, since the conversation voice in the front seat is output from the speaker in the rear seat and the conversation voice in the rear seat is output from the speaker in the front seat, the conversation in the vehicle interior is facilitated.

  However, for example, sound recorded by the front seat microphone 118 is reproduced from the rear seat speaker 121 and recorded by the rear seat microphone 119 and reproduced by the front seat speaker 120. End up.

  When the gain (loop gain) of such a system of the front seat microphone 118 -the rear seat speaker 121 -the rear seat microphone 119 -the front seat speaker 120 -the front seat microphone 118 exceeds 0 dB, it is once inputted to the front seat microphone 118. A signal called “howling” occurs because the received signal loops without being attenuated.

  In order to make it difficult for this howling to occur, the installation position of the microphone is taken into consideration so that the sound from the speaker is difficult to enter the microphone, and the directivity is given to the characteristics of the microphone. An echo canceller that cancels the sound input to the microphone is used.

Since the signal input to the microphone includes not only the voice but also a music signal from the CD player 114 and a noise signal in the passenger compartment, howling may occur due to these signals.
Therefore, a music canceller that cancels or suppresses them and a noise suppression process are used.

  A signal input to the front seat microphone 118 installed in the front seat passes through the music canceller 103, the echo canceller 101, the noise suppressor 105, the multiplier 107, and the adder 116, and from the rear seat speaker 121 installed in the rear seat. It is regenerated in the car 117.

  On the other hand, the signal input to the rear seat microphone 119 installed in the rear seat passes through the music canceller 104, the echo canceller 102, the noise suppressor 106, the multiplier 108, and the adder 115 in the same manner before being installed in the front seat. It is reproduced from the seat speaker 120 into the car 117.

  From the front seat speaker 120 and the rear seat speaker 121, not only signals input to the rear seat microphone 119 and front seat microphone 118, but also music signals that are output signals of the CD player 114 are respectively added by the adders 115 and 116. It is added to the microphone signal and played back.

  Since the operation of the conversation support apparatus 100 with respect to the front seat side and rear seat side audio signals is the same, the operation until the signal input to the front seat microphone 118 is reproduced from the rear seat speaker 121 will be described here. To do.

  The front seat microphone 118 includes not only the voice of the front seat passenger, but also the music signal reproduced from the front seat speaker 120, the voice of the rear seat passenger input to the rear seat microphone 119, Noise signal etc. are input.

  The signal input to the front seat microphone 118 is first input to the music canceller 103. The other input signal of the music canceller 103 is a music signal from the CD player 114.

  The music canceller 103 removes a signal having a high correlation with the music signal reproduced by the CD player 114 from the signal input to the front seat microphone 118.

  Here, the operation of the music canceller 103 will be described with reference to FIG. FIG. 5 is a block diagram illustrating an example of the music canceller 103.

  In FIG. 5, 500 is a music canceller, 501 is a music signal input terminal, 502 is a microphone signal input terminal, 503 is an output terminal, 504 is an adaptive filter, 505 is a coefficient updater for updating the coefficient of the adaptive filter 504, and 506 is a microphone. This is a subtracter that subtracts the output signal of the adaptive filter 504 from the signal input from the signal input terminal 502.

  Next, the operation of the music canceller 500 configured as described above will be described. The input signal of the front seat microphone 118 is input from the microphone signal input terminal 502 and becomes one input of the subtractor 506.

  On the other hand, the music signal from the CD player 114 is input from the music signal input terminal 501 and input to the adaptive filter 504. The adaptive filter 504 is composed of, for example, a FIR (Finite Impulse Response) filter, and the coefficient of the FIR filter is updated by the coefficient updater 505 so that the filter characteristics have desired amplitude and phase characteristics.

  The adaptive filter 504 changes the amplitude characteristic and phase characteristic of the input signal and outputs the result. The output signal of the adaptive filter 504 becomes the other input of the subtractor 506. The subtracter 506 subtracts the output signal of the adaptive filter 504 from the microphone signal and outputs the result.

  The subtracter 506 is connected to the output terminal 503, and the output signal of the subtracter 506 becomes the output of the music canceller 500.

  The coefficient updater 505 updates the coefficient of the adaptive filter 504 from the music signal input from the music signal input terminal 501 and the output signal of the subtractor 506 using, for example, an LMS (Least Mean Square) algorithm.

