JPH07143600A - Sound crosstalk controller - Google Patents

Abstract

PURPOSE:To reduce the circuit scale of the sound crosstalk controller by converting the sampling frequency and reducing the operation volume of an FIR filter. CONSTITUTION:A sampling conversion ratio (n) of a thinning filter 12 is obtained in accordance with the route length of sound crosstalk between a speaker 4 and a listener in the adjacent position and the operation speeds of FIR filters 14 and 15. When audio information 12 and R2 are supplied through LPFs 23 and 24, an artificial audio signal corresponding to crosstalk of audio information L1 is generated by FIR filters 14 and 15 and interpolating filters 18 and 19 and is synthesized with the normal audio signal by an adder. This signal is given to speakers 5L and 5R through DA converters 29 and 30, LPFs 31 and 32, and power amplifiers 33 and 34. Thus, sounds free from crosstalk are outputted at the listening position.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an acoustic crosstalk control device capable of independent loudspeaking in close proximity.

[0002]

2. Description of the Related Art Recently, there has been proposed a method of canceling unnecessary acoustic crosstalk by outputting a control sound from a speaker using a digital signal processing technique. Conventionally, this type of device is called a noise canceller or an acoustic crosstalk controller.

Here, a conventional acoustic crosstalk control device will be described with reference to the schematic block diagram shown in FIG. In FIG. 5, it is assumed that a plurality of seats are provided adjacent to each other and a speaker is attached to each seat. Then, it is assumed that the listener 1 seated in a certain seat listens to audio information (hereinafter referred to as an audio signal) such as audio or music provided independently in each seat through a speaker. The voice signal L1 of the adjacent seat is a FIR (Finite Impulse Responce) filter 2
a, 2b and the delay device 3. The voice signal delayed by the delay unit 3 for a predetermined time is converted into voice by the speaker 4,
Given to the listener of that seat. Moreover, this voice is also heard in the listener's ear 1.

In the FIR filter 2a, the speaker 4
The transfer function is convoluted to the left ear EL of the listener 1 who is present. Similarly, the FIR filter 2b convolves the transfer function up to the right ear ER of the listener 1 who is present, including the speaker 4, and the phase is inverted and added to the original signal. In this way, when the listener 1 in the seat listens to the sound from the left and right channel speakers 5L and 5R, the sound signal from the speaker 4, which is a crosstalk sound, is canceled by the control sound output from the FIR filters 2a and 2b. Only the sound of will be heard.

[0005]

However, in the above-mentioned conventional acoustic crosstalk control device, the spatial transmission characteristic Ga from the speaker 4 to the left ear 5R, the right ear 5L of the listener 1 is
In order to accurately simulate Gb, there is a drawback that the calculation amount of the FIR filters 2a and 2b becomes large.
For this reason, it is difficult to realize an acoustic crosstalk control device at low cost, and up until now, a quantitative method for reducing the circuit scale has not been clarified.

The present invention has been made in view of such conventional problems, and an object of the present invention is to realize an acoustic crosstalk control device capable of reducing the circuit scale and reducing the cost.

[0007]

According to the invention of claim 1 of the present application, when different audio information is provided to a plurality of adjacent listening positions, the listener at the first position is close to the first position. The first AD converter is an acoustic crosstalk control device for suppressing the crosstalk sound radiated from the second position, which converts the audio information given to the listener at the second position from analog to digital at the sampling frequency Fs. And a thinning filter that thins out the output of the first AD converter at a frequency of Fs / n (n> 1), an FIR filter that inputs a signal of the thinning filter and performs a convolution operation, and a tap coefficient of the FIR filter , A first interpolation filter for interpolating the output signal of the FIR filter into a signal of the sampling frequency Fs, and a first A
A delay device that delays the output of the D converter for a predetermined time, a first DA converter that performs analog conversion of the output signal of the delay device, and a second position that is provided at the second position and that converts the output of the first DA converter into an acoustic signal. The sampling frequency Fs is set to the audio information to be given to the first speaker and the listener at the first position.
A second AD converter that performs analog-to-digital conversion by means of an adder, an adder that adds the signal of the first interpolation filter and a signal of the second AD converter, and a second DA converter that performs analog conversion of the output of the adder , A second speaker which is provided at a second position and which converts an output of the second DA converter into an acoustic signal, and the coefficient memory has a first A
A first speaker including a first speaker from the output section of the D converter;
Let G be the transfer function to the listener's ear at the position
From the output of the D converter, including the second speaker, the first
When the transfer function to the ear of the listener at the position is set to C, a tap coefficient equivalent to the transfer function G / C is held, and the head center of the listener at the first position from the first speaker is held. To r [m], the calculation speed of the FIR filter is u
When [sec / tap] and the sound velocity are c [m / sec], the sampling conversion ratio n of the thinning filter is set to the minimum integer n that satisfies the expression (1).

