JPH1051890A - Audio signal transmission circuit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a coefficient arithmetic unit for a convolver to obtain a coefficient of the convolver provided to the audio signal transmission circuit synchronizing signal in which distortion of a sound image of a high sound frequency especially having been unavoidable by a speaker and a headphone is eliminated to allow the user to enjoy a natural audio signal. SOLUTION: Changeover devices 7L, 7R are controlled to actually measure a response characteristic of speakers 4L, 4R thereby obtaining a correction filter coefficient of convolvers 2L, 2R and the obtained coefficient is given to the convolvers 2L, 2R, in which convolution is conducted. Thus, in the case of controlling correction of the speakers, an impulse response f0(t) with a specific characteristic depends on the speaker characteristic and the filter coefficient of the convolvers 2L, 2R is selected so that a prescribed medium frequency band has a flat characteristic and only peak levels at a higher frequency band are suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio signal transmission circuit, and more particularly, to an audio signal transmission circuit which corrects the response characteristics of a speaker of a high-fidelity audio device and reproduces a faithful original sound quality. The present invention relates to a convolver coefficient calculating device for calculating a convolver coefficient.

[0002]

2. Description of the Related Art Conventionally, in a speaker system,
By using a digital filter, the sound pressure and group delay characteristics of the speaker system are flattened, the response characteristics of the speaker are corrected, and a device for reproducing faithful original sound quality has been made.

[0003]

When the sound pressure / group delay characteristics of the loudspeaker system are flattened, the following three factors are the main factors preventing the flattening. 1) Factors due to diffraction and reflection of the speaker cabinet. 2) Factor due to phase rotation by analog filter. 3) A factor due to the divided vibration of the diaphragm (including the surrounding support system) in the speaker unit. Among them, the factors 1) and 2) can be corrected without much problem.

However, the factor 3) is a very difficult correction. The divisional vibration that causes the factor 3) tends to occur particularly in a high frequency band. In such a high frequency, the divisional vibration occurs according to each unique vibration mode of the speaker cone, and undulations and the like occur. Complicated vibration occurs. It is difficult to suppress such complicated vibrations simply by averaging the sound pressure levels. Therefore, on the high frequency side for viewers,
There was a problem that high fidelity sound sources could not be reproduced. Therefore, an object of the present invention is to provide an audio signal transmission circuit that solves such a problem.

[0005]

The present invention has been made based on the following basic concept. Ideally, a speaker uses a band in which the diaphragm vibrates as a unit.In practice, however, there are many cases in which a band that vibrates with peaks and dips must be used. ,
The above factors 1) and 2) are considered to be bands where the split vibration hardly occurs and the entire cone vibrates, and can be solved by averaging peaks and dips.

Next, above the problematic band, since the divided vibrations mentioned in the above 3) occur, the speaker cone vibrates as a whole, and furthermore, undulations occur due to the respective natural vibrations. Therefore, simply averaging the peaks and dips as in the solutions 1) and 2) above may cause complicated undulations. Therefore, in this band, the idea is to suppress only the sound pressure above the average level and leave the sound pressure below the average level as it is, to prevent the occurrence of complicated undulations and achieve a substantially flat level. It is.

As a means for solving the problem, a convolver for convolutionally processing a signal from a sound source in accordance with a set coefficient and outputting the convolver, and a predetermined measurement system including a transducer are used. Signal to give an impulse response waveform h previously measured at a measurement location near the transducer.
Based on (t), the impulse response f0 of the specific characteristic of the impulse response waveform h (t) at the measurement position
In the predetermined middle band of (t), the band lower than the predetermined band is flattened, and only the level higher than the predetermined level in the higher band is suppressed to the predetermined level, and the calculated inverse filter coefficient is calculated. And a coefficient supply unit for supplying the audio signal to the convolver.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described with reference to preferred embodiments. FIG. 1 is a configuration diagram showing one embodiment of the audio signal transmission circuit of the present invention. In FIG. 1, sound sources 1L and 1R of a two-channel stereo system are predetermined audio signal sources. Convolvers 2L and 2R for correcting the response characteristics of the speakers to simultaneously correct the amplitude and phase characteristics, and the amplifiers 3L and 3R are provided between the speakers 4L and 4R and the sound sources 1L and 1R.