  The LMS algorithm is an algorithm that sequentially updates the coefficients of the FIR filter so that the sum of squares of errors is minimized.

  If the coefficient of the adaptive filter 504 is updated using the error as the output signal of the subtractor 506 that subtracts the music signal from the microphone signal, the coefficient of the adaptive filter 504 is set so that only the signal highly correlated with the music signal is removed from the microphone signal. Can be set.

  In this way, the music canceller 500 operates so as to remove only a highly correlated signal from the microphone signal and output it.

  Next, the operation of the echo canceller 101 will be described. The output signal of the music canceller 103 becomes one input of the echo canceller 101.

  The other input of the echo canceller 101 is an output signal of the echo canceller 102 for the rear seat. The output signal of the echo canceller 102 is a signal after the signal input to the rear seat microphone 119 is processed by the music canceller 104 and the echo canceller 102.

  The echo canceller 101 operates so as to remove a signal highly correlated with the voice of the passenger in the rear seat from the front seat microphone signal.

  Here, the operation of the echo canceller 101 will be described with reference to FIG. FIG. 6 is a block diagram illustrating an example of the echo canceller 101.

  In FIG. 6, 600 is an echo canceller, 601 is a rear-seat microphone signal input terminal, 602 is a front-seat microphone signal input terminal, 603 is an output terminal, 604 is an adaptive filter, and 605 is a coefficient updater that updates the coefficient of the adaptive filter 604. , 606 is a subtracter for subtracting the output signal of the adaptive filter 604 from the signal input from the front seat microphone signal input terminal 602.

  Next, the operation of the echo canceller 600 configured as described above will be described. An input signal to the front seat microphone 118 is input from the front seat microphone signal input terminal 602 and becomes one input of the subtractor 606.

  On the other hand, an input signal to the rear seat microphone 119 is input from the rear seat microphone signal input terminal 601 and input to the adaptive filter 604.

  The adaptive filter 604 is composed of, for example, an FIR filter as in the music canceller, and updates the coefficient of the FIR filter with the coefficient updater 605 so that the filter characteristics have desired amplitude and phase characteristics.

  The adaptive filter 604 changes and outputs the amplitude characteristic and phase characteristic of the input signal. The output signal of the adaptive filter 604 becomes the other input of the subtractor 606.

  The subtractor 606 subtracts the output signal of the adaptive filter 604 from the front seat microphone signal and outputs the result. The subtractor 606 is connected to the output terminal 603, and the output signal of the subtracter 606 becomes the output of the echo canceller 600.

  The coefficient updater 605 updates the coefficient of the adaptive filter 604 from the rear seat microphone signal input from the rear seat microphone signal input terminal 601 and the output signal of the subtractor 606 using, for example, the LMS algorithm in the same manner as the music canceller. .

  If the coefficient of the adaptive filter 604 is updated using the error in the LMS algorithm as the output signal of the subtractor 606, the coefficient of the adaptive filter 604 is set so as to remove only the signal highly correlated with the rear seat microphone signal from the front seat microphone signal. can do.

  Here, the signals input to the echo canceller 101 are the front seat microphone signal and the rear seat microphone signal, but since the music signals are removed by the music cancellers 103 and 102, the front seat microphone signal has the front seat boarding signal. The voice of the passenger and the voice of the passenger in the rear seat are greatly present in the rear-seat microphone signal.

  However, the front seat microphone signal also includes the voice of the passenger in the rear seat reproduced from the front seat speaker 120. Therefore, the echo canceller 600 operates so as to remove and output only the voice of the passenger in the rear seat and a high signal from the front seat microphone signal.

  Next, the operation of the noise suppressor 105 will be described. The output signal of the echo canceller 101 is input to the noise suppressor 105. The noise suppressor 105 operates so as to suppress a stationary noise signal from the input signal.

  Here, the operation of the noise suppressor 105 will be described with reference to FIG. FIG. 7 is a block diagram illustrating an example of the noise suppressor 105. In FIG. 7, 700 is a noise suppressor, 701 is an input terminal, 702 is an output terminal, 703 is a multiplier that changes the gain of the input signal, 704 is a noise level estimator that estimates the level of noise included in the input signal, and 705. Is a coefficient calculator for calculating the coefficient of the multiplier 703, and 706 is a coefficient updater for updating the coefficient of the multiplier 703.