According to a second aspect of the present invention, an oscillator for giving a random voice signal to the first AD converter and a crosstalk sound for the listener at the first position are suppressed. Instead, at the time of identification to newly set the tap coefficient of the coefficient memory, the adaptive FIR filter for performing the convolution operation of the signal of the thinning filter and the output of the thinning filter are input as the reference signal, and the crosstalk of the first speaker is input. A microphone that detects a synthesized sound at the listening position of the sound and the control sound of the second speaker, and the output of the microphone is input as an error signal, and the tap coefficient of the adaptive FIR filter is updated so that the error input becomes small. A coefficient updater and a second interpolation filter for interpolating the output signal of the adaptive FIR filter into the signal of the sampling frequency Fs. If, comprising a, an operation speed of the adaptive FIR filter and the FIR filter u [sec / ta
p], the calculation speed of the coefficient updater is v [sec / tap],
When the sound velocity is c [m / sec], the sampling conversion ratio n of the thinning filter is set to the smallest integer n that satisfies the expression (2).

According to the invention of claim 3 of the present application, when different audio information is provided to a plurality of adjacent listening positions, the listener at the first position is provided with a second position close to the first position. An acoustic crosstalk control device for suppressing radiated crosstalk sound, comprising: a first AD converter for converting analog / digital conversion of audio information given to a listener at a second position at a sampling frequency Fs; and a first AD converter. A thinning filter for thinning out the output of the converter at a frequency of Fs / n (n> 1), a first FIR filter for inputting a signal of the thinning filter, and performing a convolution operation with a first tap coefficient; The first FIR memory that stores the tap coefficient of the second FIR filter and the signal of the thinning filter is input to the first FIR memory, and the second FIR filter that performs the convolution operation with the second tap coefficient. Filter, a second coefficient memory for storing the second tap coefficient and giving it to the second FIR filter, a first interpolation filter for interpolating the output signal of the first FIR filter into a signal of the sampling frequency Fs, Second
The output signal of the FIR filter of the sampling frequency Fs
A second interpolation filter for interpolating the signal, a delay device for delaying the output of the first AD converter for a predetermined time, and a first DA converter for analog-converting the output signal of the delay device,
A first speaker provided at the second position for converting the output of the first DA converter into an acoustic signal, and audio information of either the left or right channel given to the listener at the first position is sampled at the sampling frequency Fs. A second AD converter that performs analog-to-digital conversion at the first position, a third AD converter that performs analog-to-digital conversion at the sampling frequency Fs on the other left and right channel audio information given to the listener at the first position, A first adder for adding the signal of the interpolation filter and the signal of the second AD converter; a second adder for adding the signal of the second interpolation filter and the signal of the third AD converter; A second DA converter for analog-converting the output of the first adder, a third DA converter for analog-converting the output of the second adder, and a second D-converter
The output of the A converter is converted into an acoustic signal, and the second speaker provided on one of the left and right of the second position and the third DA
The output of the converter is converted into an acoustic signal, and a third speaker provided on the other of the left and right of the second position is provided. The transfer function to the one ear of the listener at the first position including the first speaker is set to G1, and the output function of the first AD converter is set to the one of the listener at the first position including the second speaker. When the transfer function to the ear is C1, it holds a first tap coefficient equivalent to the transfer function G1 / C1, and the second coefficient memory stores the first tap coefficient from the output section of the first AD converter. The transfer function to the other ear of the listener at the first position including the speaker of G2 is G2, and the other ear of the listener at the first position including the third speaker is output from the output unit of the first AD converter. When the transfer function up to is C2, the transfer function G2 / C2 is It is intended to hold the value of the second tap coefficient, the distance from the first speaker to the listener's head center of the first position r [m], the first and second FIR
The calculation speed of the filter is u [sec / tap], and the speed of sound is c
When [m / sec] is set, the sampling conversion ratio n of the thinning filter is set to the smallest integer n that satisfies the expression (3).

According to a fourth aspect of the present invention, an oscillator for giving a random voice signal to the first AD converter and a crosstalk sound for the listener at the first position are suppressed. Instead, at the time of identification to newly set the tap coefficients of the first and second coefficient memories, an adaptive FIR filter that performs a convolution operation on the signal of the thinning filter and the output of the thinning filter are input as reference signals, First and second microphones that detect a synthesized sound of the crosstalk sound of the first speaker and the control sounds of the second and third speakers in both ears of the listener,
A coefficient updater that inputs the outputs of the first and second microphones as an error signal and updates the tap coefficient of the adaptive FIR filter so that the error input becomes small, and the first and second coefficients set by the coefficient updater Switch the tap coefficient of
Alternatively, the first switch provided to the second coefficient memory, the third interpolation filter for interpolating the output signal of the adaptive FIR filter into the signal of the sampling frequency Fs, and the output of the third interpolation filter depending on the left or right channel And a third switch for switching the outputs of the first and second microphones to output to the coefficient updater. The calculation speed of the first and second FIR filters and the adaptive FIR filter is u [sec / tap], the calculation speed of the coefficient updater is v [sec / tap], and the sound velocity is c [m / se].
c], the sampling conversion ratio n of the thinning filter is set to the smallest integer n that satisfies the expression (2).

According to the invention of claim 5 of the present application, the sampling conversion ratio n of the thinning filter is the frequency with respect to the angle at which both ears of the listener at the first position are viewed from the first speaker provided at the second position. When Fd is a frequency whose characteristic is lower than the sound pressure of the central axis of the first speaker by 6 dB or more,
It is characterized by setting to an integer n that satisfies Expression (4).

