A signal processing method for increasing the perception angle of a two-speaker crosstalk cancellation system

Abstract

The application discloses a signal processing method for increasing the perception angle of a two-speaker crosstalk cancellation system. The method first determines the azimuth angle of the left and right speakers actually arranged in the two-speaker crosstalk cancellation system, performs first crosstalk cancellation filtering processing, and plays back virtual speakers (not actually existing) with a playback angle of ±90°. Due to the limitation of playback, for a virtual sound source in the lateral direction, the actual perceived virtual sound image azimuth angle is often smaller than the set target angle of the virtual sound source. Therefore, the actual perceived angle of the virtual speaker with the angle of ±90° needs to be obtained, and the two virtual speakers are subjected to second crosstalk cancellation filtering processing based on the actual perceived angle, so as to increase the control ability of the lateral sound image, generate a virtual sound image with a target angle, and finally feed the signals subjected to the two times of crosstalk cancellation processing to the speakers, so as to increase the perception angle of the crosstalk cancellation system and improve the positioning performance of the system.

Description

Technical Field

[0001] This invention relates to the field of virtual sound signal playback based on loudspeakers, and more specifically to a signal processing method for increasing the sensing angle of a two-loudspeaker crosstalk cancellation system. Background Technology

[0002] Binaural signals (stereo signals) can be obtained through binaural synthesis or artificial head pickup, and are typically reproduced using headphones or speakers. When binaural signals are reproduced through headphones, no additional signal processing is required, but negative phenomena such as confusion between front and back, and head positioning may occur for the listener. Reproducing binaural signals directly through speakers avoids these phenomena, but introduces crosstalk from the speaker to the opposite ear. The presence of crosstalk distorts the perceived direction of sound. Therefore, binaural signals need to be processed by a crosstalk cancellation filter before being fed to the speaker. Typically, crosstalk cancellation filters are calculated based on the head-related transfer function (HRTF). The HRTF refers to the transfer function from a point sound source to the listener's two ears, measured or calculated under free-field conditions. Therefore, the corresponding crosstalk cancellation system works best in an anechoic environment. In actual binaural reproduction, the reproduction environment is often not an anechoic environment. This means that the binaural transfer function and the head-related transfer function are inconsistent in the actual environment. This inconsistency leads to a decrease in the localization performance of the crosstalk cancellation system, that is, the angle perceived by the listener is smaller than the actual reproduction angle, thus disrupting the listener's listening experience. In addition, due to the limitations of the speaker placement angle, speakers with a small placement angle (such as the classic stereo arrangement of ±30°) have poor control over lateral sound sources. It is difficult to reproduce and amplify virtual lateral sound sources (such as lateral sound sources greater than ±60°), which often results in the listener's perceived angle being smaller than the set target angle of the virtual sound source. Therefore, the poor control of lateral sound sources by speakers with a small placement angle is also a reason for the decrease in the localization performance of the crosstalk cancellation system.

[0003] In recent years, research has proposed solutions to address the reduced localization performance of crosstalk cancellation systems. For example, increasing the speaker placement angle can improve the lateral control capability of the speakers (Li Meiqing, Xie Bosun, Liu Lulu. Binaural sound pressure control and localization performance of virtual sound reproduction by two speakers); real-time measurement of the listener's binaural transfer function and calculation of crosstalk cancellation filters can generate an adaptive crosstalk cancellation system (Kabzinski T, Jax P. An adaptive crosstalk cancellation system using microphones at the ears[C] / / Audio Engineering Society Convention 147.Audio Engineering Society,2019.). These methods improve localization performance by altering the physical level of the system, such as changing the original speaker placement, which may pose difficulties for speaker placement in real-world environments; or adding additional hardware (microphones), requiring the listener to wear a pair of microphones for real-time recording, resulting in a poor listening experience and unsuitability for practical applications. In summary, these methods bring inconvenience to the practical application of crosstalk cancellation. Summary of the Invention

[0004] This invention relates to crosstalk cancellation systems and proposes a signal processing method to increase the sensing angle of a two-speaker crosstalk cancellation system. First, the azimuth angles of the actual left and right speakers in the two-speaker crosstalk cancellation system are determined, and a first crosstalk cancellation filtering process is performed. Virtual speakers (not actually existing) with a playback angle of ±90° are then used. Due to playback limitations, for lateral virtual sound sources, the actual perceived azimuth angle of the virtual sound image is often smaller than the set target angle of the virtual sound source. Therefore, it is necessary to obtain the actual perceived angle of the ±90° virtual speakers. Based on the actual perceived angle, a second crosstalk cancellation filtering process is performed on these two virtual speakers to increase the control capability over the lateral sound image, generating a virtual sound image at the target angle. Finally, the signal after the two crosstalk cancellation processes is fed back to the speakers to increase the sensing angle of the crosstalk cancellation system and improve the system's positioning performance.

