JPH0252886B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides a recording system for audio signals such as music signals and voice signals on recording media such as magnetic tapes and records, and/or an acoustic space for audio signals recorded in advance on a recording medium. The present invention relates to an audio signal transmission circuit in a reproduction system, and its purpose is to provide an audio signal transmission circuit capable of recording and/or reproducing source sound quality with high fidelity for a listener.

Generally, a Bode type control means for changing frequency amplitude characteristics is employed in a transmission line in an audio signal reproduction system as a means for changing sound quality. However, with such conventional control methods, the sound pressure at the listening position is flat, but the structure of the speaker itself is not a minimum phase shift system due to various resonances, etc., so it becomes necessary to correct the level characteristics. Even if this level characteristic is corrected, the waveform will not be transmitted correctly due to phase distortion.
There was a problem in that the source sound quality could not be reproduced with high fidelity for the listener. Further, by correcting in the frequency domain, there is a problem that the mutual time relationship cannot be accurately controlled when correcting a plurality of points. The present invention solves the above problems by providing a convolution integration circuit in the audio signal transmission system of the impulse response waveform h(t) at the listening position.
Impulse response of a specific characteristic at the listening position f ₀
(t) and the weighting coefficient g of the above convolution integration circuit
(n), the expanded matrix H obtained from h(t), its transposed matrix H ^T , and the matrix F 0 with f ₀ (t) as one column, which satisfies HH ^T G = H ^T _{F 0 .} Each element of a matrix G consisting of columns is set to g(n).
In the present invention, by finding a solution that satisfies the determinant using the above-described configuration, an arbitrary response waveform can be uniquely obtained at the listening position on the time axis.

The present invention will be described below with reference to drawings of embodiments.

FIG. 1 shows an embodiment of the present invention.
In the figure, 1 is a delay circuit having a large number of output terminals, 2a to 2n are multipliers that control weighting coefficients for a plurality of outputs from the delay circuit 1, and 3 is an adder. A convolution/integration circuit (hereinafter referred to as a convolver) 5 is configured to vary the transmission characteristics of an audio signal applied to the convolver. 6 is a power amplifier, and 7 is a speaker.

In the transmission circuit having such a configuration, in order to obtain a weighting coefficient that minimizes the square of the difference between the impulse responses obtained at the input terminal 4 of the convolver 5 and the listening position 8 at the listening position 8, the multipliers 2a to 2 are used. It is configured to change the value of 2n. In such a state, the discrete coefficient of the impulse response of the convolver 5 is expressed as g ₁ ,
If g ₂ ...g _n-1 , the discrete response at listening position 8 is
f ₀ , f ₁ ... f _o+n-2 is expressed by the following formula.

f _P = _o+n-2 〓 ⁱ⁼⁰ higi-p (1) However, hi is the transfer characteristic p=0...n+m-2 Expressing equation 1 as a matrix, Therefore, Equation 2 can be further expressed as F=HG. Here, input impulse F ₀ and winding position 8
^Taking _the square of _{the difference} in impulse response _F ^at G ^T H ^T −F ^T ₀ ) (HG−F ₀ ) =G ^T H ^T HG−F ^T ₀ HG−G ^T H ^T F ₀ +F ^T F ₀ ...(3), and the evaluation function P is the minimum. When the impulse response G of the convolver 5 is determined, ∂ _P /∂ _G =2H ^T HG−2H ^T F ₀ (4) where T indicates a transposed matrix.

Then, from equation 4, H ^T HG = H ^T F ₀ (5).

That is, the transmission characteristics are changed by setting the coefficients of the multipliers 2a to 2n as shown in Equation 5 above,
The sound pressure frequency characteristics at the listening position become flat,
It is also possible to reproduce the source sound quality with high fidelity.

In the above explanation, the case where the characteristics of the transmission line in the reproduction system is changed has been explained, but the transmission characteristic variable means 5 is provided in the transmission line in the recording system for recording the audio signal from the microphone on the recording medium. can also be implemented in the same manner and produce similar effects. In this case, the weighting coefficient of the convolver provided in the transmission path in the recording system may be calculated and controlled by comparing the sound from the microphone with the output sound of the recording system composed of a recording medium.