[0012]

According to the invention of claim 1 of the present application having such a feature, the number of taps of the FIR filter is converted by converting the audio information given to the speaker at the second position into the sampling frequency Fs / n by the thinning filter. To reduce. Then, the tap coefficient is made equivalent to the transfer function of the crosstalk sound from the second position to the first position. The sampling conversion ratio n is obtained from the acoustic crosstalk path length, the calculation speed of the FIR filter, and the sampling interval. Further, the output of the FIR filter is converted into the original sampling frequency by the interpolation filter, and the sampling frequency of the voice information at the first position is matched and added. In this way, a monaural acoustic signal that cancels the crosstalk sound is generated. Moreover, the crosstalk sound from the adjacent seat can be suppressed by directly using the components for sound reproduction provided for the listener such as the speaker.

Further, according to the invention of claim 2 of the present application, by converting the audio information given to the speaker at the second position into the sampling frequency Fs / n by the thinning filter,
Reduce the number of taps in the adaptive FIR filter. Then, the tap coefficient is determined by the identification operation of the adaptive FIR filter and the coefficient updater so as to be equivalent to the transfer function of the acoustic crosstalk from the speaker at the second position to the listener at the first position. The sampling conversion ratio n is determined by the sum of the calculation time of the path length of the acoustic crosstalk, the adaptive FIR filter, and the coefficient updater, and the sampling interval of the input voice information. In a normal state, the output of the FIR filter in which the transfer function of acoustic crosstalk is set is converted into the original sampling frequency by the interpolation filter, and the sampling frequency of the audio information at the first position is matched and added. In this way, a monaural acoustic signal that cancels the crosstalk sound is automatically generated. Moreover, the crosstalk sound from the adjacent seat can be suppressed by directly using the components for sound reproduction provided for the listener such as the speaker.

Further, according to the invention of claim 3 of the present application, by converting the audio information given to the speaker at the second position into the sampling frequency Fs / n by the thinning filter,
The number of taps of the first FIR filter and the second FIR filter is reduced. Then, the tap coefficient is made equivalent to the transfer function of the crosstalk sound from the second position to the first position. The sampling conversion ratio n is obtained from the acoustic crosstalk path length, the calculation speed of the FIR filter, and the sampling interval. Further, the output of the FIR filter is converted into the original sampling frequency by the interpolation filter, and the sampling frequency of the voice information at the first position is matched and added. This produces a stereo acoustic signal that cancels the crosstalk sound. Moreover, the crosstalk sound from the adjacent seat can be suppressed by directly using the components for sound reproduction provided for the listener such as the speaker.

Further, according to the invention of claim 4 of the present application, by converting the audio information to be given to the speaker at the second position into the sampling frequency Fs / n by the thinning filter,
Reduce the number of taps in the adaptive FIR filter. Then, the tap coefficient is determined by the identification operation of the adaptive FIR filter and the coefficient updater so as to be equivalent to the transfer function of the acoustic crosstalk from the speaker at the second position to the listener at the first position. The sampling conversion ratio n is determined by the sum of the calculation time of the path length of the acoustic crosstalk, the adaptive FIR filter, and the coefficient updater, and the sampling interval of the input voice information. In a normal state, the output of the FIR filter in which the transfer function of acoustic crosstalk is set is converted into the original sampling frequency by the interpolation filter, and the sampling frequency of the audio information at the first position is matched and added. In this way, a stereo acoustic signal that cancels the crosstalk sound is automatically generated. Moreover, the crosstalk sound from the adjacent seat can be suppressed by directly using the components for sound reproduction provided for the listener such as the speaker.

Further, according to the invention of claim 5 of the present application, the number of taps of the adaptive FIR filter is reduced by converting the audio information given to the speaker at the second position into the sampling frequency Fs / n by the thinning filter. Then, the tap coefficient is set by the identification operation of the adaptive FIR filter and the coefficient updater so as to be equivalent to the transfer function of the acoustic crosstalk from the speaker at the second position to the listener at the first position. Further, the transfer function is determined only by the signal within the frequency band where the acoustic crosstalk occurs due to the directivity of the speaker. Sampling conversion ratio n of thinning filter
Is calculated by the ratio of the frequency Fd at which the sound pressure is reduced by 6 dB to the listener at the first position and the Nyquist frequency Fs / 2. The output of the FIR filter is converted to the original sampling frequency by the interpolation filter, and the sampling frequency of the audio information at the first position is matched and added. In this way, a monaural or stereo acoustic signal that cancels the crosstalk sound is automatically generated. Moreover, the crosstalk sound from the adjacent seat can be suppressed by directly using the components for sound reproduction provided for the listener such as the speaker.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An acoustic crosstalk control device according to a first embodiment of the present invention will be described with reference to FIG. Figure 1
FIG. 3 is a block diagram showing the overall configuration of the acoustic crosstalk control device of the first embodiment. In this embodiment, the sampling frequency of the audio signal of the adjacent seat is converted to 1 / n, and the circuit scale of the FIR filter is reduced.

In FIG. 1, an audio input signal L1 is an audio signal output from the first signal generating source to a listener in the adjacent seat, and is first applied to the LPF 10. The sampling frequency for converting an analog audio signal to a digital signal is F
s, the LPF 10 has a Nyquist frequency (Fs /
2) An anti-aliasing filter that attenuates the above high frequencies. The output of the LPF 10 is given to a first AD converter (indicated by A / D in the figure) 11. The AD converter 11 outputs the output of the LPF 10 to the sampling frequency Fs.
It is a circuit that performs digital conversion at (sampling period Ts), and its output is given to the thinning filter 12 and the delay device 13.