[0005] The objective of this invention is achieved by at least one of the following technical solutions.

[0006] A signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system includes the following steps:

[0007] S1: Set up a coordinate reference system, determine the azimuth angles of the actual loudspeaker L and loudspeaker R, and perform the first crosstalk cancellation process on the paths of loudspeaker L and loudspeaker R. The target sound sources for crosstalk cancellation are virtual loudspeakers V at ±90°.R and virtual speaker V L The target signals are the virtual loudspeaker V. R and virtual speaker V L input signal and Calculate the input signals of speaker L and speaker R at this time;

[0008] S2: Obtain the ±90° virtual speaker V respectively. L and virtual speaker V R Perspective and And the transfer function obtained based on the corresponding sensing angle;

[0009] S3: Based on virtual speaker V R and virtual speaker V L The transfer function of the perceived angle for the virtual loudspeaker V R and virtual speaker V L The path undergoes a second crosstalk cancellation process to obtain the virtual speaker V at various target angles. R and virtual speaker V L input signal and That is, the target signal of the actual loudspeaker L and loudspeaker R;

[0010] S4: Transfer the target signal obtained in step S3 Substituting the input signals of speaker L and speaker R, we obtain the total input signal S of speaker L and speaker R respectively. L S R The signal is fed to the corresponding loudspeakers L and R to reproduce the sound source at the target angle.

[0011] Further, in step S1, a planar reference coordinate system is set up, defining the center position of the listener's head as the origin O, the x-axis pointing vertically from the origin to the right of the listener, and the y-axis pointing horizontally from the origin to the front of the listener.

[0012] Furthermore, in step S1, the actual left and right loudspeakers are loudspeaker L and loudspeaker R, respectively, and their azimuth angles are γ. L and γ R azimuth angle γ L and γ R The arrangement angle is determined by the arrangement angle of the loudspeaker L and the loudspeaker R respectively. The arrangement angle is the sum of the absolute values ​​of the azimuth angles of the loudspeaker L and the loudspeaker R, and the arrangement angle generally does not exceed 120°.

[0013] Further, in step S1, the paths of loudspeaker L and loudspeaker R undergo a first crosstalk cancellation process, with the target sound sources for crosstalk cancellation being virtual loudspeakers V at ±90°. R and virtual speaker V L The target signals are the virtual loudspeaker V. L and virtual speaker V R input signal The formulas for calculating the input signals of loudspeakers L and R when the target sound source direction is ±90° are as follows:

[0014]

[0015] In formula (1), and The virtual loudspeaker V represents the target sound source at -90°. L The input signals for speakers L and R are as follows. and The virtual loudspeaker V represents the target sound source at a 90° angle. R The input signals H of speakers L and R. L,-90 and H R,-90 This represents the far-field binaural transfer function of a sound source with an azimuth angle of -90°. The far field generally refers to the distance between the sound source and the listener being more than 1 meter. L,90 and H R,90 This represents the far-field binaural transfer function of a sound source with an azimuth angle of 90°. Virtual speaker V R and virtual speaker V L The input signal, due to The solution is unknown at this stage, and can only be fully solved in subsequent stages. and and This represents the crosstalk cancellation matrix corresponding to the H2 matrix. Abbreviated as C2, the crosstalk cancellation matrix C2 is the azimuth angle γ L γ R The inverse matrix of the far-field sound source correlation transfer function matrix H2 is given. The method for matrix inversion is not specifically specified, including direct inversion or pseudo-inversion. The azimuth angle is γ. L γ R The far-field source correlation transfer function matrix H2 is expressed as:

[0016]

[0017] In formula (2), H2 LL H2 represents the transfer function from loudspeaker L to the left ear. RLH2 represents the transfer function from speaker R to the left ear. LR H2 represents the transfer function from speaker L to the right ear. RR The submatrix C2 in C2 represents the transfer function from speaker R to the right ear. LL C2 RL C2 LR C2 RR Indicates H2 LL H2 RL H2 LR H2 RR The corresponding crosstalk cancellation matrix.

