Sound source direction virtualization method, apparatus, device, and medium

Abstract

The application discloses a sound source direction virtualization method and device, terminal equipment and a computer readable storage medium. The method is applied to a sound playback device. The method obtains a first occlusion transfer function and a second occlusion transfer function. The first occlusion transfer function is a transfer function between binaural ears of a user obtained by measuring the user. The second occlusion transfer function is a transfer function between binaural ears of a simulated human body obtained by measuring the simulated human body. The method calculates a correlation function between the second occlusion transfer function and a target transfer function corresponding to a to-be-virtualized sound source angle obtained in advance. The method compensates the target transfer function according to the first occlusion transfer function, the second occlusion transfer function and the correlation function, obtains a personalized transfer function corresponding to the user, and uses the personalized transfer function to virtually play a sound source signal to obtain a hearing effect corresponding to the to-be-virtualized sound source angle. The application improves the adaptability and accuracy of a sound playback device in simulating a sound source direction.

Description

Technical Field

[0001] This invention belongs to the field of signal processing technology, and particularly relates to a virtual method, apparatus, terminal device, and computer-readable storage medium for sound source direction. Background Technology

[0002] With the rapid development of technology, head-mounted open-back audio playback devices such as AR glasses, VR glasses, and Bluetooth audio players for glasses temples have appeared on the market.

[0003] Currently, existing open-back head-mounted sound playback devices can virtualize the direction of a sound source to any direction, allowing the listener to perceive that the sound is coming from a specific direction, thus achieving a spatial audio function. The main method involves creating an HRTF (Head Related Transfer Function) database at arbitrary angles by measuring on a simulated human body beforehand. This database returns the transfer function or filter coefficients for each audio channel based on the spatial angle. Each audio channel is processed by filters with corresponding coefficients, and then synthesized into stereo audio for the left and right channels, which is then played through speakers. The listener can then determine that the sound comes from a specific direction. However, the difference in perception between the two ears due to different directions is mainly caused by ear position, auricular reflection, head and torso obstruction, and reflection. Clearly, different people will not have identical ear shapes and sizes, head shapes and sizes, body shapes and sizes, and torso height and build. Therefore, using a fixed HRTF to virtualize the direction of a sound source is not accurate enough. Summary of the Invention

[0004] The main objective of this invention is to provide a method, apparatus, terminal device, and computer-readable storage medium for simulating sound source direction. The aim is to improve the adaptability and accuracy of sound source direction simulation in sound playback devices.

[0005] To achieve the above objectives, the present invention provides a method for simulating the direction of a sound source, which is applied to a sound playback device. The method includes:

[0006] Obtain a first occlusion transmission function and a second occlusion transmission function, wherein the first occlusion transmission function is the transmission function between the user's ears obtained by measuring the user, and the second occlusion transmission function is the transmission function between the ears of the simulated human body obtained by measuring the simulated human body;

[0007] Calculate the correlation function between the second occlusion transfer function and the target transfer function corresponding to the pre-acquired virtual sound source angle;

[0008] The target transmission function is compensated based on the first occlusion transmission function, the second occlusion transmission function, and the related function to obtain the personalized transmission function corresponding to the user. The personalized transmission function is used to simulate the auditory effect corresponding to the angle of the virtual sound source signal to be played.

[0009] Optionally, the sound playback device includes a first speaker, a first microphone, and a second microphone. The first speaker and the first microphone are installed at a first position of the sound playback device, and the second microphone is installed at a second position of the sound playback device. One of the first position and the second position is the left ear position of the sound playback device, and the other is the right ear position of the sound playback device.

[0010] The steps to obtain the first occlusion transfer function include:

[0011] When measuring a user, a preset audio signal played by a first speaker is recorded through the first microphone and the second microphone to obtain a first sound signal recorded by the first microphone and a second sound signal recorded by the second microphone;

[0012] Calculate the first sound pressure level corresponding to the first sound signal and the second sound pressure level corresponding to the second sound signal;

[0013] The ratio of the second sound pressure level to the first sound pressure level is used as the first occlusion transmission function corresponding to the user.