The thinning filter 12 is a circuit for converting the sampling frequency of the digital signal into Fs / n (n> 1 is an integer), and its output is given to the FIR filters 14 and 15. Since the first FIR filter 14 transforms the output signal of the thinning filter 12 by the first transfer function,
This is a circuit for convolving the coefficients stored in the first coefficient memory 16. Similarly, the second FIR filter 15 is a circuit for convolving the coefficients stored in the second coefficient memory 17 for converting the output signal of the thinning filter 12 with the second transfer function. The FIR filters 14 and 15 are configured by connecting a delay unit and a multiplier in an M set of ladder shapes,
This number M is called the tap number. FIR filters 14 and 15
Are provided to the first and second interpolation filters 18 and 19, respectively. The interpolation filters 18 and 19 are circuits for interpolating the output signals of the FIR filters 14 and 15, respectively, and converting their sampling frequencies into Fs.

The delay device 13 is a circuit for delaying the digital signal output from the AD converter 11 for a predetermined time, and its output is given to the first DA converter (denoted by D / A in the figure) 20. The DA converter 20 is a circuit for converting the digital output signal of the delay device 13 into an analog signal, and its output is given to the LPF 21. The LPF 21 is a filter for removing high frequency components of the output signal of the DA converter 20, and its output is given to the power amplifier 22. LPF2
The output of No. 1 is power-amplified by the power amplifier 22, and is acoustically converted by the first speaker 4. This sound is heard by the listener in the seat (second position). However, it can also be heard in the ear of the listener P in the present seat (first position).

The left and right channel audio input signals L2 and R2 for servicing the present listener are output from the second signal source and input to the LPFs 23 and 24, respectively. L
The PFs 23 and 24 have the same input audio signal L as the LPF 10.
2 and R2 are anti-aliasing filters excluding high frequencies above the Nyquist frequency, and their outputs are given to the second and third AD converters 25 and 26, respectively. The AD converters 25 and 26 are circuits that digitally convert an input analog signal at a sampling frequency Fs.

The outputs of the AD converters 25 and 26 are given to the adders 27 and 28, respectively. The first adder 27 is AD
It is a circuit that adds the output signal of the converter 25 and the output signal of the interpolation filter 18. The second adder 28 is A
This is a circuit for adding the output signal of the D converter 26 and the output signal of the interpolation filter 19. The outputs of the adders 27 and 28 are given to the second and third DA converters 29 and 30, respectively. DA converters 29 and 30 are adders 2 respectively
It is a circuit for analog-converting the digital signals output by 7, 28, and their outputs are given to the LPFs 31, 32. LPFs 31 and 32 are DA converters 29 and 30, respectively.
Is a filter for removing the high frequency components of the output signal of the above, and their outputs are given to the power amplifiers 33 and 34. The left and right channel audio signals, which are power-amplified by the power amplifiers 33 and 34, respectively, are the second speaker 5L and the third speaker 5R.
Is converted into sound. This sound is heard by the listener P present in the seat.

The operation of the acoustic crosstalk control apparatus of the first embodiment thus constructed will be described. First, the audio signal of the adjacent seat is acoustically converted from the speaker 4 via the delay device 13. This sound leaks to the left ear EL of the listener P through the sound transmission system 35a due to the characteristic of the transfer function G1 (crosstalk). Similarly, the transfer function G is passed through the acoustic transmission system 35b.
The sound leaks to the right ear ER of the listener P with the characteristic of 2.

The transfer function of the audio signal from the input section of the thinning filter 12 to the left ear EL of the listener P via the speaker 5L is C1, and the transfer function of the input section of the thinning filter 12 to the listener 5P via the speaker 5R. Let C2 be the transfer function of the audio signal to the right ear ER. Next, the tap coefficient of the FIR filter 14 is set to G1 / C1 and the FIR filter 15
The tap coefficient of is set to G2 / C2. At this time, the sampling conversion ratio is n, the sampling frequency of the output signal of the thinning filter 12 is Fs / n [Hz], and the sampling interval is nTs [sec].

On the other hand, the number of taps M of the FIR filters 14 and 15 is determined by the distance r [m] between the speaker 4 and the listener P present.
Determined by That is, when the sound velocity is c [m / sec], the time length in the signal time series to be processed by the FIR filters 14 and 15 is r / c. Therefore, the number of taps M
Is given by the following equation (5).

[Equation 5]

On the other hand, the total operation speed of the delay unit and the multiplier for each tap forming the FIR filters 14 and 15 is u
If [sec / tap] is set, two FIs with M taps are set.
The calculation time in the R filters 14 and 15 must be within one sampling interval of the audio signal. Therefore, the following expression (6) is established.

[Equation 6]

By substituting the equation (5) into the equation (6), the following equation (7) is obtained.

[Equation 7] Solving the equation (7) for n, the following equation (8) or (3)
The formula is obtained.

[Equation 8] Usually, the signal processing band is set as wide as possible, and therefore the minimum integer n that satisfies the expression (8) or (3) is the optimum value.

As described above, the optimum sampling conversion ratio n
By determining, the circuit scale of the FIR filters 14 and 15 can be reduced. Although the above configuration and operation have been described with respect to the case where there is one audio signal in the adjacent seat, the same can be applied to the case of a plurality of signals such as stereo. In this case F
When the number of IR filters is K, the sampling conversion ratio n is given by the following expression (9).

