Signal processing method, device, equipment, system, storage medium and program product

Abstract

The invention provides a signal processing method, device, equipment, system, storage medium and program product, and the method comprises the steps: adjusting an input signal based on an audible sound signal outputted by the input signal until a difference value between the outputted audible sound signal and an expected signal is in a preset difference value range. And then, the input signal corresponding to the audible sound signal of which the difference value with the expected signal is within the preset range is used as the signal input to the parametric array loudspeaker, so that the influence of environmental noise is relatively small, and the distortion compensation effect is improved.

Description

Technical Field

[0001] The present application relates to the field of sound processing, and in particular to a signal processing method, apparatus, device, system, storage medium and program product. Background Art

[0002] The parametric array speaker uses ultrasound as a carrier to modulate the audible sound signal on the sideband of the ultrasonic carrier. Due to the nonlinear propagation effect of air, the ultrasound will continuously self-demodulate the audible sound signal during the propagation process, and due to the high directivity of ultrasound, the audible sound signal will be accurately transmitted to the target location.

[0003] The self-demodulation process is nonlinear. Affected by nonlinear factors such as temperature, humidity, pressure, scattering, and heat propagation loss of the air medium, there is a serious distortion between the input signal and the output audible sound signal of the parametric array speaker, which affects the sound quality of the parametric array speaker.

[0004] How to compensate for the distortion between the input signal and the output audible sound signal of the parametric array loudspeaker is a relatively important issue. Summary of the invention

[0005] The present application provides a signal processing method, apparatus, device, system, storage medium and program product to solve the problem of signal distortion.

[0006] In a first aspect, the present application provides a signal processing method, the method comprising:

[0007] Based on the audible sound signal outputted from the input signal, the input signal is adjusted until the difference between the outputted audible sound signal and the expected signal is within a preset range;

[0008] An input signal corresponding to an audible sound signal whose difference with the expected signal is within a preset range is used as a signal input to the parametric array speaker.

[0009] In some embodiments, before the audible sound signal output based on the input signal adjusts the input signal, the method further comprises:

[0010] The input signal is processed based on a nonlinear filtering model to output an audible sound signal.

[0011] In some embodiments, the input signal and the expected signal both include a plurality of sampling points, and the plurality of sampling points of the input signal correspond one-to-one to the plurality of sampling points of the expected signal.

[0012] In some embodiments, the difference between the output audible sound signal and the expected signal is within a preset range, including:

[0013] The difference between the amplitude of at least some sampling points of the output audible sound signal and the amplitude of the corresponding sampling points in the expected signal is within a preset range.

[0014] In some embodiments, the step of adjusting the input signal until the difference between the output audible sound signal and the expected signal is within a preset range comprises:

[0015] The input signal is adjusted by an adaptive filter, and an audible sound signal is output based on the adjusted input signal until the difference between the output audible sound signal and the expected signal is within a preset range.

[0016] In some embodiments, the adaptive filter comprises a normalized least mean square adaptive filter.

[0017] In some embodiments, when the difference is outside a preset range, adjusting the input signal by a normalized least mean square adaptive filter includes:

[0018] When the difference is less than the minimum value of the preset range, reducing the weight vector coefficient of the normalized minimum mean square adaptive filter to reduce the input signal;

[0019] When the difference is greater than the maximum value of the preset range, the weight vector coefficient of the normalized minimum mean square adaptive filter is increased to increase the input signal.

[0020] In some embodiments, after the difference between the output audible sound signal and the expected signal is within a preset range, the method further includes:

[0021] The input signal corresponding to the audible sound signal whose difference with the expected signal is within a preset range is modulated, so that the modulated signal is embedded in a high-frequency carrier to output a highly directional sound wave.

[0022] In some embodiments, the process of establishing the nonlinear filtering model includes:

[0023] Identify first-order time-domain kernels and second-order time-domain kernels;

[0024] Using the first-order time domain kernel and the second-order time domain kernel, correcting the third-order time domain kernel;

[0025] The nonlinear filtering model is constructed using the first-order time domain kernel, the second-order time domain kernel and the modified third-order time domain kernel.

[0026] In some embodiments, the identifying the first-order time domain kernel and the second-order time domain kernel comprises:

[0027] The first-order frequency domain kernel and the second-order frequency domain kernel are identified using the two-tone method;

[0028] Based on the first-order frequency domain kernel and the second-order frequency domain kernel, a first-order time domain kernel and a second-order time domain kernel are obtained by inverse Fourier transform.

[0029] In some embodiments, the modifying the third-order time domain kernel by using the first-order time domain kernel and the second-order time domain kernel comprises:

[0030] Calculate a first-order output signal based on the first-order time domain kernel, and calculate a second-order output signal based on the second-order time domain kernel;

[0031] Subtract the first-order output signal and the second-order output signal from the expected signal to obtain a third-order expected signal;

[0032] The third-order time domain kernel is modified based on the third-order desired signal until the difference between the third-order output signal output based on the modified third-order time domain kernel and the third-order desired signal is within a target range.

