All-pass filter for loudspeaker group delay compensation and design method thereof

Abstract

The present application relates to all-pass filter for loudspeaker group delay compensation and its design method. A design method of all-pass filter for loudspeaker group delay compensation can include: determining a target compensation curve based on a group delay frequency response curve of a loudspeaker to be compensated and a target group delay frequency response curve; determining the number and order of sub-filters for implementing the target compensation curve; dividing the target compensation curve into corresponding frequency bands according to the number and order of sub-filters, and then determining the parameters of the sub-filters for compensating the corresponding frequency bands; calculating the group delay frequency response curve of the all-pass filter based on the parameters of each sub-filter, and determining the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve; and optimizing the number and / or parameters of the sub-filters to reduce the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and more particularly to a design method for an all-pass filter for speaker group delay compensation, an all-pass filter designed according to the method, and an audio signal processing system including the all-pass filter. Background Technology

[0002] Currently, portable mobile devices such as mobile phones and tablets primarily use minimum phase equalizers to compensate for amplitude frequency response curves to optimize sound quality. Minimum phase equalizers adjust speaker amplitude response with low computational cost and latency, but inevitably, they introduce phase distortion along with the amplitude frequency response. This causes a difference in phase between the harmonics and the fundamental frequency, resulting in the loss of some musical details and affecting fidelity. On the other hand, some low-end speakers naturally possess significant nonlinearity in their phase frequency response, which is difficult to remedy using a minimum phase equalizer.

[0003] The group delay frequency response curve is also a way to observe the phase characteristics of a loudspeaker. It is defined as the negative value of the derivative of the phase frequency curve with respect to frequency, and its description is essentially equivalent to that of the phase frequency curve. Therefore, compensation or equalization for non-ideal phase response can also be described by compensation or equalization of the group delay frequency response characteristics. Generally speaking, to achieve high-fidelity playback of audio signals, the phase frequency response of the loudspeaker is required to be linear, which means that the group delay curve is a constant at all frequencies. In practical calculation and analysis, using the group delay curve as the compensation object is more intuitive and easier to understand in terms of its physical meaning. Therefore, many studies also use the processing of the group delay frequency response curve to achieve compensation of the system's phase frequency characteristics.

[0004] When the amplitude frequency response of a loudspeaker is already determined or does not wish to be significantly altered, all-pass filters are primarily used to compensate for and equalize the speaker's group delay response. However, designing an all-pass filter for compensating for speaker group delay with low computational complexity, low latency, and high precision, while meeting consumers' demands for high-fidelity sound quality from loudspeakers, remains one of the pressing issues that the industry needs to address. Summary of the Invention

[0005] When the amplitude frequency response of a loudspeaker is already determined or does not wish to be significantly altered, an all-pass filter can be used to compensate for and equalize the loudspeaker's group delay response. This invention provides a design method for an all-pass filter, which can be used for high-precision, low-computational-load, and low-latency real-time compensation of the loudspeaker's group delay frequency response. This invention also provides an audio signal processing system including such an all-pass filter.

[0006] An exemplary embodiment of the present invention provides a design method for an all-pass filter for loudspeaker group delay compensation, comprising: determining a target compensation curve based on the group delay frequency response curve of the loudspeaker to be compensated and a target group delay frequency response curve; determining the number of sub-filters and the order of each sub-filter for the all-pass filter used to achieve the target compensation curve; dividing the target compensation curve into corresponding frequency bands according to the number and order of the sub-filters, and then determining the parameters of the sub-filters used to compensate the corresponding frequency bands; calculating the group delay frequency response curve of the all-pass filter based on the parameters of each sub-filter, and determining the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve; and optimizing the number and / or parameters of the sub-filters to reduce the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve.

[0007] In some exemplary embodiments, the method further includes combining two or more low-order sub-filters into a high-order sub-filter.

[0008] In some exemplary embodiments, the group delay frequency response curve of the loudspeaker to be compensated is determined by measuring or calculating the group delay frequency response curve of the loudspeaker after processing with a minimum phase equalizer.

[0009] In some exemplary embodiments, determining the target compensation curve includes: determining the target compensation curve based on the difference between the target group delay frequency response curve and the group delay frequency response curve of the loudspeaker to be compensated, wherein the target compensation curve has a group delay value greater than or equal to zero at each frequency point.

