WO2015085924A1

WO2015085924A1 - Automatic equalization method for loudspeaker

Info

Publication number: WO2015085924A1
Application number: PCT/CN2014/093456
Authority: WO
Inventors: 叶超
Original assignee: 苏州上声电子有限公司
Priority date: 2013-12-11
Filing date: 2014-12-10
Publication date: 2015-06-18
Also published as: CN103634726B; CN103634726A

Abstract

Provided is an automatic equalization method for a loudspeaker, which improves the sound replay performance of a loudspeaker system in a full frequency band. The method comprises: measuring impulse responses at one or more location points in a room by means of a loudspeaker, obtaining a frequency response at each location point and a low-frequency lower limit of the loudspeaker system, utilizing a self-adaptive optimization algorithm to obtain an equalization filter, and compensating the loudspeaker system. With regard to a low-frequency signal lower than a lower limit frequency of the loudspeaker system, the missing fundamental principle based on psychoacoustics is utilized, and a high order harmonic component of a fundamental frequency signal is generated and is superposed with a delayed original audio signal after gain control, and thus the sound replay capability of the loudspeaker system in the full frequency band is improved.

Description

Automatic speaker equalization method

Technical field

The present invention relates to the field of audio signal processing technologies, and in particular, to a speaker automatic equalization system, which aims to correct the frequency characteristics of a listening position in a room, improve the sound reproduction capability of the speaker system, and improve the sound quality. More specifically, this equalization method includes a virtual bass enhancement technique that generates harmonic components of low frequency components by nonlinearity to improve the sensing ability of low frequency components.

Background technique

The ideal sound reproduction system should have a relatively flat frequency response in the whole frequency band, but due to the limitation of the manufacturing process in the production process, the speaker system cannot have an ideal frequency response, and there is a certain distortion. On the other hand, due to the influence of the room modality and the interaction between the speaker system and the room, true sound reproduction cannot be achieved completely at the listening position. Therefore, the sound field correction technique is required to equalize the speaker system such that the frequency response at one or more position points approaches an ideal straight curve to ensure true playback of the original signal.

At present, the existing equalization techniques have a graphic equalizer and a parametric equalizer, which are mainly a set of cascaded peak or slope filters, and the center frequency of each filter corresponds to an octave or a 1/3 octave. The frequency band is controlled by adjusting the gain of each filter to achieve correction of the entire frequency band. This method is relatively straightforward, simple to implement, and easy to operate, but it needs to be familiar with the sound characteristics of each frequency band, so that it can be debugged more accurately, and the cascade of each filter is superimposed, which may cause the amplitude of certain frequencies to be unacceptable. Control the situation. In a more practical situation, the frequency response of the speaker system at one or more points is first measured by a microphone, and then the equalizer design is performed according to the measured curve. The equalizer is in the form of FIR (Finite Impulse Response) or IIR ( An infinite impulse response filter that filters the input signal such that an approximately flat frequency response is obtained at each location. However, since the frequency resolution of the low frequency band is low, in order to improve the low frequency resolution, it is necessary to increase the order of the filter, which increases the computational complexity. In addition, for a small-caliber speaker unit, if the direct equalization method is used to increase the energy of the low-frequency signal, the playback signal may be distorted and the speaker system may be damaged. The virtual bass enhancement technology based on the psychoacoustic fundamental frequency missing principle can solve this problem well. By using the non-linear effect of the human ear to obtain sound, the perception of low frequency sound can be improved subjectively, and the low frequency playback capability of the small aperture speaker can be improved. .

Summary of the invention

SUMMARY OF THE INVENTION It is an object of the present invention to provide an automatic speaker equalization method for compensating for the defects of the speaker itself and the influence of the room, and controlling the loudness characteristics of the audio signal to improve the sound reproduction performance at each position.

In order to achieve the above object, the present invention provides a speaker automatic equalization method, which in turn includes the following steps:

1) using a microphone to measure the transfer function of the speaker system electrical input signal to a plurality of locations in the room;

2) determining a prototype function of the plurality of transfer functions according to a weighting relationship of the plurality of position points;

3) design an equalization filter of the prototype function;

4) determining the low frequency lower limit frequency of the speaker system by the prototype function, and performing low pass filtering on the original input signal to obtain a low frequency fundamental frequency signal lower than the lower limit frequency;

5) generating a higher harmonic signal of the low frequency fundamental frequency signal by using a nonlinear algorithm;

6) After the high-order harmonic signal is controlled by the dynamic range, it is superimposed with the delayed original input signal and fed to the equalization filter;

7) The equalization filter output signal drives the speaker unit through the power amplifier.