Reference numeral 15 denotes switching control of a switch, which will be described later, to measure the response characteristics of the speakers 4L and 4R to obtain a correction filter coefficient of the convolver, and to apply the obtained correction filter coefficient to the convolvers 2L and 2R to perform convolution operation. This is a control unit that controls so as to correct the speaker.

A memory 16 is provided for storing a correction filter coefficient obtained based on the actual measurement of the response characteristics of the speakers 4L and 4R. The response characteristics of the speakers 4L and 4R are actually measured based on the control of the control unit 15. Sometimes, the convolvers 2L and 2R are separated from the audio signal transmission path, and switches 7L and 7R are provided for switching the audio signal transmission path so as to provide the convolvers 2L and 2R when correcting the response characteristics of the speakers 4L and 4R. Have been.

That is, the configuration shown in FIG.
Measure the impulse response of the speakers 4L, 4R.
In order to cancel out the R characteristic and correct it to a flat characteristic, the filter coefficients of the convolvers 2L and 2R provided in the transmission line of the audio signal are calculated and controlled, so that the amplitude and phase are simultaneously corrected and the predetermined band of the transmission characteristic is obtained. Only at a constant level to improve the sound quality and reproduce the faithful original sound quality.
For example, distortion of a sound image, which cannot be avoided by a speaker or headphones, is removed so that a natural audio signal can be enjoyed.

Here, the filter coefficients of the convolvers 2L and 2R are calculated as coefficient data by the measurement system shown in FIG. That is, FIG. 2 shows a state where the switches 7L and 7R are connected to the terminals ga and ha, respectively, and the convolvers 2L and 2R are not provided in FIG. With the microphone 8 provided, the speakers 4L, 4R at this position are provided.
Of the convolvers 2L and 2R provided in the transmission path of the audio signal to cancel the response characteristics of the speakers 4L and 4R and correct the flatness of the response characteristics of the speakers 4L and 4R. 4L, 4R
FIG. 2 is a system configuration diagram for realizing an ideal impulse response in which the amplitude and phase characteristics are corrected by correcting the response characteristics of FIG.

In FIG. 2, reference numeral 11 denotes a digital I / O board for transmitting an ideal impulse as digital data;
Reference numeral 6 denotes a D / A converter for D / A converting the output of the digital I / O board 11, reference numeral 7 denotes an amplifier that amplifies the converted signal and inputs the amplified signal to a speaker 4L (or 4R), and reference numeral 8 denotes a speaker 4L (or 4R). ), An amplifier 9 for amplifying the signal captured by the microphone 8, and an A / D converter 10 for A / D converting the amplified output.
The output from the A / D converter 10 is a digital I / O board 11 and a computer 12
Through the speaker 4L (or 4L) before being corrected.
R) is measured, and a filter coefficient is calculated and output as coefficient data based on the measured impulse response waveform. Note that the characteristics of the microphone 8 are corrected in the course of calculation as necessary.

That is, the I / O board 11 constitutes a measuring signal generating means for generating a measuring signal, and includes a D / A converter 6, an amplifier 7, a speaker 4L, a microphone 8 to
The configuration of the path of the workstation 13 constitutes a response characteristic measuring means for obtaining an amplitude characteristic and a phase characteristic which are responses of the audio signal transmission system based on the measurement signal. Computing means for determining a target characteristic replaced so as to average the amplitude of the band and for obtaining a filter coefficient of a convolver provided in the audio signal transmission system so that a response of the audio signal transmission system converges on the target characteristic. This realizes a system that cancels the characteristics of the speaker and corrects the characteristics to be substantially flat to improve the sound quality.