[0010] In some exemplary embodiments, the target compensation curve is the sum of the difference between the target group delay frequency response curve and the group delay frequency response curve of the loudspeaker to be compensated, plus a constant.

[0011] In some exemplary embodiments, determining the number of sub-filters of the all-pass filter used to achieve the target compensation curve and the order of each sub-filter includes: calculating the integral of the target compensation curve with respect to frequency to obtain the phase change; and determining the number of sub-filters of the all-pass filter and the order of each sub-filter such that the sum of the phase changes of each sub-filter on the positive half-axis of frequency is equal to or close to the phase change of the target compensation curve.

[0012] In some exemplary embodiments, when the target compensation curve is divided into corresponding frequency bands according to the number and order of the sub-filters, the integral of the target compensation curve over the frequency band to be compensated by the sub-filters is substantially equal to the phase change of the corresponding sub-filters.

[0013] In some exemplary embodiments, determining the parameters of the sub-filter used to compensate for the corresponding frequency band includes: calculating the complex coefficients of the transfer function of the corresponding sub-filter based on the start frequency, end frequency and sampling frequency of the frequency band; and multiplying the transfer function of the sub-filter with complex coefficients by a transfer function with conjugate complex coefficients to obtain the real coefficients of the transfer function of twice the order.

[0014] In some exemplary embodiments, calculating the group delay frequency response curve of the all-pass filter based on the parameters of each sub-filter includes: cascading each sub-filter in order corresponding to each frequency band of the target compensation curve to form the all-pass filter; and using the parameters of each sub-filter to calculate the group delay frequency response curve of the formed all-pass filter.

[0015] In some exemplary embodiments, optimizing the number and / or parameters of sub-filters includes adjusting the number of sub-filters and / or the parameters of at least one sub-filter until the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve falls within a predetermined range.

[0016] Another exemplary embodiment of the present invention provides an all-pass filter for delay compensation of a loudspeaker group, wherein the all-pass filter is designed according to the method described above.

[0017] Another exemplary embodiment of the present invention provides an audio signal processing system, including: a minimum phase equalizer for adjusting the gain of each frequency point of an audio signal; and an all-pass filter for compensating for the group delay of each frequency point of the audio signal, wherein the all-pass filter is designed according to the method described above.

[0018] The embodiments of the present invention can achieve the following beneficial technical effects:

[0019] 1. This invention starts from the physical characteristics of the all-pass filter and calculates the number of sub-filters and initial parameter values ​​of the all-pass filter based on the area covered by the group delay error frequency response curve. This avoids the stability problems caused by using various global search methods to obtain the optimal number of compensation filters and greatly reduces the amount of computation.

[0020] 2. After obtaining the initial values ​​of the compensation filter parameters, the accuracy of the compensation filter can be further optimized by combining the optimization algorithm. A trade-off can also be made between performance and computational load, such as compensation accuracy, number of filters and maximum delay.

[0021] 3. The embodiments of the present invention are not limited to compensating the group delay of each frequency point of the loudspeaker system into a horizontal straight line to achieve an ideal linear phase effect and obtain a more realistic listening experience. Instead, they can also be compensated into a target group delay curve of any shape to change the harmonic delay characteristics of the loudspeaker, thereby producing a variety of unique tonal effects.

[0022] The above and other features and advantages of the present invention will become apparent from the following description of specific embodiments taken in conjunction with the accompanying drawings. Attached Figure Description

[0023] Figure 1 This is a flowchart of an all-pass filter design method according to an exemplary embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram illustrating examples of the time delay frequency response curves of the loudspeaker group and the target group.

[0025] Figure 3A and Figure 3B This is a schematic diagram of the target compensation curve determined by some exemplary embodiments of the present invention.

[0026] Figure 4A and Figure 4B This is a schematic diagram illustrating how the target compensation curve is divided into multiple frequency bands in some exemplary embodiments of the present invention.

[0027] Figure 5 This is a schematic diagram of an audio signal processing system according to an exemplary embodiment of the present invention.

[0028] Figure 6 This is a schematic diagram of the loudspeaker group delay frequency response curves before and after group delay compensation using an all-pass filter according to an exemplary embodiment of the present invention. Detailed Implementation

[0029] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Note that the drawings may not be drawn to scale. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments of this application, and this application is not limited to the exemplary embodiments described herein.