[Equation 9]

When only one speaker 5 is provided at the first position (monaural), the FIR filter and the interpolation filter, and the circuit system (LPF, AD converter) for processing the audio information to the listener P at the first position. , DA converter,
LPF, power amplifier) may be one system. In this case, the sampling conversion ratio n is set to a value that satisfies the expression (1).

In the equation (9), the following conditions are set as an example. Sampling frequency Fs = 48KHz FIR calculation speed u = 100ns
ec Distance between speaker 4 and listener P r = 2 m Number of FIR filters (in case of stereo) K = 4 In this case, n> 2.3
2, and n = 3 is the sampling conversion ratio.

Next, an acoustic crosstalk control device according to the second embodiment of the present invention will be described with reference to FIG. FIG. 2 is a block diagram showing a part of the configuration of the acoustic crosstalk control device of the second embodiment. FIG. 2 includes the same FIR filters 14 and 15 (not shown) as in FIG. 1 and coefficient memories 16 and 17 connected to them, and can be adaptively set according to the acoustic crosstalk of the present seat and the adjacent seat. The same parts as those in the first embodiment are designated by the same reference numerals, and newly added circuit portions are designated by different reference numerals.

In the present embodiment, the sampling frequency of the audio signal of the adjacent seat (second position) is converted into 1 / n, and the adaptive F
It is configured to reduce the circuit size of the IR filter. In FIG. 2, an oscillator 40 is a circuit that generates a random signal, and is switched with an input audio signal (not shown) so that the LPF
Input to 10. The LPF 10 is an anti-aliasing filter that attenuates high frequencies above the Nyquist frequency of random signals.

The AD converter 11 is a circuit for digitally converting the output of the LPF 10 at the sampling frequency Fs, and its output is given to the thinning filter 12 and the delay device 13. The thinning filter 12 is a circuit that converts the sampling frequency of a digital signal into Fs / n (n is an integer), and its output is an adaptive FIR (AFIR) filter 41.
Given to. The AFIR filter 41 is a circuit that convolves the output signal of the thinning filter 12 with the first and second transfer functions and adaptively performs conversion processing to generate a control signal.
Similar to the FIR filters 14 and 15, it is configured to include M sets of delay devices and multipliers. The amplification factor of each multiplier is called a tap coefficient, and the optimum tap coefficient is stored in the coefficient memories 16 and 17.

The coefficient updater 42 connected to the AFIR filter 41 inputs the reference signal from the thinning filter 12, and uses the LMS (least mean square) method or the like to make the sound of the oscillator 40 the smallest at the position of the listener P. It is a circuit that finds the optimum tap coefficient so that the tap coefficient of the AFIR filter 41 is updated. The tap coefficient set by the AFIR filter 41 is stored in one of the coefficient memories 16 and 17 via the first switch 43. The switch 43 is used for the coefficient memory 16 when processing the left channel audio signal.
It is a switch for switching to the side, and for processing the audio signal of the right channel, it is switched to the side of the coefficient memory 17.

The output of the AFIR filter 41 is given to the third interpolation filter 18a. The interpolation filter 18a is a circuit that interpolates an input signal and converts it into a sampling frequency Fs. The interpolation filter 18a may also be used as the interpolation filter 18 of FIG. The second switch 44 is a circuit that switches the signal of the interpolation filter 18a and supplies its output to the DA converter 29 or the DA converter 30. The switch 44 connects to the DA converter 29 in the first signal processing, and connects to the DA converter 30 in the next signal processing.
The output signal of the DA converter 29 is given to the speaker 5L via the LPF 31 and the power amplifier 33 as in the first embodiment, and the output signal of the DA converter 30 is also the LPF 32.
It is given to the speaker 5R via the power amplifier 34.

The control sound emitted from the speaker 5L and the crosstalk sound emitted from the speaker 4 enter the error detection microphone 45. Further, the control sound emitted from the speaker 5R and the crosstalk sound emitted from the speaker 4 enter the error detection microphone 46. The error detection microphones 45 and 46 are monitoring microphones that detect the sound pressures incident on the left ear EL and the right ear ER of the listener P as error sounds.

The outputs of the error detection microphones 45 and 46 are input to the third switch 47. The switch 47 is a switch that switches to the error detection microphone 45 when processing a left channel audio signal, and switches to the error detection microphone 46 when processing a right channel audio signal. The output of the switch 47 is input to the coefficient updater 42 described above as an error signal. Above switches 43, 4
4 and 47 operate in conjunction with each other, and switching continues until the identification operation is completed.

The operation of the acoustic crosstalk control device of the second embodiment having the above configuration will be described. AFIR
In order to perform the identification operation of the filter 41, first, a random audio signal is output from the oscillator 40. This signal is LPF
10, the AD converter 11, the delay device 13, the DA converter 21, and the power amplifier 22, and then given to the speaker 4.
Sound is converted. The sound of the speaker 4 is the sound transmission system 35a.
Then, the light enters the error detection microphone 45 via. The transfer function of the path from the input unit of the delay device 13 to the error detection microphone 45 passing through the speaker 4 is G1, and the acoustic transmission system 48a from the speaker 5L to the error detection microphone 45 is used. If the transfer function from the input unit of the thinning filter 12 to the error detection microphone 45 via the speaker 5L and the acoustic transmission system 48a is C1, G1 / C1 is set in the coefficient memory 16.