[0033] In some embodiments, the modifying the third-order time domain kernel based on the third-order desired signal comprises:

[0034] The third-order time domain kernel is modified based on the third-order desired signal by using a normalized least mean square algorithm.

[0035] In some embodiments, processing the input signal based on a nonlinear filtering model to output an audible sound signal includes:

[0036] The first-order output signal, the second-order output signal, and the third-order output signal whose difference with the third-order expected signal is within a target range are added together to obtain and output an audible sound signal.

[0037] In a second aspect, the present application provides a signal processing device, the device comprising:

[0038] An adjustment module, configured to adjust the input signal based on the audible sound signal output by the input signal until the difference between the output audible sound signal and the expected signal is within a preset range;

[0039] The output module is used to use the input signal corresponding to the audible sound signal whose difference with the expected signal is within a preset range as a signal input to the parametric array speaker.

[0040] In a third aspect, the present application provides a signal processing device, including: a memory and a processor;

[0041] The memory is used to store instructions; the processor is used to call the instructions in the memory to execute the signal processing method in the first aspect and any possible design of the first aspect.

[0042] In a fourth aspect, the present application provides a signal processing system, comprising the above-mentioned signal processing device and a parametric array speaker;

[0043] The signal processing device inputs an input signal corresponding to an audible sound signal whose difference with an expected signal is within a preset range to the parametric array speaker.

[0044] In a fifth aspect, the present application provides a computer-readable storage medium, in which computer instructions are stored. When at least one processor of a signal processing device executes the computer instructions, the signal processing device executes the signal processing method in the first aspect and any possible design of the first aspect.

[0045] In a sixth aspect, the present application provides a computer program product, which includes computer instructions. When at least one processor of a signal processing device executes the computer instructions, the signal processing device executes the signal processing method in the first aspect and any possible design of the first aspect.

[0046] The signal processing method, apparatus, device, system, storage medium and program product provided by the present application adjust the input signal based on the audible sound signal output by the input signal until the difference between the output audible sound signal and the expected signal is within a preset range, and use the input signal corresponding to the audible sound signal whose difference with the expected signal is within the preset range as the signal input to the parametric array speaker, which is less affected by environmental noise and improves the distortion compensation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 A flowchart of a signal processing method provided by an embodiment of the present application;

[0049] Figure 2 A schematic diagram of adjusting an input signal based on an adaptive filter provided in an embodiment of the present application;

[0050] Figure 3 An output signal comparison diagram provided by an embodiment of the present application;

[0051] Figure 4 A flowchart of a signal processing method provided by another embodiment of the present application;

[0052] Figure 5 A schematic diagram of constructing a nonlinear filtering model provided in one embodiment of the present application;

[0053] Figure 6 An output signal comparison diagram provided by another embodiment of the present application;

[0054] Figure 7 A schematic diagram of the structure of a signal processing device provided in one embodiment of the present application;

[0055] Figure 8 A schematic diagram of the hardware structure of a signal processing device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0057] Sound is one of the most important media for people to transmit information, and the environment of daily life is full of various sounds. With the significant improvement of economic and cultural levels, people have higher and higher requirements for the quality of life and working environment, and noise control and sound information protection are becoming more and more important.

[0058] Different situations require different sounds. For example, libraries require a quiet sound environment, while performance venues require a clear sound environment. Even for the same sound, different audiences may have different needs. For example, for outdoor voice prompts, hearing-impaired people require higher loudness than normal hearing people. Another example is that the sound of a vehicle's horn needs to be clearly transmitted to the target group without disturbing others.

[0059] Therefore, traditional speakers that can only emit omnidirectional sound waves cannot meet people's needs, and the technology of constructing a local sound field is becoming increasingly important.

[0060] Currently, directional sound technology can provide different audio information to different users. With the help of this technology, personalized sound transmission can be achieved and the sound environment can be designed more freely.

[0061] For example, parametric array speakers can emit low-frequency sound in a compact structure. They use ultrasound as a carrier to modulate the audible sound signal on the sideband of the ultrasonic carrier. Due to the nonlinear propagation effect of air, the ultrasound will continuously self-demodulate the audible sound signal during the propagation process, and due to the high directivity of ultrasound, the audible sound signal will be accurately transmitted to the target location.

[0062] The parametric array speaker can also adjust the arrangement of the ultrasonic transducer array according to the needs of different applications to obtain different sound beam shapes, which has high flexibility. Therefore, it has a wide range of applications and can achieve a variety of functions, with broad application prospects, market demand and value.

[0063] Since the parametric array speaker uses ultrasonic signals to self-demodulate in the air to produce highly directional audible sound, and the self-demodulation process is nonlinear, it is affected by nonlinear factors such as the temperature, humidity, pressure, scattering, and heat propagation loss of the air medium. The input signal and output audible sound of the parametric array speaker are severely distorted, affecting the sound quality of the parametric array speaker.