[0018] Furthermore, in step S2, due to the playback limitations of the speaker system, for a lateral virtual sound source, the actually perceived virtual sound image azimuth angle is often smaller than the set target angle of the virtual sound source, thus obtaining the virtual speaker V. L and virtual speaker V R Perspective and Perspective Virtual speaker V L The corresponding binaural transfer function is denoted as H1. LL and H1 LR Perspective Virtual speaker V R The corresponding binaural transfer function is denoted as H1. RL and H1 RR .

[0019] Further, in step S2, the virtual speaker V is obtained. L and virtual speaker V R Methods for perceiving from different perspectives include subjective experimental listening or existing literature data.

[0020] Further, in step S3, the target angle refers to the target azimuth angle of the virtual sound image reproduced in the crosstalk cancellation system, which is located in the front half of the horizontal plane, that is, the angle range is [-90°, 90°], and the binaural transfer function of the virtual sound source is H. L H R It changes with the azimuth angle of the sound source;

[0021] Based on virtual loudspeaker V L and virtual speaker V R The transfer function of the perceived angle direction for the virtual loudspeaker V L and virtual speaker V R The path undergoes a second crosstalk cancellation process to obtain the virtual speaker V at various target angles. L and virtual speaker V R input signal That is, the target signal of the actual loudspeaker L and loudspeaker R.

[0022] Furthermore, in step S3, the virtual speaker V L and virtual speaker V R input signal It can be obtained from the following formula:

[0023]

[0024] Where H L H R Represents the far-field transfer function of a sound source at any target angle. Virtual speaker V L and virtual speaker V R The input signal, E0, represents any single-channel acoustic signal;

[0025] This represents the crosstalk cancellation matrix corresponding to matrix H1, abbreviated as C1. Crosstalk cancellation matrix C1 is the crosstalk cancellation matrix with an azimuth angle of... The inverse matrix of the far-field sound source correlation transfer function matrix H1 is given. The method for matrix inversion is not specifically specified, including direct inversion or pseudo-inversion. The azimuth angle is... The far-field source correlation transfer function matrix H1 is expressed as:

[0026]

[0027] In formula (4), H1 LL Indicates the perception angle as Virtual speaker V L The transfer function to the left ear, H1 RL Indicates the perception angle as Virtual speaker V R The transfer function to the left ear, H1 LR Indicates the perception angle as Virtual speaker V L The transfer function to the right ear, H1 RR Indicates the perception angle as Virtual speaker V R The transfer function to the right ear, and the submatrix C1 of C1. LL C1 RL C1 LR C1 RR Indicates H1 LL H1 RL H1 LR H1 RR The corresponding crosstalk cancellation matrix.

[0028] Furthermore, in step S4, the virtual speaker V in formula (3) is... L and virtual speaker V R input signal Substituting into formula (1), calculate the total input signal S of loudspeaker L and loudspeaker R respectively. L S R As shown in the following formula:

[0029]

[0030] In formula (5), S L S R S represents the total input signal of speaker L and speaker R, respectively. L S R These represent the virtual speaker V. L and virtual speaker V R The case where the path is subjected to a second crosstalk cancellation.

[0031] Further, in step S4, the total input signal S of the loudspeakers L and R is... L S R By feeding the corresponding loudspeakers L and R, the virtual sound image at the target angle can be reproduced, and the lateral sensing angle can be significantly increased compared with the initial two-loudspeaker crosstalk cancellation system.

[0032] Compared with the prior art, the advantages and beneficial effects of the present invention include:

[0033] 1. The system's lateral sound source control capability is improved through two crosstalk cancellation processes, and the signal processing is simple and convenient;

[0034] 2. After adjusting the speaker input signal using the above signal processing method, when the target angle is large, the actual perceived angle of the listener can be significantly increased, effectively improving the positioning performance of the crosstalk cancellation system.