Further, the method of measuring the speaker system transfer function mentioned in the above described scheme may employ a frequency sweep signal or a maximum length sequence (MLS), or other impulse response measurement method. The selected measurement location point should be the preferred listening location in the room, or cover a preferred listening area, such as a home theater or various seat locations within the car.

Further, in a specific example, the calculation process of the prototype function of determining a plurality of transfer functions is as follows:

Suppose that M transfer functions H _i (e ^jω ), i=1, 2, Λ, M are measured at M position points, and the prototype function is a characteristic function that characterizes the common trend of M transfer functions, which can be calculated in the following two ways. :

Using the weighted rms value of M transfer functions as a prototype function

Where a _i is a weighting coefficient, and different location points can be weighted according to actual conditions. For example, in a home theater, the sound quality of the position of the screen is emphasized, and other positions are considered. Another example is that in the interior of the car, the front or rear seats can be weighted differently according to actual needs. When a _i =1, i=1, 2, Λ, M, the prototype function is the root mean square value of the M transfer functions; or

Using the weighted arithmetic mean of M transfer functions as a prototype function

Where b _i is a weighting coefficient, and different location points can be weighted according to actual conditions. When b _i =1, i=1, 2, Λ, M, the prototype function is the arithmetic mean of the M transfer functions.

The prototype function describes the common features of multiple position point transfer functions. In the room sound field, the prototype function extracts the common characteristics of each position point from the aspects of direct sound, early reflection sound and reverberation sound, which can be achieved by equalizing the prototype function. Correction of the sound field for multiple locations.

Further, the design method of the prototype function equalization filter in step 3) may adopt a time domain adaptive optimization algorithm, including a minimum mean square error method, a recursive least square method, and the like. The adaptive algorithm adjusts its filter parameters by automatic iteration to meet the minimum criteria, thus achieving optimal filter coefficients.

Further, the low frequency lower limit frequency of the speaker in step 4) is determined by its physical characteristics; the low pass filter may be in the form of FIR (finite impulse response) or IIR (infinite impulse response) filter. The amplitude-frequency characteristic of the IIR filter is highly accurate. The system function can be written in the form of a closed function. It is implemented in a recursive structure with low computational complexity, but the phase characteristics are not linear, and the stability of the system needs to be considered. The amplitude and frequency characteristics of the FIR filter are lower than that of the IIR. Generally, there is no analytical expression and the computational complexity is high. The significant advantage is that the system is stable and has a linear phase.

Further, the nonlinear algorithm for generating the higher harmonic signal in step 5) may be a polynomial function, an exponential function or a power function and other nonlinear functions to generate a higher harmonic component of the input low frequency signal.

Further, the dynamic range control in the step 6) refers to dynamic control of the higher harmonic signal, and the control of the perceptible low frequency signal is realized by peak detection and gain control of the higher harmonic signal.

Further, the power amplifier of step 7) can have both analog and digital implementations. If the analog implementation is adopted, the digital signal output by the equalization filter is digital-to-analog converted into an analog signal, and then the power amplifier performs signal power amplification; if digital implementation is used, the digital signal output from the equalization filter is directly fed to the digital power amplifier. Signal power is amplified.

Further, the speaker unit described in step 7) may be a dynamic speaker of various sizes and specifications.

The advantages of the present invention over the prior art are:

A. The present invention can combine the speaker system with the acoustic characteristics of the room by on-line measurement and real-time equalization of the speaker system in the use environment, thereby improving the performance of the speaker system in a specific use environment.

B. The invention processes the low frequency and the high frequency of the frequency response of the speaker system separately, performs virtual bass enhancement for the low frequency signal below the lower limit frequency, adaptively equalizes the signal above the lower limit frequency, and improves the speaker system in the full frequency band. Sound playback capability.

C. The invention can equalize multiple points in the room, realize multi-point balance by extracting prototype functions of multiple position point transfer functions, and avoid the sound quality damage that may be caused to other points after equalization of one point in the room.

D. The present invention adopts a time domain adaptive algorithm to calculate the equalization filter, which can effectively improve the accuracy of the equalization, and adopts a time domain equalization algorithm, which can avoid the frequency domain algorithm to simultaneously consider the amplitude and phase balance, and reduce the computational complexity. .