[0064] In some embodiments, an approximate solution of the audible sound signal output by the parametric array speaker can be derived based on the Berktay equation, and then the input signal is pre-compensated based on the distortion between the approximate solution and the input signal to reduce the distortion between the input signal of the parametric array speaker and the audible sound signal. However, the Berktay equation is easily affected by environmental noise, resulting in poor compensation effect for distortion.

[0065] In other embodiments, a large amount of data obtained by testing an existing acoustic parametric array system is used to train its neural network model, and its inverse model is trained inversely. The identified inverse model is connected in series to the front of the acoustic parametric array system to realize a "pseudo-linear system". And the pre-processing of the system signal distortion is realized by adding an auxiliary adaptive PID (Proportional Integral Derivative) controller. However, this method requires a large amount of data to train the neural network, which will take a lot of time. When the data is insufficient, the neural network cannot work, and the system contains a PID controller based on a BP (back propagation) network. There are many parameters in the BP network, and a large number of thresholds and weights need to be updated each time, which will lead to a slow convergence speed. For this reason, the present application proposes a signal processing method, which continuously adjusts the input signal based on the audible sound signal output by the input signal, so that the output audible sound signal is closer to the expected signal, effectively compensates for the distortion between the input signal and the audible sound signal of the parametric array speaker, is less affected by environmental noise, and improves the compensation effect. In addition, the number of measurements is less than that of the neural network, the structure is simpler, and the amount of calculation is smaller.

[0066] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0067] An embodiment of the present application provides a scenario diagram of a signal processing method. The signal processing device continuously adjusts the input signal based on the audible sound signal output by the input signal until the difference between the output audible sound signal and the expected signal is within a preset range. The signal processing device inputs the input signal corresponding to the audible sound signal whose difference with the expected signal is within the preset range to the parametric array speaker. After receiving the input signal, the parametric array speaker outputs an audible sound signal close to the expected signal.

[0068] Figure 1 FIG. 1 is a flow chart of a signal processing method provided by an embodiment of the present application. Figure 1 As shown, with the signal processing device as the execution subject, the method of this embodiment may include the following steps:

[0069] S101 : Adjust an input signal based on an audible sound signal outputted from the input signal.

[0070] It should be noted that in the process of self-demodulating audible sound, the parametric array speaker is affected by nonlinear factors such as the temperature and humidity, pressure, heat dissipation, and heat propagation loss of the air medium, and the self-demodulated audible sound is distorted, that is, there is a difference between the input signal and the output signal of the parametric array speaker.

[0071] In the embodiment of the present application, when the difference between the expected signal and the audible sound signal is outside the preset range, the input signal is adjusted so that the expected signal can be output based on the adjusted input signal. It can be understood that when the error between the audible sound signal output based on the input signal and the expected signal is large, the input signal is adjusted.

[0072] For example, the expected signal may be an input signal, or a signal close to the input signal. For example, the difference between the expected signal and the input signal is within a preset range. Specifically, the expected signal and the input signal both include multiple sampling points, and the multiple sampling points of the expected signal correspond one-to-one to the multiple sampling points of the input signal. The difference between the expected signal and the input signal being within a preset range may include that the difference between the amplitude of a first preset number of sampling points of the expected signal and the amplitude of the sampling points corresponding to the input signal is within a preset range. The preset range and the first preset number may be determined according to actual conditions.

[0073] For example, the input signal and the expected signal have multiple sampling points, and the multiple sampling points of the input signal correspond one-to-one with the multiple sampling points of the expected signal. Accordingly, the audible sound signal also includes multiple sampling points, and the multiple sampling points of the audible sound signal correspond one-to-one with the multiple sampling points of the expected signal, so that the audible sound signal and the expected signal can be compared.

[0074] In some examples, the input signal may be adjusted until the amplitude of at least some sampling points of the output audible sound signal is within a preset range from the amplitude of corresponding sampling points in the expected signal, so that the audible sound signal is close to the expected signal.

[0075] Accordingly, when the amplitude of at least part of the sampling points of the output audible sound signal is outside the preset range with the amplitude of the corresponding sampling points in the expected signal, the input signal is adjusted. For example, when the difference between the amplitude of the second preset number of sampling points of the expected signal and the amplitude of the corresponding sampling points of the audible sound signal is outside the preset range, the input signal is adjusted. The second preset number can be determined according to actual conditions.

[0076] In some examples, when the difference between the expected signal and the audible sound signal simulated based on the input signal is outside a preset range, the input signal can be adjusted by an adaptive filter, and then the audible sound signal can be output based on the adjusted input signal until the difference between the output audible sound signal and the expected signal is within a preset range. The adaptive filter is a filter that can automatically adjust the filter parameters, and can compensate for the distortion between the expected signal and the audible sound signal in real time, so that the audible sound signal can be close to the expected signal.