[0030] Figure 1 This is a flowchart of an all-pass filter design method 100 according to an exemplary embodiment of the present invention. (Refer to...) Figure 1 Method 100 may include step 110, which is to determine the target compensation curve of the all-pass filter based on the group delay frequency response curve of the loudspeaker to be compensated and the target group delay frequency response curve.

[0031] The group delay frequency response curve of a loudspeaker, also known as the group delay profile, can be obtained after equalizing the amplitude frequency response with a minimum phase equalizer. It can be calculated by testing the loudspeaker using a swept frequency signal or a maximum length sequence (MLS) signal, or it can be directly measured using audio analyzers from companies such as Audio Precision (also known as AP). Since various methods for determining the group delay frequency response curve of a loudspeaker are known, they will not be described in detail here. Figure 2 An example of the group delay frequency response curve for a loudspeaker is shown. Figure 2 As shown, the group delay frequency response curve GD1 has a maximum group delay value of about 11ms near 100Hz. In the range of 100Hz to about 400Hz, the group delay value decreases monotonically with increasing frequency. In the range of 400Hz to about 1000Hz, the group delay value gradually oscillates and decreases with increasing frequency, and above 1000Hz, the group delay value approaches 0.

[0032] Based on the group delay frequency response curve of the loudspeaker to be compensated, the target group delay frequency response curve after compensation can be determined. Generally speaking, in order to achieve high-fidelity playback, the phase frequency response of the loudspeaker and other acoustic reproduction system is required to be linear. Since the group delay curve is the negative derivative of the phase frequency curve with respect to frequency, the group delay curve is required to be a constant at all frequencies. Figure 2 An example of a target group delay frequency response curve GD2, determined based on the group delay frequency response curve GD1 to be compensated, is shown. The target group delay frequency response curve GD2 can be a horizontal straight line, representing a group delay value that is a constant greater than or equal to the maximum group delay value of the group delay frequency response curve GD1 to be compensated. Of course, the target group delay frequency response curve is not limited to... Figure 2 The example shown can be, for example, a curve. For instance, in some frequency bands, such as the low-frequency band, the group delay value of the target group delay frequency response curve GD2 can be greater than the maximum group delay value of the group delay frequency response curve GD1 to be compensated; while in other frequency bands, such as the mid-frequency band and / or the high-frequency band, the group delay value of the target group delay frequency response curve GD2 can be less than the maximum group delay value of the group delay frequency response curve GD1 to be compensated. The group delay value of the target group delay frequency response curve GD2 at various frequency points can be flexibly determined according to the desired playback effect. For example, to improve loudness, the group delay value of several specific frequency bands can be increased. Embodiments of the present invention are not limited to any shape of the target group delay frequency response curve GD2. It should also be understood that, for simplicity and ease of implementation, it is preferable that the group delay value of the target group delay frequency response curve GD2 at the same frequency point can be greater than or equal to the group delay value of the group delay frequency response curve GD1 to be compensated.

[0033] In step 110, a target compensation curve can be determined based on the loudspeaker's group delay frequency response curve GD1 to be compensated and the target group delay frequency response curve GD2. This target curve can represent the group delay compensation effect that the all-pass filter aims to achieve. In some embodiments, the target compensation curve can be determined based on the difference between the loudspeaker's target group delay frequency response curve GD2 and the group delay frequency response curve GD1 to be compensated. For example, Figure 3A An example of a target compensation curve GD_comp is shown, where the target compensation curve GD_comp can be simply determined as the difference between the target group delay frequency response curve GD2 and the group delay frequency response curve GD1 to be compensated, i.e., GD_comp = GD2 - GD1. In some embodiments, to ensure that the target compensation curve GD_comp has a non-negative (i.e., greater than or equal to zero) group delay value at each frequency point, the target compensation curve GD_comp can also be subtracted from its minimum value. In another embodiment, the target compensation curve GD_comp can also be added to the group delay value at each frequency point by the same non-negative constant value d0, i.e., GD_comp = GD2 - GD1 + d0, as shown. Figure 3B As shown, increasing the target group delay d0 can reduce the total compensation error and improve compensation accuracy, but it obviously brings greater system delay, resulting in poorer real-time performance. Therefore, a trade-off between compensation accuracy and real-time performance can be considered to select an appropriate constant value for d0.