Similarly, the transfer function from the input section of the delay device 13 to the error detection microphone 46 via the speaker 4 and the sound transmission system 35b is set to G2, and the sound transmission system from the speaker 5R to the error detection microphone 46 is set to 48b.
And If the transfer function from the input section of the thinning filter 2 to the error detection microphone 46 via the speaker 5R and the acoustic transmission system 48b is C2, G2 / C2 is set in the coefficient memory 17. This coefficient memory 16, 17
The contents of the above are held and the same signal processing as in the first embodiment is performed.

The sampling frequency of the output signal of the thinning filter 12 is Fs / n [Hz], and the sampling interval is nT.
s [sec]. Further, the tap number M of the FIR filters 14 and 15 and the AFIR filter 41 is determined by the distance r [m] between the speaker 4 and the listener P present at the present seat. Also in this case, the expression (5) is established as in the first embodiment.

The computation time of the circuit portion that executes the convolution operation of the AFIR filter 41 is u [sec / tap], and the computation time of the circuit portion that executes the operation of the coefficient updater 42 is v [sec / tap]. Sound velocity is c [m / sec]
As long as the sum of these two calculation times is within one sampling interval after thinning. This condition is
It is given by the equation (0).

[Equation 10] Therefore, the following equation (11) is obtained from the equations (5) and (10).

[Equation 11] By solving the equation (11) for n, the following equation (12) or equation (2) is obtained.

[Equation 12]

Normally, the bandwidth of signal processing is set as wide as possible, and therefore the minimum integer n satisfying the expression (12) or (2) is the optimum value. As described above, according to the second embodiment, the circuit scale of the AFIR filter 41 and the coefficient updater 42 can be increased by obtaining the optimum sampling conversion ratio n even when the calculation amount in the identification processing exceeds the calculation amount in the normal time. Can be made smaller.

Further, according to the present embodiment, when the speaker 5 at the first position and the speaker 4 at the second position are replaced with ones having other characteristics or the characteristics of the acoustic space of the listener P change. , FIR filters 14 and 15 can be installed only by installing error detection microphones 45 and 46 at the time of system identification.
The tap coefficient of can be easily reset. If a monaural sound is reproduced at the first position, the switches 43, 4
No need for 4,47.

Next, an acoustic crosstalk control apparatus according to the third embodiment of the present invention will be described with reference to FIGS. 1, 3 and 4. In this embodiment, the sampling frequency of the audio signal is converted to 1 / n, and the circuit scale of the FIR filter is further reduced. The block structure is the first
The configuration is the same as that of the embodiment, and the same parts are omitted for the configuration and operation.

FIG. 3 is a graph of sound pressure frequency characteristics of the speaker 4, and FIG. 4 is a graph showing directivity. As shown in FIG. 3 and FIG. 4, the frequency characteristic with respect to the angle θd ° at which the listener P present at the seat is seen from the speaker 4 is lower in the high range than 0 ° on the central axis. Therefore, the influence of acoustic crosstalk is reduced in the high frequency range, and it is not necessary to control the listener P present in the seat. Therefore, from the viewpoint of hearing, the frequency that falls by 6 dB or more from the frequency characteristic on the central axis is set to Fd.
In such a case, the acoustic crosstalk may be controlled in the low frequency range below the frequency Fd where the leakage is large. Therefore, the Nyquist frequency Fs / which is the upper limit of the output of the thinning filter
(2n) may be set to Fd or more.

From this, the following equation (13) is established.

[Equation 13] As an example, sampling frequency Fs = 48K
When the directivity frequency of Hz −6 dB Fd = 5 KHz, n <4.8, and n = 4 is the sampling conversion ratio. As described above, according to the third embodiment, the crosstalk suppressing process is not necessary for the high frequency acoustic signal having high directivity. Therefore, the optimum sampling conversion ratio n becomes large, and the circuit scale of the FIR filters 14 and 15 can be further reduced.

[0047]

As described above, according to the invention described in claim 1 of the present application, the sampling interval of the thinning filter is obtained from the path length of the acoustic crosstalk and the calculation speed of the FIR filter. The sampling conversion ratio n is determined from this value, and the convolution calculation of the crosstalk sound input to the listener at the first position by the FIR filter is performed. This way F
The number of taps of the IR filter is reduced, and the circuit scale is greatly reduced. The output of the FIR filter is converted to the original sampling frequency by the interpolation filter and added to the regular voice information to be given to the listener at the first position.
In this way, monaural sound without crosstalk is output from the second speaker. Therefore, the listener can clearly hear his / her own voice information regardless of the presence or absence of the sound radiated from another adjacent speaker.

According to the invention of claim 2 of the present application, each speaker and each position are corresponding to the installation positions of the first and second speakers by using the random voice signal from the oscillator and the adaptive FIR filter. The acoustic transfer function between listeners can be adaptively obtained. In this case as well, similar to the effect of the invention of claim 1, the number of taps of the FIR filter and the adaptive FIR filter is reduced, and the circuit scale is greatly reduced. The tap coefficient once set is held in the coefficient memory and is used to suppress the crosstalk sound that normally occurs. And even if you change the sound reproduction environment such as the position of the speaker, you can easily set the tap coefficient corresponding to this.
The crosstalk sound from the close position can be automatically controlled.