[0077] As an implementation method, the adaptive filter is a NLMS (Normalization least-mean-square) adaptive filter. The NLMS adaptive filter has strong adaptability, high real-time performance, fast convergence speed, and small steady-state error, so as to make the audible sound signal closer to the expected signal.

[0078] Specifically, the weight vector coefficients of the NLMS adaptive filter are:

[0079] ω(n)=[ω1,ω2,…,ω M ] T (1)

[0080] Among them, M is the filter length, that is, the order of the filter;

[0081] The input vector at time n is:

[0082] x(n)=[x(n-0),…,x(n-M+1)] T (2)

[0083] The output of the adaptive filter at time n is:

[0084] y(n)=x(n) T ω(n) (3)

[0085] The error signal is:

[0086] e(n)=VF(y(n))-E(n) (4)

[0087] Wherein, VF(·) represents a nonlinear filtering model, VF(y(n)) represents an audible sound signal output based on an input signal, and E(n) represents a desired signal;

[0088] The formula for updating the filter weight vector coefficients is:

[0089]

[0090] Where μ is the step size of the adaptive filter, δ is a number less than 1, usually 0.0001, to prevent the inner product of the input data vector x(n) from being too small and causing system mutation or oscillation.

[0091] Therefore, the weight vector coefficient ω of the NLMS adaptive filter can be adjusted M The input signal y(n) is adjusted. Then, an error signal e(n) is calculated between the expected signal E(n) and the audible sound signal E(n)′ corresponding to the adjusted input signal until the error signal e(n) is within a preset range.

[0092] For example, when the difference is less than the minimum value of the preset range, the output audible sound signal is greater than the expected signal, and the weight vector coefficient of the normalized minimum mean square adaptive filter can be reduced to reduce the input signal so that the audible sound signal output based on the adjusted input signal is close to the expected signal. The output audible sound signal being greater than the expected signal may include that the difference between the amplitude of the second preset number of sampling points in the output audible sound signal and the amplitude of the corresponding sampling points in the expected signal is outside the preset range.

[0093] When the difference is greater than the maximum value of the preset range, the expected signal is greater than the output audible sound signal, and the weight vector coefficient of the normalized minimum mean square adaptive filter can be increased to increase the input signal so that the audible sound signal output based on the adjusted input signal is close to the expected signal. The expected signal is greater than the output audible sound signal may include that the difference between the amplitude of the second preset number of sampling points in the expected signal and the amplitude of the corresponding sampling points in the simulated audible sound signal is outside the preset range.

[0094] In other implementations, the adaptive filter may also include other adaptive filters such as LMS (Least-mean-square) adaptive filters.

[0095] S102: until the difference between the output audible sound signal and the expected signal is within a preset range.

[0096] It should be noted that step S101 may be executed repeatedly until the difference between the output audible sound signal and the expected signal is within a preset range.

[0097] In some embodiments, after the difference between the audible sound signal output based on the adjusted input signal and the expected signal is within a preset range, the input signal corresponding to the audible sound signal whose difference with the expected signal is within the preset range is modulated before entering the parametric array speaker, so that the modulated signal is embedded in a high-frequency carrier and a highly directional sound wave is output, so that the parametric array speaker can accurately control the propagation direction of the sound and reduce the diffusion and interference of the sound. For example, ultrasonic waves (usually at a frequency above 20kHz) can be used as carrier signals, and audio signals (usually in the frequency range of 20Hz to 20kHz) can be embedded in a high-frequency carrier through modulation.

[0098] For example, the input signal may be modulated by amplitude-shift modulation (AM), double-sideband modulation (DSB), single-sideband modulation (SSB), or the like.

[0099] S103: using an input signal corresponding to an audible sound signal whose difference with an expected signal is within a preset range as a signal input to a parametric array speaker.

[0100] Accordingly, the reference speaker can output an audible sound signal that is close to the expected signal.

[0101] The signal processing method provided in the embodiment of the present application adjusts the input signal based on the audible sound signal output by the input signal, and continuously adjusts until the difference between the audible sound signal output based on the adjusted input signal and the expected signal is within a preset range, and uses the input signal corresponding to the audible sound signal whose difference with the expected signal is within the preset difference range as the signal input to the parametric speaker, thereby improving the compensation effect.

[0102] An embodiment of the present application provides a signal processing method. Before adjusting the input signal based on the audible sound signal outputted from the input signal, the method of this embodiment may include the following steps:

[0103] S201 : Process an input signal using a nonlinear filtering model and output an audible sound signal.

[0104] In this embodiment, the nonlinear filtering model can process the distortion of the input signal within at least three orders to output an audible sound signal corresponding to the input signal. For example, the audible sound signal simulated by the nonlinear filtering module is the audible sound signal output by the parametric array speaker based on the input signal.