[0034] As mentioned earlier, the target compensation curve GD_comp represents the group delay compensation effect that the all-pass filter aims to achieve. In the following step 120, the number of sub-filters included in the all-pass filter and the order of each sub-filter can be determined based on the target compensation curve GD_comp, i.e., the group delay compensation effect that the all-pass filter aims to achieve. The group delay curve is the negative value of the derivative of the phase-frequency curve with respect to frequency. Therefore, the integral of the target compensation curve GD_comp with respect to frequency (i.e., the frequency range between the minimum and maximum frequencies of the target compensation curve GD_comp) corresponds to the phase change of the all-pass filter, also known as the phase of the all-pass filter. In step 120, the integral of the target compensation curve GD_comp with respect to frequency, i.e., the area of ​​the region between the target compensation curve GD_comp and the frequency horizontal axis, can be calculated first to determine the phase change of the all-pass filter. The integral of the group delay curve of a predetermined order all-pass filter over the full frequency range (-π, π) is known; that is, the integral of the group delay curve of an M-order all-pass filter over the full frequency range is 2Mπ, where M is an integer greater than or equal to 1. However, in audio processing applications, generally only the characteristics of the all-pass filter in the positive half-axis frequency range (0, π) are considered. Therefore, the integral of the group delay curve of an M-order all-pass filter over the positive half-axis frequency range (0, π) is Mπ. In step 120, the number and order of the sub-filters included in the all-pass filter can be determined such that the sum of the integrals of the group delay curves of these sub-filters over the positive half-axis frequency range, which is also the sum of the phase changes of these sub-filters over the positive half-axis frequency range, is equal to or close to the integral of the target compensation curve GD_comp with respect to frequency, i.e., the phase change corresponding to the target compensation curve GD_comp. In some embodiments, the order of the sub-filters can be a power of 2, such as 1st, 2nd, 4th, 8th, etc., and so on.

[0035] For example, assuming the integral of the target compensation curve GD_comp with respect to frequency is 7π, then in step 120, the all-pass filter can be determined to include 7 first-order sub-filters, or 3 second-order sub-filters and 1 first-order sub-filter, or 1 fourth-order sub-filter, 1 second-order sub-filter, and 1 first-order sub-filter. When the integral of the target compensation curve GD_comp with respect to frequency is not an integer multiple of π, it can be rounded down to an integer multiple of π before determining the number and order of the sub-filters. In other embodiments, the integral of the target compensation curve GD_comp with respect to frequency can also be rounded up or rounded to an integer multiple of π before determining the number and order of the sub-filters. For example, when the integral of the target compensation curve GD_comp with respect to frequency is 7.8π, it can be rounded down to 7π, rounded up to 8π, or rounded to 8π before determining the corresponding number and order of the sub-filters. In other words, the sum of the phase changes of the sub-filters can be equal to or close to the integral of the target compensation curve GD_comp with respect to frequency. Here, "close" means that the difference between the two is less than π, that is, the integral of the group delay curve of the first-order sub-filter on the positive half-axis of frequency. From the perspective of saving computation, it is preferable to round down the integral of the target compensation curve GD_comp with respect to frequency, because the group delay compensation effect in the high-frequency band of audio is not as good as that in the low-frequency band. Therefore, it is generally preferred to compensate the mid-low frequency band, while the last frequency band (i.e., a small segment of the target compensation curve GD_comp on the high-frequency side corresponding to a fractional multiple of the integral of π) can be ignored and not compensated.

[0036] In some embodiments, when the integral of the target compensation curve GD_comp with respect to frequency is a fractional multiple of π, the group delay constant d0 added to the entire target compensation curve GD_comp can also be adjusted (see...). Figure 3B This is done so that the integral of the target compensation curve GD_comp with respect to frequency is an integer multiple of π. Then, as described above, the number and order of the sub-filters included in the all-pass filter are determined.