  Next, the operation of the noise suppressor 700 configured as described above will be described. The gain of the signal input to the input terminal 701 is changed by the multiplier 703 and output from the output terminal 702.

  The signal input to the input terminal 701 is also input to the noise level estimator 704, and the noise level is estimated.

  When the input signal is a signal in which noise is superimposed on speech, the noise level estimator 704 estimates the noise level by processing such as minimum value detection, for example.

  By performing such processing, it is possible to detect a steady noise level in a time interval in which no voice exists.

  The noise level estimation method may be a method of estimating by taking an average of noise levels in sections other than the voice section.

  Also in this case, a steady noise level can be detected. The noise level estimated by the noise level estimator 704 becomes one input of the coefficient calculator 705.

  The other input of the coefficient calculator 705 is an input signal input from the input terminal 701. A coefficient calculator 705 calculates a coefficient to be set in the multiplier 703.

The coefficient can be calculated as follows, for example. Now, assuming that the amplitude value of the input signal is X and the noise amplitude value estimated by the noise level estimator 704 is N, the amplitude value Y of the output signal is
Y = X-N
By doing so, the amplitude value of the output signal can be obtained by subtracting the amplitude value of the noise signal from the amplitude value of the input signal. Dividing both sides of the above equation by X,
Y / X = (X−N) / X
Therefore,
Y = H · X
It can be expressed as However, H is
H = (X−N) / X
It is.

When this input signal is multiplied by H as a coefficient of the multiplier, the output signal is obtained by subtracting the noise signal.

  However, since these expressions are expressed as amplitude values, the phase component of the output signal is the same as the phase component of the input signal.

  In the above description, one multiplication is performed for the entire signal, but the input signal is divided into a plurality of frequency bands, the noise level is estimated for each frequency band, and the multiplication process is performed. good.

  In this case, more detailed control is possible, and the noise suppression effect is improved.

  Next, the gain of the signal whose noise has been suppressed by the noise suppressor 105 is changed by the multiplier 107 and becomes one input of the adder 116.

  The other input of the adder 116 is a music signal from the CD player 114. The adder 116 removes the music signal by the music canceller 103, removes the voice of the passenger in the rear seat output from the front seat speaker 120 by the echo canceller 101, and removes the noise by the noise suppressor 105. The voice of the passenger in the front seat recorded by the microphone 118 and the music signal are added and output. The output signal of the adder 116 is reproduced from the rear seat speaker 121 into the car 117.

  On the other hand, input signals to the front seat microphone 118 and the rear seat microphone 119 are input to the noise level analyzer 109. The noise level analyzer 109 analyzes the noise level in the vehicle 117 from the microphone signal. The analysis of the noise level may be performed in the same manner as the noise level estimation of the noise suppressor 105 described above.

  Although the noise level analyzer 109 is provided to analyze the noise level, when the noise suppressors 105 and 106 are provided, the noise level estimated by the noise suppressors 105 and 106 is used as the noise level analysis result. It is also good.

  In this case, it is possible to reduce the amount of processing required for noise level analysis and to reduce the system.

  The signal input to the noise level analyzer 109 is the signal of the front seat microphone 118 and the rear seat microphone 119, but only one of the signals may be used.

  Also in this case, since the noise level analysis is the analysis of one microphone signal, it is possible to reduce the processing amount for the noise level analysis.

  The noise level analyzed by the noise level analyzer 109 becomes one input of the control amount calculator 111. The other input of the control amount calculator 111 is the level of the music signal analyzed by the music signal level analyzer 110.

  The music signal level analyzer 110 inputs the music signal from the CD player 114 and analyzes the music signal level.

  As an analysis method of the music signal level analyzer 110, the energy of the music signal may be calculated or the frequency analysis of the music signal may be performed at regular time intervals.

  The control amount calculator 111 is a multiplier based on the noise level analyzed by the noise level analyzer 109, the music level analyzed by the music signal level analyzer 110, and the howling margin set by the howling margin setting unit 112. The coefficients set in 107 and 108 are calculated.

  The coefficient calculated by the control amount calculator 111 is set as a coefficient of the multipliers 107 and 108 by the coefficient updater 113.