[0105] In some embodiments, considering that the distortion components above the third order of the parametric array loudspeaker are very small, only the first three orders may be considered. Specifically, the nonlinear filter is constructed based on the first-order time domain kernel, the second-order time domain kernel and the third-order time domain kernel. The first-order time domain kernel represents the component of the linear output of the parametric array loudspeaker, the second-order time domain kernel represents the second harmonic distortion, and the third-order time domain kernel represents the third harmonic distortion.

[0106] Accordingly, the nonlinear filtering model can simulate the third-order distortion of the input signal to simulate the audible sound signal corresponding to the input signal, thereby improving the accuracy of the simulated audible sound signal.

[0107] Moreover, in the parametric array loudspeaker, the near field and the third-order nonlinear effect are closely related. The near field area is the area with the strongest nonlinear effect. Therefore, the third-order nonlinear effect is stronger in the near field area, and the audible sound waves generated are more obvious. Therefore, the nonlinear filter can simulate the distortion of the near field to simulate an audible sound signal that is closer to the output signal of the parametric array loudspeaker. Here, the output signal of the parametric array loudspeaker is an audible sound signal output based on the input signal, which is distorted compared to the input signal.

[0108] Therefore, if Figure 2 As shown, the weight vector coefficient ω of the NLMS adaptive filter can be adjusted M The input signal y(n) is adjusted, and the nonlinear filter model outputs the corresponding audible sound signal E(n)′ based on the adjusted input signal. Then, the error signal e(n) between the expected signal E(n) and the audible sound signal E(n)′ corresponding to the adjusted input signal is calculated until the error signal e(n) is within a preset range.

[0109] In the above embodiment, the nonlinear filtering model can approximate the parametric array loudspeaker. After the nonlinear filtering model simulates the output of the parametric array loudspeaker, the input signal is dynamically adjusted through the adaptive filter to make it close to the expected signal.

[0110] For example, when the difference is less than the minimum value of the preset range, the simulated audible sound signal is greater than the expected signal, and the weight vector coefficient of the normalized minimum mean square adaptive filter can be reduced to reduce the input signal so that the audible sound signal simulated by the nonlinear filter based on the adjusted input signal is close to the expected signal. The simulated audible sound signal being greater than the expected signal may include that the difference between the second preset number of sample values ​​in the simulated audible sound signal and the corresponding sample values ​​in the expected signal is outside the preset range.

[0111] When the difference is greater than the maximum value of the preset range, the expected signal is greater than the simulated audible sound signal, and the weight vector coefficient of the normalized minimum mean square adaptive filter can be increased to increase the input signal so that the audible sound signal simulated by the nonlinear filter based on the adjusted input signal is close to the expected signal. The expected signal is greater than the simulated audible sound signal may include that the difference between the second preset number of sample values ​​in the expected signal and the corresponding sample values ​​in the simulated audible sound signal is outside the preset range.

[0112] In order to further verify the solution of the present application, the audible sound signal simulated by the Berktay model and the audible sound signal simulated by the nonlinear filtering model are compared, as shown in FIG. Figure 3 As shown, Figure 3 The frequency domain diagram of the audible sound signal corresponding to the frequency of 1kHz is shown. Figure 3 It can be seen that compared with the output signal of the Berktay model, the audible sound signal simulated by the nonlinear filtering model is closer to the expected signal.

[0113] The signal processing method provided in this embodiment processes the input signal using a nonlinear filtering model, outputs an audible sound signal, and adjusts the input signal based on the audible sound signal. The adjustment is continued until the difference between the audible sound signal output based on the adjusted input signal and the expected signal is within a preset range. The input signal corresponding to the audible sound signal whose difference with the expected signal is within the preset difference range is used as the signal input to the parametric speaker, thereby improving the compensation effect.

[0114] Figure 4 FIG. 4 shows a signal processing method provided by another embodiment of the present application, such as Figure 4 As shown, the signal processing method provided in the embodiment of the present application includes:

[0115] S301 . Identify a first-order time domain kernel and a second-order time domain kernel.

[0116] It should be noted that if a nonlinear system meets the four conditions of limited input energy, time-invariance, continuity and causality, its input-output relationship can be expressed by the Volterra series. For discrete systems, its input-output equations can be given in the form of polynomial series. The Volterra series model is expressed as follows:

[0117]

[0118] Among them, x(n) is the input signal of the system, y(n) is the output signal of the system, the discrete Volterra series model is also called the nonlinear filtering model, h p is called the p-order time domain kernel, M p is the memory length of the p-order time-domain kernel.

[0119] Since the distortion components above the third order of the parametric array loudspeaker are very small, we can only consider the first three order time domain kernels and take the discrete Fourier transform to obtain:

[0120]

[0121] Where X(·) is the Fourier transform of the input signal, Y(·) is the Fourier transform of the output signal, and H p (·) is the p-th order frequency domain kernel.