FIG. 2 shows another embodiment of the present invention. In FIG. 2, 11 is a left speaker, 12 is a right speaker, 13 is a convolver provided in the signal transmission path of the left speaker 11, and 14 is a right speaker. The convolver 15 provided in the signal transmission path of the speaker 12 is connected to the input end, 1
6 is the listener. Here, convolver 13,1
4 is constructed in the same manner as in the case of FIG. 1 described above.

In such a configuration, let the input waveforms to the front left and right speakers 11 and 12 be i _L , i _R , the impulse responses h ₁ and h ₂ , and the input waveform to the virtual speaker 17 at the desired location be i _S impulse response h ₃ ,
h ₄ , the sound reaching both ears of the listener 16 from the front left and right speakers 11 and 12 is g L , where g _L is the sound reaching the left ear, and g _R is _the sound reaching the right ear. =h ₁ *i _R +h ₂ *i _L ...(6) g _R =h ₂ *i _R +h ₁ *i _L ...(7) and further from the virtual speaker 17 to the listener 16
The sound reaching both ears is the sound reaching the left ear f _L ,
If the sound reaching the right ear is f _R , then f _L = h ₄ *i _S ……(8) f _R = h ₃ *i _S ……(9).

If the impulse responses of the convolvers 13 and 14 provided in the signal transmission paths of the left and right speakers 11 and 12 are a and b, respectively, then the sound at the left and right ears of the listener 16 is g _L = h ₂ * b + h ₁ * a ...(10) g _R = h ₁ * b + h ₂ * a ... (11). Also, assuming that i _S at this time is an impulse response, f _L = h ₄ ... (12) f _R = h ₃ ... (13), and from equations 10, 11, 12, and 13, g _L By adjusting the convolvers 13 and 14 so that ≒f _L and g _R ≒ f _R , the two speakers 11 and 12 can reproduce sound as if the sound source (virtual speaker 17) is at the desired position. can.

Here, (i) the impulse responses of h ₁ , h ₂ , h ₃ and h ₄ are each n discrete impulses, (ii) the impulse responses of a and b are m discrete impulses, and further evaluation is performed. Let the function P be the impulse h ₁ , h ₂
Let it be the square of the sound pressure difference between h ₃ and h ₄ . That is, the evaluation function P is P= _o+n-1 〓 ⁱ⁼¹ (g _Ri −f _Ri ) ² + (g _Li −f _Li ) ² (14).

Here, formulas 10 and 11 are changed to the following formula.

Moreover, the above equations 14 and 15 can be rearranged as follows.

Furthermore, equation 16 can also be expressed as the following equation.

g=HX (17) Also, the vector d consisting of impulses h ₃ and h ₄ is defined as follows.

Therefore, the evaluation function P is P=(g-d) ^t (g-d) ...(18) Therefore, P=(X ^t H ^t -d ^t )(HX-d) =X ^t H ^t HX-( X ^t H ^td+dt HX)+d ^t d Here, the constants of the convolution integration circuits a and b are determined by finding the extreme value of X that makes the evaluation function P 0. That is, ∂P/∂X=2H ^t HX−2H ^t d=0 Therefore, H ^t HX=H ^t d (19) By finding X, the constants of a and b are determined.
By controlling the convolvers 13 and 14 as described above,
The sound source can be moved to any position using the two speakers 11 and 12, and the source sound quality can be reproduced with high fidelity.