  Next, a control amount calculation method by the control amount calculator 111 using the howling margin set by the howling margin setter 112 will be described.

In the howling margin setting unit 112, a maximum gain that is not howling is set from the gains of the system that can generate howling obtained by measurement or calculation in advance.
Alternatively, the maximum gain that is not howling is obtained by measurement or the like. If there is music or noise in the car 117, it becomes difficult to hear the conversation at the front and rear seats. Therefore, the control amount calculator 111 uses the noise level analyzed by the noise level analyzer 109 or the music analyzed by the music signal level analyzer 110. In accordance with the level, the control amount is calculated so as to increase the level of the sound output from the front seat speaker 120 and the rear seat speaker 121. At this time, the control amount exceeds the howling margin set by the howling margin setting unit 112. There is no control amount.

  By setting the control amount in this way, it is possible to prevent howling even if the sound level is increased according to the noise level or the music level.

  Note that the howling margin set by the howling margin setting unit 112 may be set so that the multipliers 107 and 108 have different coefficients.

  In this case, for example, by making the coefficient of the multiplier 108 a larger howling margin than the coefficient of the multiplier 107, the sound of the passenger's voice in the rear seat reproduced at the front seat can be increased. This is advantageous in that it can be set to give priority to the rear seats.

  Also, different howling margins may be set for each case when there is no noise or music, when there is only noise, when there is only music, or when there is noise and music.

  In this case, since a detailed howling margin is set, there is an effect that howling is more difficult.

  According to the conversation support apparatus of the first embodiment of the present invention as described above, in the apparatus that supports the conversation in the vehicle interior using the microphone and the speaker, the gain when the sound recorded by the microphone is output by the speaker is increased. A multiplier for changing, a noise level analyzer for analyzing the noise level in the passenger compartment, a music level analyzer for analyzing the level of a music signal reproduced in the passenger compartment, and a control amount calculation for calculating a coefficient of the multiplier And a howling margin setting unit for setting a maximum gain amount that does not perform howling when the coefficient of the multiplier is changed, the howling is performed when controlling the level of conversational voice according to the level of noise or music. Since control in consideration of the margin can be performed, howling can be made difficult.

(Embodiment 2)
Next, a conversation support apparatus according to a second embodiment of the present invention is shown in FIG.

In FIG. 2, the conversation support apparatus 200 includes a front seat microphone 217 and a front seat speaker 219 installed in the front seat in the car 216, and a rear seat microphone 218 and a rear seat speaker 220 installed in the rear seat in the car 216. And connected.

  The conversation support device 200 is also connected to an audio signal generator such as a CD player 213.

  The conversation support apparatus 200 removes echo components input to the front seat microphone 217 and the rear seat microphone 218, and music cancellers 203 and 204 that remove music signals input to the front seat microphone 217 and the rear seat microphone 218. Similarly, echo cancellers 201 and 202, noise suppressors 205 and 206 for suppressing noise signals input to the front seat microphone 217 and rear seat microphone 218, and gains of signals input from the front seat microphone 217 and rear seat microphone 218 Output signals of the noise suppressors 205 and 206, and adders 214 and 215 for adding music signals from the CD player 213 to input signals to the front seat microphone 217 and the rear seat microphone 218, respectively. Set the coefficients of the voice interval detector 209 that detects the voice interval from the voice and the multipliers 207 and 208 that do not perform howling. And a control amount calculator 210 for calculating coefficients to be set in the multipliers 207 and 208 from the output signals of the speech section detector 209 and the howling margin setter 211, and a control amount calculator 210. The coefficient updater 212 sets the calculated coefficient in the multipliers 207 and 208.

  The conversation support apparatus 200 may be realized by a DSP or a microcomputer that can perform digital signal processing.

  The operation of the conversation support apparatus configured as described above will be described with reference to FIG.

  Echo cancellers 201 and 202, music cancellers 203 and 204, noise suppressors 205 and 206, multipliers 207 and 208, coefficient updater 212, CD player 213, adders 214 and 215, front seat microphone 217, rear seat microphone 218, The operations of the front seat speaker 219 and the rear seat speaker 220 are the same as those in the first embodiment.

  Next, a control amount calculation method in the control amount calculator 210 will be described.