DRAWINGS

1 is a flow chart of signal processing of the speaker automatic equalization system of the present invention;

2 is a flow chart of the sound field equalization process in FIG. 1;

3 is a schematic diagram of calculating an equalization filter using an adaptive algorithm in FIG. 2;

4 is a flow chart of signal processing for realizing virtual bass boosting in FIG. 1;

Figure 5 is a flow chart of signal processing for implementing dynamic range control in Figure 1;

6 is a time-domain graph of an equalization filter according to an embodiment of the present invention;

7A is a time-domain impulse response graph of a speaker system according to an embodiment of the present invention;

7B is a time-domain impulse response graph of a speaker system after equalization according to an embodiment of the present invention;

8A is a graph showing a frequency response curve of a speaker system according to an embodiment of the present invention;

8B is a graph showing a frequency response of a speaker system after equalization according to an embodiment of the present invention;

detailed description

The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

The invention firstly measures the impulse response of the speaker system at a plurality of position points in the room through a microphone, determines a prototype function from the plurality of impulse responses, calculates an equalization filter of the prototype function by using an adaptive optimization algorithm, and determines the speaker system according to the prototype function. The low frequency lower limit frequency is used to perform virtual bass boost on the input signal below this frequency to achieve automatic equalization of the speaker system in the full frequency band.

The speaker automatic equalization system according to the present invention as shown in FIG. 1 is mainly composed of a sound source 101, a virtual bass enhancement module 102, a dynamic range control 103, a delay unit 104, an equalization filter 105, and a power amplifier 106. It is composed of a speaker unit 107 and the like. The sound source 101 is connected to the input end of the virtual bass enhancement 102 for enhancing the bass below the lower limit frequency of the speaker system; the output of the virtual bass enhancement module 102 is connected to the input of the dynamic range control 103, The signal of the virtual bass processing is dynamically controlled to remove noise; the output of the dynamic range control 103 is added to the output of the delay unit 104, and then connected to the input end of the equalization filter 105, and the input signal is equalized and then sent to the power. The amplifier 106 amplifies the equalized signal and drives the speaker unit 107 to sound.

The calculation process of the equalization filter 105 in FIG. 1 is shown in FIG. 2. The specific implementation step is to first measure the impulse response of a plurality of position points in the room by using a microphone, and obtain the prototype function by using the above formula (1) or (2). 202, since the prototype function 202 is generally a non-minimum phase system, it can be divided into a minimum phase system 203 and an all-pass system 204, respectively obtaining amplitude information 205 and phase information 206; and then frequency transforming it using equation (3) 207. Transform the prototype function from a linear frequency to a bending frequency to improve low frequency resolution.

In the formula (3), z ^-1 is a frequency domain delay unit, and λ is a bending factor, and the value ranges from -1<λ<1. In the bending frequency domain, the equalization filter 209 is obtained using the adaptive optimization algorithm 208.

The schematic diagram of the adaptive optimization algorithm is shown in Figure 3, where x(t) is the input signal, H(z) is the filter, P(z) is the system function, and d(t) and y(t) are the expected The signal and output signals, e(t) = d(t) - y(t), are error signals. The error signal e(t) is minimized by updating the coefficients of the filter H(z) by an optimization algorithm. Preferably, the calculation process of the adaptive optimization algorithm is specifically described by taking the minimum mean square error method as an example. Suppose the input signal vector X(n)=[x(n), x(n-1), Λ, x(n-N+1)] ^T , filter coefficient H(n)=[b ₀ (n), b ₁ (n), Λ, b _N-1 (n)] ^T , N is the filter length. Estimating the gradient vector using the instantaneous value of the square of the error of a single sample, ie

The filter coefficient is calculated as

Where μ is the step factor. The larger the μ value, the faster the algorithm converges, but the larger the steady-state error; the smaller the μ value, the slower the convergence of the algorithm, but the smaller the steady-state error.

The virtual bass boost module is shown in Figure 4. The input signal 401 first passes through the low pass filter 402, generates the higher harmonics 403 by nonlinearity, and then performs the band pass filtering 404 and the original input signal through the delay unit 405. The number 401 is superimposed to obtain an output signal 406 that is enhanced by the virtual bass. The manner in which the harmonics 403 are nonlinearly generated may be a polynomial function, an exponential function or a power function, and other nonlinear functions. Preferably, the manner in which the harmonics are generated nonlinearly may be in the form of a polynomial of equation (6).

f(x)=a ₀ +a ₁ x+a ₂ x ² +Λ+a _N x ^N (6)

Where a ₀ , Λ, a _N are constant coefficients.