According to the invention of claim 3 of the present application, the sampling interval of the thinning filter is obtained from the path length of the acoustic crosstalk and the calculation speeds of the first and second FIR filters. The sampling conversion ratio n is determined from this value, and the convolution calculation of the crosstalk sound input to the listener at the first position is performed by the first and second FIR filters. In this way, the number of taps of the FIR filter is reduced, and the circuit scale is greatly reduced. The output of each FIR filter is converted to the original sampling frequency by the interpolation filter and added to the regular voice information to be given to the listener at the first position. In this way, stereo sound without crosstalk is output from the second and third speakers. Therefore, the listener can clearly hear his / her own voice information regardless of the presence or absence of the sound radiated from another adjacent speaker.

According to the invention of claim 4 of the present application, corresponding to the installation positions of the first to third speakers, the random voice signal from the oscillator and the adaptive FIR filter are used, and each speaker and each position are arranged. The acoustic transfer function between listeners can be adaptively obtained. In this case as well, similar to the effect of the invention of claim 1, the number of taps of the FIR filter and the adaptive FIR filter is reduced, and the circuit scale is greatly reduced. The tap coefficient once set is held in the first and second coefficient memories, and is used for suppressing the crosstalk sound that normally occurs. Even if the sound reproduction environment such as the position of the speaker is changed, the tap coefficient corresponding thereto can be easily set, and the crosstalk sound from the close position can be automatically controlled.

According to the invention of claim 5 of the present application,
Only the low frequency band where the crosstalk sound is generated due to the directivity of the speaker is the signal processing target of the FIR filter. Therefore, in addition to the effect of the invention of claim 1 or 3,
The sampling conversion ratio n of the thinning filter is further increased, and the number of taps of the FIR filter can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of an acoustic crosstalk control device according to first and third embodiments of the present invention.

FIG. 2 is a block diagram showing a partial configuration of an acoustic crosstalk control device according to a second embodiment of the present invention.

FIG. 3 is a directional characteristic for each frequency of the speaker in the third embodiment.

FIG. 4 is a directional characteristic diagram of a speaker according to a third embodiment.

FIG. 5 is a block diagram showing a configuration of a conventional acoustic crosstalk control device.

[Explanation of symbols]

4,5R, 5L Speaker 10,21,23,24,31,32 Low pass filter 11,25,26 AD converter 12 Decimation filter 13 Delay device 14,15 FIR filter 18,18a, 19 Interpolation filter 20,29,30 DA Converter 22, 33, 34 Power amplifier 27, 28 Adder 35a, 35b, 48a, 48b Acoustic transmission system 40 Oscillator 41 Adaptive AIR filter 42 Coefficient updater 43, 44, 47 Switch 45, 46 Error detection microphone EL Left ear ER right Ear P listener

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication location H03H 17/02 L 8842-5J 21/00 8842-5J H04R 3/00 310 (72) Inventor Tamura Tadashi 1006, Kadoma, Kadoma-shi, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. (72) Inventor, Intermediate interest: 1006, Kadoma, Kadoma-shi, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (5)