  First, the voice section detector 209 includes a signal input to the front seat microphone 217 installed in the front seat of the car 216 that has passed through the music canceller 203, the echo canceller 201, and the noise suppressor 205, and the music canceller 204, echo canceller. 202, a voice section which is a section in which a front seat passenger or a rear seat passenger speaks from a signal input to a rear seat microphone 218 installed in the rear seat of the car 216 that has passed through the noise suppressor 206 Is detected.

  A voice threshold is set by setting a level threshold, and a section where a signal larger than the threshold exists or a section before and after a signal larger than the threshold exists as a voice section is detected. can do.

  Note that not only the magnitude of the level but also whether or not there is a voice feature (for example, pitch) may be detected.

  In this case, there is an effect that the voice section can be detected with higher accuracy. In the howling margin setting device 211, when only the front seat passenger speaks, only the rear seat passenger speaks, or both the front seat and rear seat passengers speak For each of these, a howling margin is set in advance by measurement and calculation.

This is because when both front and rear passengers are speaking, there are two systems in which the sound input to the microphone is played from the speaker and input to the microphone, but only one of them speaks. This is because when there is only one system, the howling margin can be increased.

  The control amount calculator 210 uses the information on which of the front and rear seat passengers detected by the voice section detector 209 and the howling margin set by the howling margin setting unit 211 to use the multiplier 207. , 208 is calculated. As described above, when one of the front and rear passengers is speaking, the control amount calculator 210 calculates the coefficient of the multiplier 207 or 208 connected to the microphone on the speaking side. Set larger.

  Note that the coefficient of the multiplier on the utterance side may be increased and the coefficient of the multiplier on the utterance side may be decreased.

  In this case, since the gain of the system on the side where the voice is not spoken can be reduced, there is an effect that the howling becomes more difficult.

  Further, when both of the front and rear seat passengers are speaking, the control amount calculator 210 is more than the coefficient set in the multiplier 207 or 208 on the speaking side when either one is speaking. A small coefficient is set in the multipliers 207 and 208.

  Note that the howling margin set by the howling margin setting unit 112 when both the front and rear passengers are speaking may be set so that the multipliers 107 and 108 have different coefficients.

  In this case, it is possible to increase the voice of the passenger on the side where a large coefficient is set, and to set the front seat or the rear seat priority.

According to the conversation support apparatus of the second embodiment of the present invention, in the apparatus that supports the conversation in the vehicle interior using the microphone and the speaker, the gain when the sound recorded by the microphone is output by the speaker is increased. A multiplier for changing, a voice interval detector for detecting a voice interval from a microphone signal, a control amount calculator for calculating a coefficient of the multiplier, and a maximum gain amount that does not howling when the coefficient of the multiplier is changed. By providing the howling margin setting device to be set, it is possible to control how to take into account howling margins according to the situation of the passenger's conversation, so that howling can be made difficult.
(Embodiment 3)
Next, a conversation support apparatus according to a third embodiment of the present invention is shown in FIG.

  In FIG. 3, the conversation support apparatus 300 includes a front seat microphone 317 and a front seat speaker 319 installed in the front seat in the car 316, and a rear seat microphone 318 and a rear seat speaker 320 installed in the rear seat in the car 316. And connected.

The conversation support device 300 is also connected to an audio signal generator such as a CD player 313. The conversation support apparatus 300 removes echo components input to the front seat microphone 317 and the rear seat microphone 318 as well as music cancellers 303 and 304 that remove music signals input to the front seat microphone 317 and the rear seat microphone 318. Similarly, echo cancellers 301 and 302, noise suppressors 305 and 306 for suppressing noise signals input to the front seat microphone 317 and the rear seat microphone 318, and gains of signals input from the front seat microphone 317 and the rear seat microphone 318 Multipliers 307 and 308, adders 314 and 315 for adding music signals from the CD player 313 to input signals to the front seat microphone 317 and the rear seat microphone 318, and output signals from the noise suppressors 305 and 306 Residual echo analyzer 30 for analyzing echo components that could not be removed by echo cancellers 302 and 303 from Then, howling margin setting unit 311 for setting the coefficients of multipliers 307 and 308 that do not perform the calculation, and coefficients set in multipliers 307 and 308 are calculated from the output signals of residual echo analyzer 309 and howling margin setting unit 311. The control amount calculator 310 and a coefficient updater 312 that sets the coefficients calculated by the control amount calculator 310 in the multipliers 307 and 308 are configured.