[0035] 3. In this invention, the system does not require additional hardware facilities, and the proposed signal processing method will not cause timbre distortion in the system. Attached Figure Description

[0036] Figure 1 A flowchart of the signal processing for increasing the sensing angle of the two-speaker crosstalk cancellation system in an embodiment of the present invention;

[0037] Figure 2 Speaker arrangement diagram of the crosstalk cancellation system in this embodiment of the invention;

[0038] Figure 3 The average sensing angle results under different methods in the embodiments of the present invention are shown in the figure. Detailed Implementation

[0039] The present invention will be further described below with reference to embodiments and accompanying drawings, but the implementation of the present invention is not limited thereto.

[0040] Example:

[0041] A signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system, such as Figure 1 As shown, it includes the following steps:

[0042] S1: Set up a coordinate reference system, determine the azimuth angles of the actual loudspeaker L and loudspeaker R, and perform the first crosstalk cancellation process on the paths of loudspeaker L and loudspeaker R. The target sound sources for crosstalk cancellation are virtual loudspeakers V at ±90°. R and virtual speaker V L The target signals are the virtual loudspeaker V. R and virtual speaker V L input signal and Calculate the input signals of speaker L and speaker R at this time;

[0043] Set up a planar reference coordinate system, defining the center position of the listener's head as the origin O, with the x-axis pointing vertically from the origin to the right of the listener, and the y-axis pointing horizontally from the origin to the front of the listener.

[0044] The actual left and right loudspeakers are loudspeaker L and loudspeaker R, respectively, with azimuth angles γ and γ. L and γ R azimuth angle γ L and γ R The angles of arrangement of loudspeakers L and R are determined respectively, and the angles of arrangement generally do not exceed 120°.

[0045] In this embodiment, the actual left and right speakers are denoted as speaker L and speaker R. Speaker L and speaker R are arranged in a classic stereo configuration with an azimuth angle of ±30° and a distance of 1.5m from the listener. The coordinate system and the arrangement of the speakers and virtual speakers are as follows. Figure 2 As shown, speakers L and R, and virtual speaker V L and V R The virtual speakers are symmetrically distributed and located at the actual perceived angle. and The target virtual sound source P can be located in any direction within the range of [-90°, 90°].

[0046] The first crosstalk cancellation process is performed on the paths of loudspeakers L and R. The target sound sources for crosstalk cancellation are virtual loudspeakers V at ±90°. R and virtual speaker V LThe target signals are the virtual loudspeaker V. L and virtual speaker V R input signal The formulas for calculating the input signals of loudspeakers L and R when the target sound source direction is ±90° are as follows:

[0047]

[0048] In formula (1), and The virtual loudspeaker V represents the target sound source at -90°. L The input signals for speakers L and R are as follows. and The virtual loudspeaker V represents the target sound source at a 90° angle. R The input signals H of speakers L and R. L,-90 and H R,-90 This represents the far-field binaural transfer function of a sound source with an azimuth angle of -90°. The far field generally refers to the distance between the sound source and the listener being more than 1 meter. L,90 and H R,90 This represents the far-field binaural transfer function of a sound source with an azimuth angle of 90°. Virtual speaker V R and virtual speaker V L The input signal, due to The solution is unknown at this stage, and can only be fully solved in subsequent stages. and and This represents the crosstalk cancellation matrix corresponding to the H2 matrix, abbreviated as C2. The crosstalk cancellation matrix C2 is the crosstalk cancellation matrix with an azimuth angle of γ. L γ R The inverse matrix of the far-field sound source correlation transfer function matrix H2 is given. The method for matrix inversion is not specifically specified, and may include, but is not limited to, direct inversion and pseudo-inversion methods. The azimuth angle is γ. L γ R The far-field source correlation transfer function matrix H2 is expressed as:

[0049]

[0050] In formula (2), H2 LL H2 represents the transfer function from loudspeaker L to the left ear. RL H2 represents the transfer function from speaker R to the left ear. LR H2 represents the transfer function from speaker L to the right ear. RR The submatrix C2 in C2 represents the transfer function from speaker R to the right ear. LL C2 RL C2LR C2 RR Indicates H2 LL H2 RL H2 LR H2 RR The corresponding crosstalk cancellation matrix.