[0037] It is understandable that floating-point operations incur some precision loss. While fewer operations are performed during group delay compensation, the higher the order of the sub-filter, the greater the precision loss due to floating-point operations. Therefore, the highest preset sub-filter order can be determined by considering the floating-point precision of the audio processing system performing group delay compensation, and the order of the sub-filters determined in step 120 must not exceed this preset highest order. It is known that the precision of single-precision floating-point operations is lower than that of double-precision floating-point operations, and the precision of double-precision floating-point operations is lower than that of extended-precision floating-point operations. In some embodiments, if the audio processing system only supports single-precision floating-point operations, the order of the sub-filters determined in step 120 must not exceed order 2 or 4; if the audio processing system supports double-precision floating-point operations, the order of the sub-filters determined in step 120 may not exceed order 4 or 8; if the audio processing system supports extended-precision floating-point operations, the order of the sub-filters determined in step 120 may not exceed order 8 or 16. It should be understood that the highest order given here is merely an example, and the invention is not limited thereto. In step 120, the order of the determined sub-filters need not exceed the predetermined highest order. In some embodiments, each sub-filter may also be determined to be a first-order all-pass filter in step 120.

[0038] Continue to refer to Figure 1 In step 130, the target compensation curve GD_comp can be divided into multiple corresponding frequency bands according to the number and order of the sub-filters determined in step 120, and then the parameters of the sub-filters used to compensate the corresponding frequency bands can be determined. Figure 4A and Figure 4B Examples of dividing the target compensation curve GD_comp into multiple frequency bands are shown, where the phase change corresponding to each frequency band to be compensated, i.e., the integral of the target compensation curve GD_comp in that frequency band, can be substantially equal to the phase change of the corresponding sub-filter. In this application, "substantially equal to" can mean that the difference between two values ​​is less than or equal to 10% of one of the values, preferably less than or equal to 5% of one of the values. Figure 4AIn the example, the target compensation curve GD_comp is divided into 5 frequency bands. In the first 4 frequency bands starting from the low-frequency side, the integral of the target compensation curve GD_comp in each band is approximately equal to the phase change of the corresponding sub-filter, which in this example is the phase change π of the first-order sub-filter. However, the integral of the target compensation curve GD_comp in the 5th frequency band (the last band on the high-frequency side) is less than π, insufficient to correspond to a lowest-order (i.e., first-order) sub-filter, and the compensation effect on the high-frequency side is not significant. Therefore, this frequency band can be ignored without compensation, saving computational resources. Of course, if in step 120 it is determined that the target compensation curve GD_comp corresponds to 5 first-order sub-filters (e.g., the integral of the target compensation curve GD_comp on the frequency is rounded up or rounded to the nearest larger integer multiple of π), then... Figure 4A The fifth frequency band shown can correspond to the fifth first-order sub-filter. That is, the integral of the target compensation curve GD_comp at a certain frequency band can also be less than the phase change of the sub-filter used to compensate for that frequency band. Alternatively, in some embodiments, for example when Figure 4A When the integral value corresponding to the 5th frequency band of the target compensation curve GD_comp shown is relatively small, a sub-filter can also be allocated to the 4th and 5th frequency bands, that is to say... Figure 4A The fourth and fifth frequency bands shown can be merged into one frequency band and assigned a sub-filter. In this case, the integral of the target compensation curve GD_comp over the merged frequency band can also be greater than the phase change of the sub-filter used to compensate for the merged frequency band. It can be understood that, generally speaking, in step 130, the target compensation curve GD_comp can be divided into multiple frequency bands and the corresponding sub-filters can be determined according to the number and order of the sub-filters determined in step 120. However, when dividing the frequency bands, at least on the high-frequency side, a flexible approach can be adopted, and it is not limited to any specific precise correspondence.

[0039] exist Figure 4B In the embodiment, when the constant value d0 is adjusted so that the integral of the target compensation curve GD_comp at frequency is an integer multiple of π (6 times in this example), and six first-order sub-filters are determined in step 120, the target compensation curve GD_comp can be divided into six frequency bands in step 130, and the integral of the target compensation curve GD_comp at each frequency band is equal to the phase change of the corresponding sub-filter. Here, by comparison... Figure 4A and Figure 4B It is understandable that when the group delay value increases by d0, the integral of the target compensation curve GD_comp over frequency increases, thus increasing the number of determined sub-filters, or the order of the sub-filters can be increased, thereby achieving higher compensation accuracy.