  The conversation support apparatus 300 may be realized by a DSP or a microcomputer that can perform digital signal processing.

  About the conversation assistance apparatus comprised as mentioned above, the operation | movement is demonstrated using FIG.

  Echo cancellers 301 and 302, music cancellers 303 and 304, noise suppressors 305 and 306, multipliers 307 and 308, coefficient updater 312, CD player 313, adders 314 and 315, front seat microphone 317, rear seat microphone 318, The operations of the front seat speaker 319 and the rear seat speaker 320 are the same as in the first embodiment.

  Next, a control amount calculation method in the control amount calculator 310 will be described. First, the residual echo analyzer 309 analyzes the amount of residual echo that could not be removed by the echo cancellers 301 and 302 using the output signals of the noise suppressors 305 and 306. For example, the amount of residual echo may be determined by analyzing how much sound other than the sound desired to be recorded by a microphone exists.

  For example, it is possible to analyze the amount of residual echo by setting a threshold value for the level and analyzing a sound smaller than the threshold value. The signal to be analyzed by the residual echo analyzer 309 is not the output signal of the echo cancellers 301 and 302 but the output signal of the noise suppressors 305 and 306. This is because a small residual echo is obtained by the noise suppression processing by the noise suppressors 305 and 306. This is because it may be suppressed together with noise.

  Note that the residual echo analysis may be detected by adding not only the magnitude of the level but also whether or not a voice feature exists.

  In this case, it is possible to distinguish whether a signal smaller than the threshold value to be analyzed is voice or not, and the residual echo analysis accuracy is improved.

  When the residual echo is large, the sound input to the microphone is output from the speaker and the gain of the system input to the microphone is large, which facilitates howling.

  Therefore, in the howling margin setting unit 311, a howling margin corresponding to the amount of residual echo is set such that the howling margin is large when the residual echo is small.

  Then, the control amount calculator 310 calculates coefficients to be set in the multipliers 307 and 308 from the output signal of the residual echo analyzer 309 and the output of the howling margin setting unit 311. The control amount calculator 310 increases the coefficient of the multipliers 307 and 308 when the residual echo level analyzed by the residual echo analyzer 309 is small, and conversely, the coefficient of the multipliers 307 and 308 when the residual echo level is large. Set to a smaller value.

According to the conversation support apparatus of the third embodiment of the present invention, in the apparatus that supports the conversation in the vehicle interior using the microphone and the speaker, the gain when the sound recorded by the microphone is output by the speaker is increased. A multiplier to be changed, an echo canceller that removes an echo component generated when the sound output from the speaker is input to the microphone again, a residual echo analyzer that analyzes an echo component that could not be removed by the echo canceller, By providing a howling margin setting device that sets a howling margin according to the amount of residual echo and a control amount calculator that calculates the coefficient of the multiplier, control in consideration of the howling margin according to the amount of residual echo can be performed. Therefore, howling can be made difficult.
(Embodiment 4)
Next, a conversation support apparatus according to a fourth embodiment of the present invention is shown in FIG.

  In FIG. 4, the conversation support apparatus 400 includes a front seat microphone 417 and a front seat speaker 419 installed in the front seat in the car 416, and a rear seat microphone 418 and a rear seat speaker 420 installed in the rear seat in the car 416. And connected.

  The conversation support device 400 is also connected to an audio signal generator such as a CD player 413. The conversation support apparatus 400 removes the echo components input to the front seat microphone 417 and the rear seat microphone 418, and the music cancellers 403 and 404 that remove the music signals input to the front seat microphone 417 and the rear seat microphone 418. Similarly, echo cancellers 401 and 402, noise suppressors 405 and 406 for suppressing noise signals input to front seat microphone 417 and rear seat microphone 418, and gains of signals input from front seat microphone 417 and rear seat microphone 418 From the coefficients of the echo cancellers 401 and 402, multipliers 407 and 408 for changing the signal, adders 414 and 415 for adding the music signal from the CD player 413 to the input signals to the front seat microphone 417 and the rear seat microphone 418, and Filter coefficient analyzer 409 for estimating echo components, and multipliers 407 and 40 that do not perform howling , A control amount calculator 410 for calculating coefficients to be set in the multipliers 407 and 408 from output signals of the filter coefficient analyzer 409 and the howling margin setter 411, and a control amount The coefficient updater 412 sets the coefficient calculated by the calculator 410 in the multipliers 407 and 408.