[0051] In this embodiment, all binaural transfer functions were obtained from the KEMAR (Knowles Electronics Manikin for Acoustics Research) artificial head database of the Institute of Acoustics, South China University of Technology;

[0052] In this embodiment, the pseudo-inverse method is used to invert the H2 matrix (Kirkeby O, Nelson P A. Digital filter design for inversion problems in sound reproduction.). The inversion process can be represented as follows:

[0053] C2=(H2 H H2+λI) -1 H2 H

[0054] H2 H λ represents the conjugate transpose of the H2 matrix, and λ is the regularization coefficient. In the trade-off between stability and playback performance of the crosstalk cancellation system, the regularization coefficient λ is set to 0.001.

[0055] S2: Obtain the ±90° virtual speaker V respectively. R and virtual speaker V L Perspective and And the transfer function obtained based on the corresponding sensing angle;

[0056] Due to the playback limitations of loudspeaker systems, for lateral virtual sound sources, the perceived azimuth angle of the virtual sound image is often smaller than the set target angle of the virtual sound source. This is necessary to obtain the virtual loudspeaker V... L and virtual speaker V R Perspective and in Perspective Virtual speaker V L The corresponding binaural transfer function is denoted as H1. LL and H1 LR Perspective Virtual speaker V R The corresponding binaural transfer function is denoted as H1. RL and H1 RR .

[0057] Get the virtual speaker V L and virtual speaker V R Methods for perceiving from different perspectives include, but are not limited to, subjective experimental listening, and existing literature data.

[0058] S3: Based on virtual speaker V R and virtual speaker V L The transfer function of the perceived angle for the virtual loudspeaker V R and virtual speaker V L The path undergoes a second crosstalk cancellation process to obtain the virtual speaker V at various target angles. R and virtual speaker V L input signal and That is, the target signal of the actual loudspeaker L and loudspeaker R;

[0059] The target angle refers to the target azimuth angle of the virtual sound image reproduced in the crosstalk cancellation system. It is located in the front half of the horizontal plane, that is, the angle range is [-90°, 90°]. The binaural transfer function of the virtual sound source is H. L H R It changes with the azimuth angle of the sound source;

[0060] Based on virtual loudspeaker V L and virtual speaker V R The transfer function of the perceived angle direction for the virtual loudspeaker V L and virtual speaker V R The path undergoes a second crosstalk cancellation process to obtain the virtual speaker V at various target angles. L and virtual speaker V R input signal That is, the target signal of the actual loudspeaker L and loudspeaker R.

[0061] Virtual Speaker V L and virtual speaker V R input signal It can be obtained from the following formula:

[0062]

[0063] Where H L H R Represents the far-field transfer function of a sound source at any target angle. Virtual speaker V L and virtual speaker V R The input signal, E0, represents any single-channel acoustic signal;

[0064] This represents the crosstalk cancellation matrix corresponding to matrix H1, abbreviated as C1. Crosstalk cancellation matrix C1 is the crosstalk cancellation matrix with an azimuth angle of... The inverse matrix of the far-field sound source correlation transfer function matrix H1 is given. The method for matrix inversion is not specifically specified, and may include, but is not limited to, direct inversion and pseudo-inversion methods. The azimuth angle is given. The far-field source correlation transfer function matrix H1 is expressed as:

[0065]

[0066] In formula (4), H1 LL Indicates the perception angle as Virtual speaker V L The transfer function to the left ear, H1 RL Indicates the perception angle as Virtual speaker V R The transfer function to the left ear, H1 LR Indicates the perception angle as Virtual speaker V L The transfer function to the right ear, H1 RR Indicates the perception angle as Virtual speaker V R The transfer function to the right ear, and the submatrix C1 of C1. LL C1 RL C1 LR C1 RR Indicates H1 LL H1 RL H1 LR H1 RR The corresponding crosstalk cancellation matrix.

[0067] In this embodiment, the pseudo-inverse method is used to invert the H1 matrix. The inversion process can be represented as follows:

[0068] C1=(H1 H H1+λI) -1 H1 H

[0069] H1 H Let H1 be the conjugate transpose of the H1 matrix, and λ be the regularization coefficient, which takes the value 0.001.