[0040] It should also be understood that Figure 4A and Figure 4B The multiple frequency bands with integral values ​​of π shown are merely examples. When second-order, fourth-order, or mixed (i.e., including sub-filters of different orders) are determined in step 120, the target compensation curve GD_comp can be divided into frequency bands with corresponding integral values ​​of 2π, 4π, or different integral values. It can be understood that each determined sub-filter is used to compensate for the group delay value of each corresponding frequency band.

[0041] In step 130, after dividing the target compensation curve GD_comp into multiple frequency bands, the parameters of the sub-filters used to compensate the corresponding frequency bands can be further determined. Here, the all-pass filter used to realize the target compensation curve GD_comp is composed of multiple cascaded sub-filters, and each sub-filter is also an all-pass filter. The parameters of the corresponding sub-filters determined according to the frequency bands of the target compensation curve GD_comp may include the coefficients of the transfer function of the sub-filters. For example, the coefficients of the transfer function of a first-order sub-filter can be calculated according to the following formula.

[0042]

[0043] In Formula 1 above, f1 is the starting frequency of the corresponding frequency band, f2 is the ending frequency of the corresponding frequency band, f0 is the center frequency of the frequency band, and f... s The sampling frequency is denoted by B. B is a coefficient ranging from 0 to 1. A larger value results in a smoother overall group delay curve after compensation, but with larger compensation errors at local frequencies. A smaller value results in finer compensation at local frequencies and smaller errors, but with more spikes in the overall group delay curve after compensation. Generally, B can range from 0.3 to 0.9, preferably from 0.5 to 0.8. x(n) and y(n) are the input and output signals in the time domain, respectively, and satisfy y(n) = x(n) * h(n), where "*" indicates convolution, and h(n) is the time-domain impulse response. The time-domain input signal x(n), output signal y(n), and impulse response h(n) are transformed by z to obtain the complex domain X(z), Y(z), and H(z), respectively, where H(z) is the transfer function. For any discrete signal x(n), its z-transform is defined by the following formula 2, where z is a complex exponent and the complex exponent z = r·e jω r represents the amplitude of z; ω represents the phase of z, in radians.

[0044]

[0045] The coefficients b0, b1, and a1 of the transfer function H(z) of the first-order sub-filter can be determined based on formulas 1 and 2 above. However, these coefficients include complex coefficients b0 and a1. In practical applications, audio signals are real signals, therefore all parameters of the all-pass filter must be real parameters. In some embodiments, the transfer function H(z) can be multiplied by its conjugate complex coefficients, resulting in the transfer function H1(z), as shown in the formula below.

[0046]

[0047] Therefore, we can obtain the transfer function H2(z) of a second-order all-pass filter with all real coefficients that is basically the same as the group delay curve of a first-order all-pass filter. The real coefficients (b'0, b'1, b'2, a'1, a'2) are shown in Equation 4 below.

[0048]

[0049] After determining the parameters of each sub-filter, such as the transfer function coefficients, cascading these sub-filters yields an all-pass filter used to achieve the target compensation curve GD_comp. Next, in step 140, the group delay curve GD3 of the all-pass filter can be calculated, and its deviation D from the target compensation curve GD_comp can be determined. It should be understood that various criteria can be used to define the deviation D; for example, the deviation D can be the difference, variance, or root mean square error between the target compensation curve GD_comp and the group delay curve GD3 of the all-pass filter. For ease of description, it can be simply represented here as D = GD_comp - GD3.

[0050] The initially determined parameters of the all-pass filter may not accurately achieve the target compensation curve GD_comp, but rather there may be a significant deviation D between its group delay curve GD3 and the target compensation curve GD_comp. Therefore, in step 150, an optimization algorithm can be used to reduce the deviation D between the group delay frequency response curve GD3 of the all-pass filter and the target compensation curve GD_comp by optimizing the number and / or parameters of the sub-filters. For example, in some embodiments, the optimization algorithm can be used to continuously adjust the parameters of the all-pass filter until the deviation D between the group delay curve GD3 of the all-pass filter and the target compensation curve GD_comp falls within a predetermined sufficiently small range, that is, the group delay curve GD3 of the all-pass filter is sufficiently close to the target compensation curve GD_comp. Examples of optimization algorithms that can be used include, but are not limited to, genetic algorithms, neural network algorithms, etc. During optimization, the parameters of the sub-filters of the all-pass filter, such as the transfer function coefficients, can be adjusted, or the number of sub-filters can be adjusted, such as replacing a higher-order sub-filter with multiple lower-order sub-filters, or adding more sub-filters, etc. During optimization, it is not limited to optimizing the entire frequency band; it can also be optimized only for specific frequency bands or several frequency bands of interest.