[0122] The relationship between the time domain kernel and the frequency domain kernel is:

[0123]

[0124] From the above formula, we can see that by performing a p-dimensional Fourier transform on the p-order time domain kernel, the corresponding frequency domain kernel can be obtained.

[0125] In some embodiments, the parameter identification required by the nonlinear filtering model can adopt an adaptive iterative algorithm such as NLMS. Considering that the NLMS recognition accuracy is greatly affected by the signal-to-noise ratio of the identification signal, while the two-tone method is less affected by the signal-to-noise ratio of the identification signal, the kernel function of the identification will be more accurate. Therefore, the two-tone method is adopted to identify the first-order time domain kernel and the second-order time domain kernel.

[0126] Specifically, the input signal is x(n)=a1cos(2πnf1 / f s )+a2cos(2πnf2 / f s ), f s is the sampling frequency, the output is:

[0127]

[0128] Among them, H1(f1) and H2(f2) represent the fundamental frequency signals output by the linear part; H2(f1,f1) and H2(f2,f2) represent the harmonic signal components output by the quadratic nonlinear term; H2(f1,f2) represents the intermodulation distortion and frequency components output by the quadratic nonlinear term; H2(f1,-f2) represents the intermodulation distortion and difference frequency components output by the quadratic nonlinear term.

[0129] According to formula (9), the relationship between the frequency domain output and the frequency domain kernel can be expressed as:

[0130]

[0131] It is worth noting that there are some symmetric relationships between the frequency domain coefficients of the quadratic kernel function:

[0132]

[0133] According to formula (11), the measurement process can be simplified.

[0134] Therefore, the first-order frequency-domain kernel and the second-order frequency-domain kernel can be identified by the dual-tone method. Then, the first-order time-domain kernel and the second-order time-domain kernel can be obtained through inverse Fourier transform.

[0135] In practical applications, a harmonic signal containing two frequency components f1 and f2 is input into the parametric array system, and f1 ∈ [0.2 kHz, 0.4 kHz, …, 5.8 kHz], f2 ∈ [0.4 kHz, 0.6 kHz, …, f1 + 0.2 kHz], f1 < f2. The first-order frequency-domain kernel and the second-order frequency-domain kernel can be obtained by the dual-tone method.

[0136] S302. Use the first-order time-domain kernel and the second-order time-domain kernel to correct the third-order time-domain kernel.

[0137] In some embodiments, as Figure 5 shown, the first-order output signal y1(n) is calculated based on the first-order time-domain kernel h1(n), and the second-order output signal y2(n) is calculated based on the second-order time-domain kernel h2(n). Exemplarily, the first-order output signal y1(n) and the second-order output signal y2(n) can be calculated according to formula (6).

[0138] Then, subtract the first-order output signal y1(n) and the second-order output signal y2(n) from the desired signal E(n) to obtain the third-order desired signal d3(n). Subsequently, correct the third-order time-domain kernel h3(n) based on the third-order desired signal d3(n) until the difference e(n) between the third-order output signal y3(n) output based on the corrected third-order time-domain kernel h3(n) and the third-order desired signal d3(n) is within the target range.

[0139] Correspondingly, based on the first-order time-domain kernel, the second-order time-domain kernel, and the third-order time-domain kernel, an audible signal that is the desired signal or close to the desired signal can be output.

[0140] Exemplarily, the desired signal can be a self-demodulated audible signal collected by a data acquisition device.

[0141] As an implementation manner, the third-order time-domain kernel is corrected based on the third-order desired signal through the NLMS algorithm until the difference between the third-order output signal output by the corrected third-order time-domain kernel and the desired signal is within the preset range.

[0142] Specifically, the weight coefficient of the NLMS algorithm is:

[0143]

[0144] The input vector at time n is:

[0145]

[0146] The third-order output at time n is:

[0147] y3(n)=x3(n) T h3=h3 T x3(n) (14)

[0148] The error signal e(n) is:

[0149] e(n)=d(n)-y1(n)-y2(n)-y3(n)=d3(n)-h3 T x3(n) (15)

[0150] Among them, d3(n) = E(n) - y1(n) - y2(n), and the mean square error is:

[0151]

[0152] The minimum mean square error criterion is to find E[e 2 (n)] minimum h3 T , the LMS algorithm directly uses the single sampling data e 2 (n) replaces the mean square error E[e 2 (n)], and its iteration formula is:

[0153] h3(n+1)=h3(n)+2μe(n)x3(n) (17)

[0154] Where μ is a fixed step size factor, which is used to control the speed and stability of the iterative algorithm. In order to improve the convergence speed of the algorithm, the NLMS algorithm uses a variable step size factor instead of a fixed step size factor. The iterative equation is:

[0155] h3(n+1)=h3(n)+η(n)e(n)x3(n) (18)

[0156] in,

[0157]

[0158] η0 is the correction step constant, and 0<η0<2. In practical applications, 0.01 can be selected. Δ is a small amount introduced to prevent the stability from decreasing due to the small inner product of the input vector. In practical applications, 0.0001 can be selected. The training signal uses the input and output data obtained by the two-tone method.