Speakers 4L and 4R having the structure shown in FIG.
Is measured using the microphone 8 in an anechoic room, for example, using 4096 samples,
The measurement is performed by suppressing the error by performing the synchronous addition 1000 times.
FIG. 3 shows an impulse response waveform h (t) obtained by the measurement system, and FIG. 4A is a waveform showing an amplitude characteristic obtained by Fourier-transforming the impulse response waveform h (t). ) Shows the waveform of the target impulse response f0 (t) having the specific characteristics.

Here, the workstation 1 shown in FIG.
When calculating the filter coefficient, first, a middle band from 150 Hz to about 8000 is used as a correction band to be substantially averaged with respect to the amplitude characteristics before correction shown in FIG. And the high range is an uncorrectable band.

[0017] Of the correction band, 150H
A frequency band of about 1600 Hz from z is referred to as a piston band caused by 1) and 2) pointed out in the above-mentioned problem section (so that the speaker cone vibrates as a whole according to sound pressure without undulation. The band exceeding this level is recognized as a band in which undulation occurs in the speaker cone pointed out as the problem 3), and a level higher than the average level in this band is suppressed to the average level. Set as target characteristics and calculate filter coefficients.

That is, the impulse response f0 (t) of the specific characteristic shown in FIG. 4B, which is the amplitude characteristic corrected based on the measurement characteristic (the amplitude characteristic of the impulse response waveform h (t)), is obtained. Impulse response f0 of specific characteristics
(T) and the expanded matrix H obtained from the impulse response waveform h (t), and its transposed matrices HT and f0 (t) are 1
Each element of the determinant G consisting of one column that satisfies HTHG = HTF0 is defined as the filter coefficients g (n) of the convolvers 2L and 2R shown in FIG.

The solution of the above determinant will be described below. In this embodiment, a response waveform that satisfies the determinant is obtained by the above configuration, so that a response waveform can be uniquely obtained on the time axis. Specifically, according to the least-squares method using the Levinson algorithm (Reference: "Introduction to the Application of Digital Filters", Journal of the Acoustical Society of Japan, Vol. 43, No. 4 (1987), Haruo Hamada) It is assumed that a filter coefficient that minimizes the square of the difference between the impulse responses to be obtained is obtained.

Now, assuming that the discrete coefficients of the impulse response of the convolver are g1, g2,..., Gm-1, the discrete responses f0, f1,. Can be expressed.

[0021]

(Equation 1)

Where hi is the transfer characteristic, p is p = 0,
1, ..., n + m-2. Expressing equation (1) as a matrix,

[0023]

(Equation 2)

Equation (2) can be further expressed as F = HG. Here, the input impulse F
Taking the square of the difference between 0 and the impulse response F at the microphone position and defining the evaluation function P, P = (F-F0) T (F-F0) = (HG-F0) T (HG-F0) = ( GTHT-F0T) (HG-F0) = GTHTHG-F0THG-GTHTF0 + F0TF0, and in order to obtain the impulse response G of the convolver for minimizing the evaluation function P,

[0025]

(Equation 3)

Is calculated. Here, T represents a transposed matrix. Then, a solution G that satisfies HT HG = HT F0 (5) from equation (4) = 0 may be determined. That is, by setting the filter coefficient as in the above equation (5), the transmission characteristics are corrected, the amplitude / phase characteristics at the microphone position are averaged in the piston band, and a level higher than the average level is obtained in a higher band. Correction is performed so as to be suppressed to the average level, and other low and high frequencies are not corrected as uncorrectable bands, and have the same characteristics as the actual impulse response waveform. In this way, faithful original sound quality can be reproduced, and distortion of the sound image unavoidable with speakers or headphones can be removed. In addition, in the low and high frequency bands other than the middle band, actual speakers and the like can be used. The same characteristics as those of the measurement system can be obtained, and a natural audio signal adapted in consideration of the characteristics of the actual speaker can be enjoyed at the same time.