[Claims] 1. When providing different audio information to a plurality of adjacent listening positions, the first position is provided to the listener at the first position.
An acoustic crosstalk control device for suppressing a crosstalk sound radiated from a second position close to the second position, which converts the audio information given to the listener at the second position from analog to digital at a sampling frequency Fs. A first AD converter, a thinning filter that thins out the output of the first AD converter at a frequency of Fs / n (n> 1), and an FIR filter that inputs a signal of the thinning filter and performs a convolution operation. A coefficient memory for storing tap coefficients of the FIR filter, and an output signal of the FIR filter for sampling frequency F
a first interpolation filter for interpolating the s signal, a delay device for delaying the output of the first AD converter for a predetermined time, a first DA converter for analog-converting the output signal of the delay device, a second A first speaker provided at a position for converting the output of the first DA converter into an acoustic signal, and a second AD for performing analog / digital conversion of audio information given to the listener at the first position at a sampling frequency Fs. A converter, an adder for adding the signal of the first interpolation filter and a signal of the second AD converter, a second DA converter for analog-converting the output of the adder, and the second DA converter The output of is converted into an acoustic signal,
A second speaker provided at a second position, wherein the coefficient memory includes the first speaker and the ear of the listener at the first position from the output unit of the first AD converter. Is defined as G, and the transfer function from the output part of the first AD converter to the ear of the listener at the first position including the second speaker is defined as C. Equivalent to the tap coefficient is held, the distance from the first speaker to the head center of the listener at the first position is r [m], and the calculation speed of the FIR filter is u [sec / tap]. ], And the sound velocity is c [m / sec], the sampling conversion ratio n of the thinning filter is expressed by the following equation (1): An acoustic crosstalk control device characterized by setting to a minimum integer n that satisfies 2. An oscillator for providing a random audio signal to the first AD converter, and a random signal of the oscillator is used instead of audio information to suppress a crosstalk sound to a listener at the first position. At the time of identification in which a tap coefficient of the coefficient memory is newly set, an adaptive FIR filter that performs a convolution operation on the signal of the thinning filter and the output of the thinning filter are input as a reference signal, and the crosstalk sound of the first speaker is input. And a microphone for detecting a synthesized sound at a listening position of the control sound of the second speaker and an output of the microphone as an error signal, and updating the tap coefficient of the adaptive FIR filter so as to reduce the error input. And an output signal of the adaptive FIR filter for interpolating into a signal of sampling frequency Fs 2 interpolation filter, the calculation speed of the FIR filter and the adaptive FIR filter is u [sec / tap], the calculation speed of the coefficient updater is v [sec / tap], and the sound speed is c [m / m]. sec], the sampling conversion ratio n of the thinning filter is expressed by the following equation (2): The acoustic crosstalk control device according to claim 1, wherein a minimum integer n that satisfies the above is set. 3. When providing different audio information to a plurality of adjacent listening positions, the first position is provided to the listener at the first position.
An acoustic crosstalk control device for suppressing a crosstalk sound radiated from a second position close to the second position, which converts the audio information given to the listener at the second position from analog to digital at a sampling frequency Fs. No. 1 AD converter, a decimation filter for decimation of the output of the first AD converter at a frequency of Fs / n (n> 1), and a signal of the decimation filter, and convolution with a first tap coefficient. A first FIR filter that performs an operation, a first coefficient memory that stores the first tap coefficient and gives the first FIR filter, and a signal of the thinning filter is input, and a second tap coefficient is used. A second FIR filter for performing a convolution operation; a second coefficient memory for storing the second tap coefficient and giving it to the second FIR filter; A first interpolation filter that interpolates the output signal of the first FIR filter into a signal of the sampling frequency Fs; a second interpolation filter that interpolates the output signal of the second FIR filter into the signal of the sampling frequency Fs; A delay unit for delaying the output of the AD converter for a predetermined time, a first DA converter for converting the output signal of the delay unit into an analog signal, and a second DA converter provided at a second position for converting the output of the first DA converter into an acoustic signal. To the listener at the first position, a second AD converter for performing analog-digital conversion of the audio information of either the left or right channel given to the listener at the first position at the sampling frequency Fs, and the first position A third AD converter for analog-digital converting the audio information of the other left and right channels given to the listener at the sampling frequency Fs. Data, a first adder that adds the signal of the first interpolation filter and the signal of the second AD converter, a signal of the second interpolation filter, and a signal of the third AD converter And a second DA for analog-converting the output of the first adder.
A converter and a third DA for analog-converting the output of the second adder
A converter, and converting the output of the second DA converter into an acoustic signal,
A second speaker provided on one of the left and right of the second position, and an output of the third DA converter, which is converted into an acoustic signal,
A third speaker provided on the other of the left and right of the second position, wherein the first coefficient memory includes the first speaker from the output section of the first AD converter. The transfer function from the output of the first AD converter to one ear of the listener at the first position is set to G1 and the transfer function to the one ear of the listener at the first position is set to G1. Let C1 be the transfer function G1
/ C1 holds a first tap coefficient equivalent to the / C1, and the second coefficient memory includes a first speaker from the output section of the first AD converter, and a second position of the listener at the first position. Let G2 be the transfer function to the other ear, and from the output of the first AD converter,
When the transfer function to the other ear of the listener at the first position including the third speaker is C2, the transfer function G2 / C
A second tap coefficient equivalent to 2 is held, and the distance from the first speaker to the head center of the listener at the first position is r [m], the first and second FIR The calculation speed of the filter is u [sec / tap], and the speed of sound is c [m
/ Sec], the sampling conversion ratio n of the thinning filter is expressed by the following equation (3): An acoustic crosstalk control device characterized by setting to a minimum integer n that satisfies 4. An oscillator for providing a random audio signal to the first AD converter, and a random signal of the oscillator is used instead of audio information to suppress a crosstalk sound to a listener at the first position. At the time of identification for newly setting the tap coefficients of the first and second coefficient memories, an adaptive FIR filter that performs a convolution operation on the signal of the thinning filter and the output of the thinning filter are input as reference signals, and the first The first and second microphones for detecting a synthesized sound in both ears of the listener of the crosstalk sound of the speaker and the control sound of the second and third speakers, and the outputs of the first and second microphones. The adaptive FI is input as an error signal so that the error input becomes small.
A coefficient updater for updating the tap coefficient of the R filter; a first switch for switching between the first and second tap coefficients set by the coefficient updater and giving the first or second coefficient memory; A third interpolation filter for interpolating the output signal of the adaptive FIR filter into a signal of the sampling frequency Fs, and switching the output of the third interpolation filter according to the left or right channel, and switching to the second or third DA converter. A second switch for providing audio information; and a third switch for switching the outputs of the first and second microphones and outputting to the coefficient updater, the first and second FIR filters The calculation speed of the adaptive FIR filter is u [sec / tap], the calculation speed of the coefficient updater is v [sec / tap], and the sound velocity is c [m.
/ Sec], the sampling conversion ratio n of the thinning filter is set to the smallest integer n that satisfies the equation (2). 5. The sampling conversion ratio n of the thinning filter has a frequency characteristic with respect to an angle at which the first speaker provided at the second position looks into both ears of a listener at the first position. 6 dB from the sound pressure of the central axis of the first speaker
When Fd is the frequency that decreases more than the above, the following equation (4) The acoustic crosstalk control device according to claim 1 or 3, wherein the integer n is set to satisfy the above condition.