[0014] Optionally, the steps of obtaining the second occlusion transfer function include:

[0015] When measuring a simulated human body, the preset audio signal played by the first speaker is recorded through the first microphone and the second microphone to obtain the third sound signal recorded by the first microphone and the fourth sound signal recorded by the second microphone;

[0016] Calculate the third sound pressure level corresponding to the third sound signal and the fourth sound pressure level corresponding to the fourth sound signal;

[0017] The ratio of the fourth sound pressure level to the third sound pressure level is used as the second occlusion transmission function corresponding to the simulated human body.

[0018] Optionally, the step of compensating the target transmission function based on the first occlusion transmission function, the second occlusion transmission function, and the related function to obtain the personalized transmission function corresponding to the user includes:

[0019] The ratio of the first occlusion transfer function to the second occlusion transfer function is used as the occlusion transfer change rate between the user and the simulated human body.

[0020] The calibration function is obtained by multiplying the rate of change of occlusion transmission by the correlation function;

[0021] The target transmission function is compensated based on the calibration function to obtain the personalized transmission function corresponding to the user.

[0022] Optionally, after the step of compensating the target transmission function based on the first occlusion transmission function, the second occlusion transmission function, and the related function to obtain the personalized transmission function corresponding to the user, the method further includes:

[0023] Based on the personalized transfer function, the audio source signal to be played is subjected to frequency domain filtering to obtain the audio signal to be played.

[0024] The audio signal to be played is played through the speaker on the sound playback device.

[0025] Optionally, the step of calculating the correlation function between the second occlusion transfer function and the target transfer function corresponding to the pre-acquired virtual sound source angle includes:

[0026] Based on the angle of the virtual sound source, the target transfer function corresponding to the angle of the virtual sound source is obtained from the de novo related transfer function database;

[0027] Extract the fifth sound pressure level corresponding to each of the two ears of the simulated human body from the target transfer function;

[0028] The fifth sound pressure level and the fourth sound pressure level are input into a pre-trained target deep learning model to generate a correlation function between the second occlusion transfer function and the target transfer function.

[0029] Optionally, before the step of inputting the fifth sound pressure level and the fourth sound pressure level into a pre-trained target deep learning model to generate a correlation function between the second occlusion transfer function and the target transfer function, the method further includes:

[0030] Extract the sixth sound pressure level and the seventh sound pressure level of the second microphone in the preset occlusion transmission function and the preset transmission function corresponding to the virtual angle from the preset occlusion transmission function;

[0031] A simulated human training set is established based on the sixth and seventh sound pressure levels;

[0032] The target deep learning model is obtained by training the pre-constructed initial deep learning model based on the simulated human training set.

[0033] Furthermore, to achieve the above objectives, the present invention also provides a sound source direction virtual device, which is applied to a sound playback device, and the sound source direction virtual device includes:

[0034] The measurement module is used to acquire a first occlusion transmission function and a second occlusion transmission function, wherein the first occlusion transmission function is the transmission function between the user's ears obtained by measuring the user, and the second occlusion transmission function is the transmission function between the ears of the simulated human body obtained by measuring the simulated human body.

[0035] The correlation function module is used to calculate the correlation function between the second occlusion transmission function and the target transmission function corresponding to the pre-acquired virtual sound source angle;

[0036] The personalized transmission function module is used to compensate the target transmission function according to the first occlusion transmission function, the second occlusion transmission function and the related function to obtain the personalized transmission function corresponding to the user. The personalized transmission function is used to simulate the auditory effect corresponding to the angle of the sound source signal to be played.

[0037] In addition, to achieve the above objectives, the present invention also provides a terminal device, the terminal device comprising: a memory, a processor, and a sound source direction virtual program stored in the memory and executable on the processor, wherein the sound source direction virtual program of the terminal device, when executed by the processor, implements the steps of the sound source direction virtual method as described above.

[0038] In addition, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a sound source direction virtual program, which, when executed by a processor, implements the steps of the sound source direction virtual method as described above.