[0159] So far, the nonlinear filtering model of the parametric array loudspeaker is obtained.

[0160] In order to further verify the established nonlinear filtering model, single-frequency signals with frequency components of 1kHz and 5kHz are used to verify the performance of the nonlinear filtering model. In the verification process, the signal is modulated using the double-sideband modulation method, and the modulation coefficient and carrier frequency are selected as 0.7 and 39kHz respectively. The output signal is normalized based on the amplitude of the fundamental frequency component, and then the output response obtained by the nonlinear filtering is compared with the experimental results. The results are shown in Figure 2. Figure 6 For the convenience of comparison, the nonlinear filtering model and the Berktay model are offset in the frequency domain.

[0161] Depend on Figure 6 It can be seen that the Berktay model can only calculate the second harmonic, the VF model can calculate the third harmonic, and the VF model is more accurate.

[0162] S303, constructing a nonlinear filtering model using a first-order time domain kernel, a second-order time domain kernel and a modified third-order time domain kernel.

[0163] It should be noted that the first-order time domain kernel, the second-order time domain kernel and the modified third-order time domain kernel are used as parameters of the nonlinear filtering model, so that the nonlinear filtering model can simulate the audible sound signal corresponding to the input signal based on formula (6).

[0164] Specifically, the first-order output signal, the second-order output signal, and the third-order output signal whose difference with the third-order expected signal is within the target range are added together to obtain and output the audible sound signal.

[0165] The signal processing method provided in the present application constructs a nonlinear filtering model through a first-order time domain kernel, a second-order time domain kernel and a third-order time domain kernel, so as to simulate the distortion of the input signal within the third order, so as to simulate the audible sound signal corresponding to the input signal and improve the sound quality of the parametric array speaker.

[0166] Figure 7 A schematic diagram of the structure of a signal processing device provided by an embodiment of the present application is shown. Figure 7 As shown, the signal processing device 10 of this embodiment is used to implement the operation corresponding to the signal processing device in any of the above method embodiments. The signal processing device 10 of this embodiment includes:

[0167] An adjustment module 11, configured to adjust the input signal based on the audible sound signal output by the input signal until the difference between the output audible sound signal and the expected signal is within a preset range;

[0168] The output module 12 is used to use the input signal corresponding to the audible sound signal whose difference with the expected signal is within a preset range as a signal input to the parametric array speaker.

[0169] The signal processing device 10 provided in the embodiment of the present application can execute the above method embodiment. Its specific implementation principle and technical effects can be found in the above method embodiment, and this embodiment will not be repeated here.

[0170] Figure 8 FIG. 1 is a schematic diagram showing the hardware structure of a signal processing device provided in an embodiment of the present application. Figure 8 As shown, the signal processing device 20 is used to implement the operations corresponding to the signal processing device in any of the above method embodiments. The signal processing device 20 of this embodiment may include: a memory 21, a processor 22 and a communication interface 23.

[0171] The memory 21 is used to store computer instructions. The memory 21 may include a high-speed random access memory (RAM), and may also include a non-volatile memory (NVM), such as at least one disk memory, and may also be a USB flash drive, a mobile hard disk, a read-only memory, a disk or an optical disk.

[0172] The processor 22 is used to execute the computer instructions stored in the memory to implement the method in the above embodiment. For details, please refer to the relevant description in the above method embodiment. The processor 22 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the invention can be directly embodied as being executed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0173] Optionally, the memory 21 may be independent or integrated with the processor 22 .

[0174] The communication interface 23 may be connected to the processor 22. The processor 22 may control the communication interface 23 to implement the functions of receiving and sending signals.

[0175] The signal processing device provided in this embodiment can be used to execute the above method. Its implementation method and technical effect are similar, and this embodiment will not be repeated here.

[0176] The embodiment of the present application provides a signal processing system, comprising the above-mentioned signal processing device and a parametric array speaker. The signal processing device inputs an input signal corresponding to an audible sound signal whose difference with an expected signal is within a preset range into the parametric array speaker.

[0177] The present application also provides a computer-readable storage medium, in which computer instructions are stored. When the computer instructions are executed by a processor, they are used to implement the methods provided in the various embodiments described above.

[0178] The present application also provides a computer program product, which includes computer instructions stored in a computer-readable storage medium. At least one processor of a device can read the computer instructions from the computer-readable storage medium, and at least one processor executes the computer instructions so that the device implements the methods provided in the various embodiments described above.

[0179] An embodiment of the present application also provides a chip, which includes a memory and a processor, wherein the memory is used to store computer instructions, and the processor is used to call and execute the computer instructions from the memory, so that a device equipped with the chip executes the methods described in the various possible implementation modes above.