FIG. 3 shows a specific embodiment of the present invention. In FIG. 3, the output of the pulse generator 21 is passed through the switch 22 as the output of the terminal a or b, and is input to the amplifier 23 or the convolution integral Connected to circuits 24 and 25. The convolution integration circuits 24 and 25 are connected to speakers 28 and 29 via amplifiers 26 and 27, and audio signals are transmitted to microphones 31 and 32 placed at the entrances of both ears of a listener 30. Here, when the contact point of the switch 22 is set to the a side, the pulse signal is emitted from the speaker 33 via the amplifier 23, and the binaural microphone 31 of the listener 30,
Detected from 32. The signal is then recorded as data into a computer 36 via a microphone amplifier 34 and an analog-to-digital converter 35. On the other hand, when the switch 22 is set to the contact b side, the pulse signal passes through convolution and integration circuits (hereinafter referred to as convolvers) 24, 25, amplifiers 26, 27, and outputs sound from the speakers 28, 29, as described above. Similarly, the data is detected by microphones 31 and 32 set in both ears of the listener 30, and is set to a computer 36 via an amplifier 34 and an analog-to-digital converter 35. Here, the computer 36 calculates each element of the matrix X shown in Equation 19, and the address and data are connected as the computer output to the aforementioned convolvers 24 and 25 via the buffer circuit 37. At this time, the convolvers 24 and 25 are configured to be able to take out multiple output signals after a certain delay time with respect to, for example, an analog input signal, and each of the multiple output terminals is connected to a digital converter and an analog converter. configured to have. In this case, the digital-to-analog converter for each tap output may be selected using the address signal described above, and the level of the analog delay signal may be controlled using the data signal. Another method may be to convert the analog input into digital and then use a shift register and a digital multiplier, for example.

Moreover, the same thing can be considered even if the listener has a condensed head for sound collection.

FIG. 4 shows another embodiment of the present invention. In FIG. 5, 41 is an audio band generator such as a tape recorder, 42 is a preamplifier, and 43 and 44 are convolvers. Here, the audio signal is
The speaker 4 is powered by power amplifiers 45 and 46 via a preamplifier 42 and convolvers 43 and 44.
Obtain power to drive 7 and 48. At this time, the convolvers 43 and 44 are connected to the speakers 47 and 44 as described above.
48 and the listener 49, the load of each tap output of the convolver is recorded in advance in the memory 50.
The loading coefficients of each convolver 43, 44 at the position of an arbitrary sound image to be synthesized are recorded. Here, the potential of VRef with respect to the reference voltage is detected by, for example, a three-dimensionally changing panpot 51, which is connected to an analog-to-digital converter 52, and then to a decoder 53. The decoder 53 detects from the memory 50 the position corresponding to the digital signal detected by the analog-to-digital converter 52 and the load coefficients of the convolvers 43 and 44 for obtaining the same synthesized sound image, and selects the taps of each convolver 43 and 44. The coefficients are changed to localize the sound image to a desired position.

Note that the three-dimensional panpot 51 can also be achieved by rotating the joystick bar of a normal two-dimensional panpot.

Furthermore, the same problem may occur even if headphones are used instead of the speakers of this embodiment.

As described above, according to the present invention, a convolution integrator is provided in the audio signal transmission path, and the weighting coefficients of the convolution integrator are the enlarged matrix H generated from the listening position response waveform and its transposed matrix H ^T HH ^T G by the matrix F ₀ with the target response waveform in one column
The feature is that the value of each element of the matrix G consisting of one column that satisfies = H ^T F ₀ is set.
It is possible to find the optimal solution in the time domain simply and uniquely, rather than adaptively, without using a frequency analysis means such as a Fourier transformer to find the inverse characteristics of the two characteristics of amplitude and phase in the frequency domain. Furthermore, since the delay time of the time axis response waveform can be clearly specified, it is effective in accurately controlling the mutual time relationship of the response waveforms at multiple points.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the audio signal transmission circuit of the present invention, FIG. 2 is a block diagram showing another embodiment of the audio signal transmission circuit of the present invention, and FIGS. FIG. 6 is a block diagram showing still another embodiment of the audio signal transmission circuit of the invention. 5, 13, 14, 24, 25, 43, 44...
Convolution integral circuit.

Claims (1)

[Claims] 1 Impulse response waveform h(t) at listening position
A convolution integrator circuit is installed in the audio signal transmission system, and the enlarged matrix H obtained from the impulse responses f ₀ (t) and h(t) with specific characteristics at the listening position and its transposed matrices H ^T and f ₀ (t) are calculated. A matrix G consisting of one column that satisfies HH ^T G = H ^T F ₀ by the matrix F ₀ with one column
The weighting coefficient g of the convolution integrator circuit is
(n) An audio signal transmission circuit characterized by having the following settings.