  The conversation support apparatus 400 may be realized by a DSP or a microcomputer that can perform digital signal processing.

  The operation of the conversation support apparatus configured as described above will be described with reference to FIG.

  Echo cancellers 401 and 402, music cancellers 403 and 404, noise suppressors 405 and 406, multipliers 407 and 408, coefficient updater 412, CD player 413, adders 414 and 415, front seat microphone 417, rear seat microphone 418, The operations of the front seat speaker 419 and the rear seat speaker 420 are the same as those in the first embodiment.

  Next, a control amount calculation method in the control amount calculator 410 will be described. First, the filter coefficient analyzer 409 analyzes the amount of change in the filter coefficient size of the echo cancellers 401 and 402.

  When the echo cancellers 401 and 402 are configured using an FIR filter as described in the first embodiment, the coefficients analyzed by the filter coefficient analyzer 409 are the coefficients of the FIR filter.

  Since the coefficients of the FIR filter are sequentially updated by the adaptive algorithm, when the coefficient value is close to the optimum value, the amount of change in the coefficient value is small. On the contrary, when the distance is away from the optimum value, the change amount of the coefficient value becomes large. That is, it can be estimated that the echo is not sufficiently removed when the coefficient change amount is large, and that the echo is sufficiently removed when the coefficient change amount is small.

  The filter coefficient analyzer 409 analyzes the change in the filter coefficient size of the echo cancellers 401 and 402 in this way, and determines whether or not the echo is sufficiently suppressed. When the echo is not sufficiently suppressed, the sound input to the microphone is output from the speaker, and the gain of the system input to the microphone is large, which facilitates howling.

  Therefore, a howling margin is set in the howling margin setting unit 411 according to the amount of change in the echo canceller coefficient so that the howling margin is large when the amount of change in the echo canceller coefficient is small.

  Then, the control amount calculator 410 calculates coefficients to be set in the multipliers 407 and 408 from the output signal of the filter coefficient analyzer 409 and the output of the howling margin setting unit 411. According to the conversation support apparatus of the fourth embodiment of the present invention, in the apparatus that supports the conversation in the vehicle interior using the microphone and the speaker, the gain when the sound recorded by the microphone is output by the speaker is increased. A multiplier to be changed, an echo canceller that removes echo components generated when the sound output from the speaker is input to the microphone again, a filter coefficient analyzer that analyzes changes in the coefficient of the echo canceller, and a change in the filter coefficient By providing a howling margin setter that sets the howling margin according to the amount and a control amount calculator that calculates the coefficient of the multiplier, the howling margin according to the amount of change of the filter coefficient, that is, the amount of residual echo is taken into consideration Therefore, howling can be made difficult.

  As described above, the conversation support apparatus according to the present invention considers the howling margin when increasing the level of the conversation voice that is difficult to hear due to noise according to the noise level, or the howling margin according to the amount of echo. For example, when talking at the front and rear seats of a car, the conversation is not possible due to the distance between the front and rear seats or the presence of noise or music in the passenger compartment. It is useful as a conversation support device or the like that facilitates conversation by recording conversation voice with a microphone and reproducing it with a speaker in the passenger compartment.

100, 200, 300, 400 Conversation support apparatus 101, 102, 201, 202, 301, 302, 401, 402, 600 Echo canceller 103, 104, 203, 204, 303, 304, 403, 404, 500 Music canceller 105, 106, 205, 206, 305, 306, 405, 406, 700 Noise suppressor 107, 108. 207, 208, 307, 308, 407, 408, 703 Multiplier 109 Noise level analyzer 110 Music signal level analyzer 111, 210, 310, 410 Control amount calculator 112, 211, 311, 411 Howling margin setting unit 113, 212, 312, 412, 505, 605, 706 Coefficient updater 114, 213, 313, 413 CD player 115, 116, 214 215, 14, 315, 414, 415 Adder 117, 216, 316, 416 Car 118, 217, 317, 417 Front seat microphone 119, 218, 318, 418 Rear seat microphone 120, 219, 319, 419 Front seat speaker 121, 220 , 320, 420 Rear seat speaker 209 Voice section detector 309 Residual echo analyzer 409 Filter coefficient analyzer 501 Music signal input terminal 502 Microphone signal input terminals 503, 603, 702 Output terminals 504, 604 Adaptive filters 506, 606 Subtractor 601 Rear seat microphone signal input terminal 602 Front seat microphone signal input terminal 701 Input terminal 704 Noise level estimator 705 Coefficient calculator