[0180] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the above embodiments can still be modified, or some or all of the technical features can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A signal processing method, characterized in that: The method comprises: Based on the audible sound signal outputted from the input signal, the input signal is adjusted until the difference between the outputted audible sound signal and the expected signal is within a preset range; An input signal corresponding to an audible sound signal whose difference with the expected signal is within a preset range is used as a signal input to the parametric array speaker.

2. The method according to claim 1, characterized in that Before adjusting the input signal by outputting the audible sound signal based on the input signal, the method further comprises: The input signal is processed based on a nonlinear filtering model to output an audible sound signal.

3. The method according to claim 1, characterized in that The input signal and the expected signal both include a plurality of sampling points, and the plurality of sampling points of the input signal correspond one-to-one to the plurality of sampling points of the expected signal.

4. The method according to claim 3, characterized in that The difference between the output audible sound signal and the expected signal is within a preset range, including: The difference between the amplitude of at least some sampling points of the output audible sound signal and the amplitude of the corresponding sampling points in the expected signal is within a preset range.

5. The method according to claim 1, characterized in that The step of adjusting the input signal until the difference between the output audible sound signal and the expected signal is within a preset range includes: The input signal is adjusted by an adaptive filter, and an audible sound signal is output based on the adjusted input signal until the difference between the output audible sound signal and the expected signal is within a preset range.

6. The method according to claim 5, characterized in that The adaptive filter comprises a normalized minimum mean square adaptive filter.

7. The method according to claim 6, characterized in that The input signal is adjusted by a normalized least mean square adaptive filter, comprising: When the difference is less than the minimum value of the preset range, reducing the weight vector coefficient of the normalized minimum mean square adaptive filter to reduce the input signal; When the difference is greater than the maximum value of the preset range, the weight vector coefficient of the normalized minimum mean square adaptive filter is increased to increase the input signal.

8. The method according to any one of claims 1 to 7, characterized in that: After the difference between the output audible sound signal and the expected signal is within a preset range, the method further includes: The input signal corresponding to the audible sound signal whose difference with the expected signal is within a preset range is modulated, so that the modulated signal is embedded in a high-frequency carrier to output a highly directional sound wave.

9. The method according to claim 2, characterized in that: The process of establishing the nonlinear filtering model includes: Identify first-order time-domain kernels and second-order time-domain kernels; Using the first-order time domain kernel and the second-order time domain kernel, correcting the third-order time domain kernel; The nonlinear filtering model is constructed using the first-order time domain kernel, the second-order time domain kernel and the modified third-order time domain kernel.

10. The method according to claim 9, characterized in that The identifying of the first-order time domain kernel and the second-order time domain kernel comprises: The first-order frequency domain kernel and the second-order frequency domain kernel are identified using the two-tone method; Based on the first-order frequency domain kernel and the second-order frequency domain kernel, a first-order time domain kernel and a second-order time domain kernel are obtained by inverse Fourier transform.

11. The method according to claim 9, characterized in that The method of modifying the third-order time domain kernel by using the first-order time domain kernel and the second-order time domain kernel comprises: Calculate a first-order output signal based on the first-order time domain kernel, and calculate a second-order output signal based on the second-order time domain kernel; Subtract the first-order output signal and the second-order output signal from the expected signal to obtain a third-order expected signal; The third-order time domain kernel is modified based on the third-order desired signal until the difference between the third-order output signal output based on the modified third-order time domain kernel and the third-order desired signal is within a target range.

12. The method according to claim 11, characterized in that The modifying the third-order time domain kernel based on the third-order desired signal comprises: The third-order time domain kernel is modified based on the third-order desired signal by using a normalized least mean square algorithm.

13. The method according to claim 11 or 12, characterized in that: Processing the input signal based on a nonlinear filtering model to output an audible sound signal includes: The first-order output signal, the second-order output signal, and the third-order output signal whose difference with the third-order expected signal is within a target range are added together to obtain and output an audible sound signal.

14. A signal processing device, characterized in that: The device comprises: A processing module, configured to adjust the input signal based on the audible sound signal output by the input signal until the difference between the output audible sound signal and the expected signal is within a preset range; The output module is used to use the input signal corresponding to the audible sound signal whose difference with the expected signal is within a preset range as a signal input to the parametric array speaker.

15. A signal processing device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 13.

16. A signal processing system, characterized in that: The signal processing device and the parametric array loudspeaker according to claim 15 are included; The signal processing device inputs an input signal corresponding to an audible sound signal whose difference with an expected signal is within a preset range to the parametric array speaker.

17. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which are used to implement the method according to any one of claims 1 to 13 when executed by a processor.

18. A computer program product, characterized in that The computer program product comprises a computer program, which implements the method according to any one of claims 1 to 13 when executed by a processor.

Images