[0070] S4: Transfer the target signal obtained in step S3 Substituting the input signals of speaker L and speaker R, we obtain the total input signal S of speaker L and speaker R respectively. L S R The signal is fed to the corresponding loudspeakers L and R to reproduce the sound source at the target angle.

[0071] The virtual speaker V in formula (3) L and virtual speaker V R input signal Substituting into formula (1), calculate the total input signal S of loudspeaker L and loudspeaker R respectively. L S R As shown in the following formula:

[0072]

[0073] In formula (5), S L S R S represents the total input signal of speaker L and speaker R, respectively. L S R These represent the virtual speaker V. L and virtual speaker V R The case where the path is subjected to a second crosstalk cancellation.

[0074] The total input signal S of speaker L and speaker R L S R By feeding the corresponding speakers L and R, the virtual sound image at the target angle can be reproduced. The change in the listener's perceived angle before and after using this method can be compared. Figure 3 As shown, compared with the original two-speaker crosstalk cancellation system, the signal processing method proposed in this invention (i.e., the optimized perception angle curve in the figure) can significantly increase the listener's lateral perception angle.

[0075] As can be seen from the above examples, this invention can perform signal processing on a two-speaker crosstalk cancellation system according to the steps described above. Based on the secondary crosstalk cancellation method, it improves the system's ability to control lateral sound sources by adjusting the speaker's input signal, increases the listener's perception angle, and enhances the localization performance of the crosstalk cancellation system. Compared with traditional methods, this invention does not require additional hardware, the calculation process is simple and convenient, the localization performance is significantly improved, and it does not produce timbre distortion, thus creating a better auditory experience for listeners using the crosstalk cancellation system.

[0076] The above embodiments are implementations of the present invention with obvious effects, but the implementation of the present invention is not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention should be considered equivalent substitutions and are included within the protection scope of the present invention.

Claims

1. A signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system, characterized in that, including the following steps: S1: Set up a coordinate reference system, determine the azimuth angles of the actual loudspeaker L and loudspeaker R, and perform the first crosstalk cancellation process on the paths of loudspeaker L and loudspeaker R. The target sound sources for crosstalk cancellation are virtual loudspeakers V at ±90°. R and virtual speaker V L The target signals are the virtual loudspeaker V. R and virtual speaker V L input signal and Calculate the input signals of speaker L and speaker R at this time; S2: Obtain the ±90° virtual speaker V respectively. L and virtual speaker V R Perspective And the transfer function obtained based on the corresponding sensing angle; S3: Based on virtual speaker V R and virtual speaker V L The transfer function of the perceived angle for the virtual loudspeaker V R and virtual speaker V L The path undergoes a second crosstalk cancellation process to obtain the virtual speaker V at various target angles. R and virtual speaker V L input signal and That is, the target signal of the actual loudspeaker L and loudspeaker R; S4: Transfer the target signal obtained in step S3 , Substitute the input signals of speaker L and speaker R into the input signals to obtain the total input signals of speaker L and speaker R respectively. , The signal is fed to the corresponding loudspeakers L and R to reproduce the sound source at the target angle.

2. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S1, a planar reference coordinate system is set up, and the center position of the listener's head is defined as the origin. O , x The axis points perpendicularly from the origin to the right of the listener. y The axis points horizontally from the origin of the coordinate system directly in front of the listener.

3. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S1, the actual left and right loudspeakers are loudspeaker L and loudspeaker R, respectively, and their azimuth angles are respectively c L and c R azimuth c L and c R The arrangement angle is determined by the arrangement angle of loudspeaker L and loudspeaker R respectively. The arrangement angle is the sum of the absolute values ​​of the azimuth angles of loudspeaker L and loudspeaker R, and the arrangement angle does not exceed 120°.

4. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S1, the paths of loudspeaker L and loudspeaker R undergo a first crosstalk cancellation process. The target sound sources for crosstalk cancellation are virtual loudspeakers V at ±90°. R and virtual speaker V L The target signals are the virtual loudspeaker V. L and virtual speaker V R input signal , The formulas for calculating the input signals of loudspeaker L and R when the target sound source direction is ±90° are obtained as follows: In formula (1), and The virtual loudspeaker V represents the target sound source at -90°. L The input signals for speakers L and R are as follows. and The virtual loudspeaker V represents the target sound source at a 90° angle. R The input signals for speakers L and R are as follows. and This represents the far-field binaural transfer function of a sound source with an azimuth angle of -90°. The far field refers to the distance between the sound source and the listener being more than 1 meter. and This represents the far-field binaural transfer function of a sound source with an azimuth angle of 90°. , Virtual speaker V R and virtual speaker V L The input signal; Indicates and The crosstalk cancellation matrix corresponding to the matrix will Abbreviated as Crosstalk cancellation matrix The azimuth angle is c L , c R Far-field sound source correlation transfer function matrix H2 The inverse matrix is ​​given, and the method for finding the matrix inverse is not specifically specified, including direct inversion or pseudo-inversion. The azimuth angle is given. c L , c R Far-field sound source correlation transfer function matrix H2 Represented as: In formula (2), This represents the transfer function from speaker L to the left ear. This represents the transfer function from speaker R to the left ear. This represents the transfer function from speaker L to the right ear. This represents the transfer function from speaker R to the right ear. submatrix in , , , Indicates and , , , The corresponding crosstalk cancellation matrix.

5. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S2, due to the playback limitations of the speaker system, for a lateral virtual sound source, the actual perceived azimuth angle of the virtual sound image is often smaller than the set target angle of the virtual sound source. The virtual speaker V is then obtained. L and virtual speaker V R Perspective and Perspective Virtual speaker V L The corresponding binaural transfer function is denoted as and Perspective Virtual speaker V R The corresponding binaural transfer function is denoted as and .

6. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S2, the virtual speaker V is obtained. L and virtual speaker V R Methods for perceiving from different perspectives include subjective experimental listening.

7. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S3, the target angle refers to the target azimuth angle of the virtual sound image reproduced in the crosstalk cancellation system, which is located in the front half of the horizontal plane, that is, the angle value range is [ [90°, 90°], the binaural transfer function of the virtual sound source is H L , H R It changes with the azimuth angle of the sound source; Based on virtual loudspeaker V L and virtual speaker V R The transfer function of the perceived angle direction for the virtual loudspeaker V L and virtual speaker V R The path undergoes a second crosstalk cancellation process to obtain the virtual speaker V at various target angles. L and virtual speaker V R input signal , That is, the target signal of the actual loudspeaker L and loudspeaker R.

8. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S3, the virtual speaker V L and virtual speaker V R input signal , It can be obtained from the following formula: in H L , H R Represents the far-field transfer function of a sound source at any target angle. 、 Virtual speaker V L and virtual speaker V R The input signal, Represents any single-path acoustic signal; Indicates and The crosstalk cancellation matrix corresponding to the matrix is ​​abbreviated as: Crosstalk cancellation matrix C1 The azimuth angle is , Far-field sound source correlation transfer function matrix H1 The inverse matrix is ​​given, and the method for finding the matrix inverse is not specifically specified, including direct inversion or pseudo-inversion. The azimuth angle is given. , Far-field sound source correlation transfer function matrix H1 Represented as: In formula (4), Indicates the perception angle as Virtual speaker V L The transfer function to the left ear, Indicates the perception angle as Virtual speaker V R The transfer function to the left ear, Indicates the perception angle as Virtual speaker V L The transfer function to the right ear. Indicates the perception angle as Virtual speaker V R The transfer function to the right ear. C1 submatrix , , , Indicates and , , , The corresponding crosstalk cancellation matrix.

9. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 8, characterized in that, In step S4, the virtual speaker V in formula (3) is... L and virtual speaker V R input signal , Substitute into formula (1) to calculate the total input signals of loudspeaker L and loudspeaker R respectively. , As shown in the following formula: In formula (5), , These represent the total input signals for speaker L and speaker R, respectively. , These represent the virtual speaker V. L and virtual speaker V R The case where the path is subjected to a second crosstalk cancellation.

10. The signal processing method for increasing the sensing angle of a two-speaker crosstalk cancellation system according to claim 1, characterized in that, In step S4, the total input signal of speaker L and speaker R is... , By feeding the corresponding loudspeakers L and R, the virtual sound image at the target angle can be reproduced, and the lateral sensing angle can be significantly increased compared with the initial two-loudspeaker crosstalk cancellation system.

Images