Claims (8)

In a conversation support apparatus that is connected to a microphone and a speaker and supports conversation in a vehicle interior using the microphone and the speaker.
A multiplier for changing the gain of the sound recorded by the microphone, a noise level analyzer for analyzing the noise level in the passenger compartment, a control amount calculator for calculating a coefficient of the multiplier, and a maximum gain amount of the multiplier With howling margin setting device to set,
The control amount calculator calculates a coefficient of the multiplier according to a noise level analysis result by the noise level analyzer based on a howling margin set by the howling margin setter. apparatus. In a conversation support apparatus that is connected to a microphone and a speaker and supports conversation in a vehicle interior using the microphone and the speaker.
A multiplier for changing the gain of the sound recorded by the microphone, a music level analyzer for analyzing a music signal level reproduced in the passenger compartment, a control amount calculator for calculating a coefficient of the multiplier, It has a howling margin setting device that sets the maximum gain amount,
The control amount calculator calculates a coefficient of the multiplier according to a music signal level analysis result by the music level analyzer based on a howling margin set by the howling margin setter. Support device. In a conversation support apparatus that is connected to a microphone and a speaker and supports conversation in a vehicle interior using the microphone and the speaker.
A multiplier for changing the gain of the sound recorded by the microphone, a voice section detector for detecting the voice section of the signal recorded by the microphone, a control amount calculator for calculating the coefficient of the multiplier, and the voice section detection A howling margin setting device for setting a howling margin according to the conversation state determined from the voice section detected by the device,
The control amount calculator calculates a coefficient of the multiplier according to a conversation state discriminated from a voice section detected by the voice section detector based on a howling margin set by the howling margin setter. A conversation support device characterized by that. In a conversation support apparatus that is connected to a microphone and a speaker and supports conversation in a vehicle interior using the microphone and the speaker.
A multiplier that changes the gain of the sound recorded by the microphone, an echo canceller that removes an echo component generated when the sound output from the speaker is input again to the microphone, and an echo component that could not be removed by the echo canceller A residual echo analyzer that analyzes the coefficient of the multiplier, a control amount calculator that calculates a coefficient of the multiplier, and a howling margin setter that sets a howling margin according to the amount of residual echo analyzed by the residual echo analyzer Prepared,
The control amount calculator calculates a coefficient of the multiplier according to an amount of residual echo analyzed by the residual echo analyzer based on a howling margin set by the howling margin setter. Conversation support device. In a conversation support apparatus that is connected to a microphone and a speaker and supports conversation in a vehicle interior using the microphone and the speaker.
A multiplier that changes the gain of the sound recorded by the microphone, an echo canceller that removes echo components generated when the sound output from the speaker is input again to the microphone, and the amount of change in the filter coefficient of the echo canceller A filter coefficient analyzer, a control amount calculator for calculating a coefficient of the multiplier, and a howling margin setter for setting a howling margin according to a change amount of the filter coefficient analyzed by the filter coefficient analyzer. ,
The control amount calculator calculates a coefficient of the multiplier according to a change amount of the filter coefficient analyzed by the filter coefficient analyzer based on a howling margin set by the howling margin setter. Conversation support device. An echo canceller that removes an echo component generated when the sound output from the speaker is input to the microphone again;
4. The conversation support apparatus according to claim 1, wherein the echo canceller removes an echo component input to the microphone. A music canceller that removes a music signal reproduced in the vehicle interior from the signal input to the microphone,
6. The conversation support apparatus according to claim 1, wherein the music canceller makes it difficult for feedback by removing a music signal input to a microphone. A noise suppressor for suppressing a noise signal present in the passenger compartment among signals input to the microphone;
The conversation support apparatus according to claim 1, wherein the noise suppressor suppresses a noise signal input to a microphone.