[0051] The optimized sub-filters can be directly cascaded to form an all-pass filter for achieving the target compensation curve GD_comp. In some embodiments, two or more optimized low-order sub-filters can optionally be combined into a higher-order sub-filter, which can be achieved by multiplying their transfer functions to obtain a higher-order transfer function. Combining into higher-order sub-filters can save computation, but may result in some loss of precision when performing floating-point operations. Therefore, the order of the combined sub-filters is preferably no higher than a predetermined order, which can be determined based on the floating-point precision of the system performing the audio signal processing. For example, the higher the floating-point precision of the audio processing system, the higher the order of the sub-filters can be, specifically determined based on system capabilities and precision requirements. The combined sub-filters can then be cascaded to obtain the final all-pass filter.

[0052] The all-pass filter designed according to the above method 100 can be applied to audio processing systems. Figure 5 An example of such an audio processing system 200 is shown. (See reference...) Figure 5 The audio processing system 200 may include a minimum phase equalizer 210 and an all-pass filter 220. The minimum phase equalizer 210 can be used to adjust the gain of each frequency point of the audio signal, and the all-pass filter 220 can be used to compensate for the group delay of each frequency point of the audio signal. The all-pass filter 220 can be designed using the method 100 described above. Although Figure 5The diagram shows the full-pass filter 220 positioned downstream of the minimum phase equalizer 210, but in some embodiments, the full-pass filter 220 may also be positioned upstream of the minimum phase equalizer 210.

[0053] The audio processing system 200 may also include an automatic gain control (AGC) unit 230 and a run protection unit 240. The AGC unit 230 controls the gain of the audio signal to achieve, for example, an appropriate playback volume, while the run protection unit 240 ensures that the power of the audio signal does not exceed an upper limit threshold to avoid damage to the speaker 250 due to excessive power. Although not shown, the audio processing system 200 may also include other audio processing modules known in the field of audio processing, which will not be described in detail here. Finally, the audio processing system 200 provides the processed audio signal to the speaker 250 for playback. Although only one speaker is shown here, it should be understood that the speaker 250 may be a sound reproduction system including multiple speakers, such as a 2.1 or 5.1 system.

[0054] Figure 6 This is a schematic diagram showing the loudspeaker group delay frequency response curves before and after group delay compensation using an all-pass filter according to an exemplary embodiment of the present invention. (Refer to...) Figure 6 The group delay frequency response curve GD1 is the group delay curve before compensation, which has a much larger group delay at low frequencies than at mid and high frequencies. Although the compensated group delay curve GD3 has some fluctuations at low frequencies, its amplitude is much smaller than that of the uncompensated group delay curve GD1. Moreover, the compensated group delay curve GD3 has a basically flat and constant group delay value in the mid and high frequency bands, that is, it achieves linear phase.

[0055] The above embodiments of the present invention achieve at least the following beneficial technical effects:

[0056] 1. This invention starts from the physical characteristics of the all-pass filter and calculates the number of sub-filters and initial parameter values ​​of the all-pass filter based on the area covered by the group delay error frequency response curve. This avoids the stability problems caused by using various global search methods to obtain the optimal number of compensation filters and greatly reduces the amount of computation.

[0057] 2. After obtaining the initial values ​​of the compensation filter parameters, the accuracy of the compensation filter can be further optimized by combining the optimization algorithm. A trade-off can also be made between performance and computational load, such as compensation accuracy, number of filters and maximum delay.

[0058] 3. The embodiments of the present invention are not limited to compensating the group delay of each frequency point of the loudspeaker system into a horizontal straight line to achieve an ideal linear phase effect and obtain a more realistic listening experience. Instead, they can also be compensated into a target group delay curve of any shape to change the harmonic delay characteristics of the loudspeaker, thereby producing a variety of unique tonal effects.