[0039] This invention, through measurements of both a user and a simulated human body, obtains a first occlusion transmission function between the user's ears and a second occlusion transmission function between the simulated human's ears. Furthermore, it calculates a correlation function between the second occlusion transmission function and a pre-acquired target transmission function corresponding to the angle of the virtual sound source. Finally, it uses the first occlusion transmission function, the second occlusion transmission function, and the correlation function to perform personalized compensation on the target transmission function, thereby obtaining a personalized transmission function that can virtually reproduce the auditory effect corresponding to the angle of the virtual sound source signal. In other words, by using the personalized transmission function after personalized compensation to virtualize the sound source direction, the adaptability to the user and the accuracy in virtualizing the sound source direction are improved. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the device structure of the terminal device hardware operating environment involved in the embodiments of the present invention;

[0041] Figure 2 This is a flowchart illustrating the steps of the first embodiment of the virtual sound source direction method of the present invention;

[0042] Figure 3 This is a schematic diagram of the device application process involved in an embodiment of the virtual sound source direction method of the present invention;

[0043] Figure 4 This is a schematic diagram of the functional modules of an embodiment of the virtual sound source direction device of the present invention.

[0044] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0045] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0046] Reference Figure 1 , Figure 1 This is a schematic diagram of the hardware operating environment of the terminal device involved in the embodiment of the present invention.

[0047] It should be noted that the terminal device in the embodiments of the present invention can be a sound playback device, headphones, smartphone, personal computer, server, etc. in the field of signal processing technology, and no specific limitation is made here.

[0048] like Figure 1As shown, the terminal device may include: a processor 1001, such as a CPU; a communication bus 1002; a user interface 1003; a network interface 1004; and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0049] Those skilled in the art will understand that Figure 1 The terminal device structure shown does not constitute a limitation on the terminal device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0050] like Figure 1 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a virtual program for the direction of the sound source.

[0051] exist Figure 1 In the terminal shown, network interface 1004 is mainly used to connect to the backend server and communicate data with it; user interface 1003 is mainly used to connect to the client and communicate data with it; and processor 1001 can be used to call the sound source direction virtual program stored in memory 1005 and perform the following operations:

[0052] Obtain a first occlusion transmission function and a second occlusion transmission function, wherein the first occlusion transmission function is the transmission function between the user's ears obtained by measuring the user, and the second occlusion transmission function is the transmission function between the ears of the simulated human body obtained by measuring the simulated human body;

[0053] Calculate the correlation function between the second occlusion transfer function and the target transfer function corresponding to the pre-acquired virtual sound source angle;

[0054] The target transmission function is compensated based on the first occlusion transmission function, the second occlusion transmission function, and the related function to obtain the personalized transmission function corresponding to the user. The personalized transmission function is used to simulate the auditory effect corresponding to the angle of the virtual sound source signal to be played.

[0055] Furthermore, the sound playback device includes a first speaker, a first microphone, and a second microphone. The first speaker and the first microphone are installed at a first position on the sound playback device, and the second microphone is installed at a second position on the sound playback device. One of the first position and the second position is the left ear position of the sound playback device, and the other is the right ear position of the sound playback device. The operation of obtaining the first occlusion transfer function includes:

[0056] When measuring a user, a preset audio signal played by a first speaker is recorded through the first microphone and the second microphone to obtain a first sound signal recorded by the first microphone and a second sound signal recorded by the second microphone;

[0057] Calculate the first sound pressure level corresponding to the first sound signal and the second sound pressure level corresponding to the second sound signal;

[0058] The ratio of the second sound pressure level to the first sound pressure level is used as the first occlusion transmission function corresponding to the user.

[0059] Furthermore, the operation of obtaining the second occlusion transfer function includes:

[0060] When measuring a simulated human body, the preset audio signal played by the first speaker is recorded through the first microphone and the second microphone to obtain the third sound signal recorded by the first microphone and the fourth sound signal recorded by the second microphone;

[0061] Calculate the third sound pressure level corresponding to the third sound signal and the fourth sound pressure level corresponding to the fourth sound signal;

[0062] The ratio of the fourth sound pressure level to the third sound pressure level is used as the second occlusion transmission function corresponding to the simulated human body.