FIG. 5 is a flowchart showing a control operation for setting a filter coefficient by the workstation 13 using the measurement system shown in FIG. First, the response of the speaker is measured by the measurement system shown in FIG. 2 (step S1), and the impulse response waveform h (t) is subjected to Fourier transform (FFT: frequency-time axis transform) to obtain an amplitude characteristic (step S2). ). Next, in step S3, the low band and the high band not to be corrected are determined, and the level of each point in the band is determined.

On the other hand, in step S4, the average level of the piston zone is determined. Here, averaging of all points may be performed, or averaging at arbitrary intervals may be performed. Next, this averaging level, for example, 88 dB is set as the piston band level (step S5).

Then, ka which is presumed to cause swelling
A band where> 1 is set. That is, a frequency is set from a frequency sufficiently higher than the low-frequency resonance (a standard is an inertial control region) to a frequency at which ka = 1.

Next, in step S7, step S
The conversion result after measurement obtained in step 2 is compared with the average level. If the conversion result level is lower, the process directly proceeds to step S9. If the conversion result level is higher, the process proceeds to step S8. The level of each point, that is, the aforementioned average level is set.

In step S9, the low-frequency and high-frequency point data outside the correction band are combined with the correction data performed in step S8, and the combined data is subjected to inverse frequency transform (IFFT) (step S9). S9, 10). Finally, in step S10, the target frequency / measured frequency is calculated by the least squares method to obtain a convolver coefficient.

As described above, in the present embodiment, the sound processing is performed based on the coefficient that can obtain the above-described characteristics, so that the sound source particularly in the high-frequency range can be faithfully reproduced. In the above-described embodiment, the sound pressure above the average level is corrected so as to be suppressed to the average level in the band exceeding the piston band, but the sound pressure is not limited to the average level but is corrected so as to follow a preset envelope characteristic. May be. Further, in the above embodiment, the measurement system is provided in the audio signal transmission circuit system to determine the coefficient.However, the measurement system is separately and independently prepared, and the coefficient is obtained in advance by the method described above. The coefficients may be stored in the memory 16.

[0033]

As described above, according to the audio signal transmission circuit of the present invention, in particular, the predetermined level of the sound pressure in a high band which causes undulation in the speaker cone is suppressed to the predetermined level. The sound source in the high range can be faithfully reproduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of an audio signal transmission circuit as a first embodiment of the present invention.

FIG. 2 is a configuration diagram showing a system for measuring a filter coefficient of a convolver according to the present invention.

FIG. 3 is an explanatory diagram showing an impulse response waveform before correction.

FIG. 4 is a characteristic diagram showing an amplitude characteristic and a target characteristic obtained by Fourier-transforming the impulse response waveform of FIG. 3;

FIG. 5 is a flowchart for calculating a filter coefficient by a workstation 13 using the measurement system shown in FIG. 2;

[Explanation of symbols]

2L, 2R Convolver 4L, 4R Speaker 7, 9 Amplifier 7L, 7R Switch 8 Microphone 10 A / D converter 11 Digital I / O board (measurement signal generating means) 12 Computer 13 Workstation (arithmetic means) 15 Control unit (memory) 16 constitutes coefficient supply means) 16 memory

Claims (1)

[Claims] 1. A convolver for convolutionally processing a signal from a sound source in accordance with a set coefficient and outputting the convolver, and a predetermined measurement system including a transducer, and applying an audio signal from a predetermined sound source to the transducer. A predetermined value of the impulse response f0 (t) having a specific characteristic of the impulse response waveform h (t) at the measurement position based on the impulse response waveform h (t) measured in advance at the measurement position near the transducer. In the frequency band, the band lower than the predetermined band becomes flat, and only the level higher than the predetermined level in the higher band is suppressed to the predetermined level, and the calculated inverse filter coefficient is supplied to the convolver. Means for transmitting an audio signal.