[0059] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0060] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Each block shown in the figures can be subdivided into multiple sub-blocks, each sub-block implementing a related function or step, so that multiple sub-blocks can achieve the function implemented by a larger block before subdivision. Alternatively, multiple blocks shown in the figures can also be combined into a single block, which can achieve the functions of the multiple blocks before merging. In this application, words such as "comprising," "including," "having," etc., are open-ended terms meaning "including but not limited to," and are used interchangeably with them. The terms "or" and "and" as used herein refer to the terms "and / or," and are used interchangeably with them unless the context clearly indicates otherwise. The term "such as" as used herein refers to the phrase "such as but not limited to," and is used interchangeably with it.

[0061] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0062] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0063] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A design method for an all-pass filter for time delay compensation of loudspeaker groups, comprising: Based on the group delay frequency response curve of the loudspeaker to be compensated and the target group delay frequency response curve, the target compensation curve is determined. Determine the number of sub-filters of the all-pass filter used to achieve the target compensation curve and the order of each sub-filter; The target compensation curve is divided into corresponding frequency bands based on the number and order of the sub-filters, and then the parameters of the sub-filters used to compensate for the corresponding frequency bands are determined. The group delay frequency response curve of the all-pass filter is calculated based on the parameters of each sub-filter, and the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve is determined. as well as Optimize the number and / or parameters of the sub-filters to reduce the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve. The determination of the number of sub-filters of the all-pass filter used to achieve the target compensation curve and the order of each sub-filter includes: Calculate the integral of the target compensation curve with respect to frequency to obtain the phase change. The number of sub-filters and the order of each sub-filter are determined such that the sum of the phase changes of each sub-filter on the positive half-axis of the frequency is equal to or close to the phase change of the target compensation curve. The term "close" means that the difference between the sum of the phase changes of each sub-filter on the positive half-axis of the frequency and the phase change of the target compensation curve is less than π.

2. The method of claim 1, further comprising: Combining two or more low-order sub-filters into a high-order sub-filter.

3. The method of claim 1, wherein, The group delay frequency response curve of the loudspeaker to be compensated is determined by measuring or calculating the group delay frequency response curve of the loudspeaker after processing by the minimum phase equalizer.

4. The method of claim 1, wherein, Determining the target compensation curve includes: The target compensation curve is determined based on the difference between the target group delay frequency response curve and the group delay frequency response curve of the loudspeaker to be compensated. The target compensation curve has a group delay value greater than or equal to zero at each frequency point.

5. The method of claim 4, wherein, The target compensation curve is the sum of the difference between the target group delay frequency response curve and the group delay frequency response curve of the loudspeaker to be compensated, plus a constant.

6. The method of claim 1, wherein, When the target compensation curve is divided into corresponding frequency bands according to the number and order of the sub-filters, the integral of the target compensation curve over the frequency band to be compensated by the sub-filters is approximately equal to the phase change of the corresponding sub-filters.

7. The method of claim 1, wherein, Determining the parameters of the sub-filter used for compensation in the corresponding frequency band includes: Calculate the complex coefficients of the transfer function of the corresponding sub-filter based on the start frequency, end frequency, and sampling frequency of the frequency band; and Multiply the transfer function of the sub-filter with complex coefficients by the transfer function with conjugate complex coefficients to obtain the real coefficients of the transfer function of order twice.

8. The method of claim 1, wherein, The calculation of the group delay frequency response curve of the all-pass filter based on the parameters of each sub-filter includes: The sub-filters are cascaded in order corresponding to the frequency bands of the target compensation curve to form the all-pass filter; and The group delay frequency response curve of the all-pass filter is calculated using the parameters of each sub-filter.

9. The method of claim 1, wherein, Optimizing the number and / or parameters of sub-filters includes: Adjust the number of sub-filters and / or the parameters of at least one sub-filter until the deviation between the group delay frequency response curve of the all-pass filter and the target compensation curve falls within a predetermined range.

10. An all-pass filter for time delay compensation of a loudspeaker group, wherein, The all-pass filter is designed according to the method described in any one of claims 1-9.

11. An audio signal processing system, comprising: Minimum phase equalizer, used to adjust the gain of each frequency point of an audio signal; as well as An all-pass filter is used to compensate for the group delay at various frequency points of an audio signal. The all-pass filter is designed according to the method described in any one of claims 1-9.

Images