The dynamic range control processing flow is shown in Figure 5. The virtual bass enhanced output signal 406 is used as the dynamic range controlled input signal 501, multiplied by the peak detection 502 and the gain control 503 and the original input signal passed through the delay unit 504 to achieve dynamic range of the original input signal 501. control.

The invention will now be described in detail in conjunction with the drawings and an embodiment.

In the present embodiment, the speaker unit is 3.5 inches in size, and the impulse response and frequency response of the speaker system are first measured using a microphone as shown in Figs. 7A and 8A, respectively. The adaptive optimization method is used to obtain the equalization filter, which adopts the 300-order FIR form, and the time domain waveform is shown in Fig. 6. The pulse response and frequency response after equalization of the speaker system are shown in Figures 7B and 8B, respectively. It can be seen from the figure that after the equalization process, the impulse response of the speaker system is sharper, the frequency response is flatter, and the frequency characteristics are better. And use the virtual bass enhancement algorithm for bass compensation. Through the actual audition, the performance of the low frequency band is obviously enhanced, the music of the middle and high frequency is brighter and the sound is more natural.

Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention and not limiting. While the invention has been described in detail herein with reference to the embodiments of the embodiments of the invention Within the scope of the claims.

Claims

An automatic speaker equalization method includes the following steps in sequence:

1) using a microphone to measure the transfer function of the speaker system electrical input signal to a plurality of locations in the room;

2) determining a prototype function of the plurality of transfer functions according to a weighting relationship of the plurality of position points;

3) design an equalization filter of the prototype function;

4) determining the low frequency lower limit frequency of the speaker system by the prototype function, and performing low pass filtering on the original electrical input signal to obtain a low frequency fundamental frequency signal lower than the lower limit frequency;

5) generating a higher harmonic signal of the low frequency fundamental frequency signal by using a nonlinear algorithm;

6) After the high-order harmonic signal is controlled by the dynamic range, it is superimposed with the delayed original electrical input signal and fed to the equalization filter;

7) The equalization filter outputs a digital signal through the power amplifier to drive the speaker unit.
The automatic speaker equalization method according to claim 1, wherein the measurement position selected in step 1) is a listening position selected in the room, or a listening area covering the selected listening position.
The automatic speaker equalization method according to claim 1, wherein the prototype function in step 2) describes a common feature of a transfer function of one or more position points, and in the room sound field, the prototype function is from direct sound and early reflection. Sound and reverberation sounds extract the common characteristics of each position point. It is assumed that M transfer functions H i (e jω ), i=1, 2, Λ, M are measured at M position points, and the calculation method of the prototype function includes ,

Using the weighted rms value of M transfer functions as a prototype function

Where a i is a weighting factor, i=1, 2, Λ, M; or

Using the weighted arithmetic mean of M transfer functions as a prototype function

Where b i is a weighting coefficient, i=1, 2, Λ, M.
The automatic speaker equalization method according to claim 1, wherein the design method of the equalization filter in step 3) adopts an adaptive optimization method, including a minimum mean square error method and a recursive least squares method.
A speaker automatic equalization method according to claim 1, wherein the speaker in step 4) is low The lower frequency limit is determined by its physical characteristics.
The automatic speaker equalization method according to claim 1, wherein the step 4) employs a low pass filter in the form of a finite impulse response or an infinite impulse response filter.
The automatic speaker equalization method according to claim 1, wherein the nonlinear algorithm for generating the higher harmonic signal in the step 5) comprises a polynomial function, an exponential function or a power function, and other nonlinear functions to generate a low frequency base. The higher harmonic component of the frequency signal.
The automatic speaker equalization method according to claim 1, wherein the dynamic range control in step 6) is to dynamically control the higher harmonic signal, and the peak detection and gain control of the higher harmonic signal are performed. Achieve control of perceptible low frequency signals.
The automatic speaker equalization method according to claim 1, wherein in step 7), the power amplifier is an analog implementation, and the digital signal output by the equalization filter is digital-to-analog converted into an analog signal, and then the signal power is performed by the power amplifier. The output is amplified and used to drive the speaker unit.
The automatic speaker equalization method according to claim 1, wherein the power amplifier in step 7) is a digital implementation, and the digital signal output by the equalization filter is directly fed to the digital power amplifier for signal power amplification and output. Used to drive the speaker unit.
The automatic speaker equalization method according to claim 1, wherein the speaker unit in step 7) comprises a plurality of dynamic speakers of different sizes and specifications.