[0063] Further, the operation of compensating the target transmission function based on the first occlusion transmission function, the second occlusion transmission function, and the related function to obtain the personalized transmission function corresponding to the user includes:

[0064] The ratio of the first occlusion transfer function to the second occlusion transfer function is used as the occlusion transfer change rate between the user and the simulated human body.

[0065] The calibration function is obtained by multiplying the rate of change of occlusion transmission by the correlation function;

[0066] The target transmission function is compensated based on the calibration function to obtain the personalized transmission function corresponding to the user.

[0067] Furthermore, after the step of compensating the target transmission function based on the first occlusion transmission function, the second occlusion transmission function, and the related function to obtain the personalized transmission function corresponding to the user, the processor 1001 can also be used to call the sound source direction virtual program stored in the memory 1005 to perform the following operations:

[0068] Based on the personalized transfer function, the audio source signal to be played is subjected to frequency domain filtering to obtain the audio signal to be played.

[0069] The audio signal to be played is played through the speaker on the sound playback device.

[0070] Furthermore, the operation of calculating the correlation function between the second occlusion transfer function and the target transfer function corresponding to the pre-acquired virtual sound source angle includes:

[0071] Based on the angle of the virtual sound source, the target transfer function corresponding to the angle of the virtual sound source is obtained from the de novo related transfer function database;

[0072] Extract the fifth sound pressure level corresponding to each of the two ears of the simulated human body from the target transfer function;

[0073] The fifth sound pressure level and the fourth sound pressure level are input into a pre-trained target deep learning model to generate a correlation function between the second occlusion transfer function and the target transfer function.

[0074] Furthermore, before the step of inputting the fifth sound pressure level and the fourth sound pressure level into the pre-trained target deep learning model to generate the correlation function between the second occlusion transfer function and the target transfer function, the processor 1001 can also be used to call the sound source direction virtual program stored in the memory 1005 to perform the following operations:

[0075] Extract the sixth sound pressure level and the seventh sound pressure level of the second microphone in the preset occlusion transmission function and the preset transmission function corresponding to the virtual angle from the preset occlusion transmission function;

[0076] A simulated human training set is established based on the sixth and seventh sound pressure levels;

[0077] The target deep learning model is obtained by training the pre-constructed initial deep learning model based on the simulated human training set.

[0078] Based on the above structure, various embodiments of the virtual sound source direction method are proposed.

[0079] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the sound source direction virtualization method of the present invention. It should be noted that although the logical order is shown in the flowchart, in some cases, the sound source direction virtualization method of the present invention may execute the steps shown or described in a different order. In this embodiment, the executing entity of the sound source direction virtualization method can be a personal computer, smartphone, or other device; this is not limited in this embodiment. For ease of description, the executing entity is omitted from the description of each embodiment. In this embodiment, the sound source direction virtualization method is applied to a sound playback device, and the sound source direction virtualization method includes:

[0080] Step S10: Obtain the first occlusion transmission function and the second occlusion transmission function, wherein the first occlusion transmission function is the transmission function between the user's ears obtained by measuring the user, and the second occlusion transmission function is the transmission function between the ears of the simulated human body obtained by measuring the simulated human body.

[0081] Obtain the occlusion transfer function obtained by measuring the user (hereinafter referred to as the first occlusion transfer function for distinction) and the occlusion transfer function obtained by measuring the simulated human body (hereinafter referred to as the second occlusion transfer function for distinction), wherein the first occlusion transfer function is the transfer function between the left and right ears of the user, and the second occlusion transfer function is the transfer function between the left and right ears of the simulated human body.

[0082] In one feasible implementation, the first occlusion transmission function includes a left ear occlusion transmission function U_r2l and a right ear occlusion transmission function U_l2r, and the second occlusion transmission function includes a left ear occlusion transmission function H_r2l and a right ear occlusion transmission function H_l2r.

[0083] It should be noted that the first occlusion transfer function mentioned above can be used to measure the occlusion and reflection effect of the head and torso of the user currently wearing the sound playback device on the sound signal. Similarly, the second occlusion transfer function mentioned above can be used to measure the occlusion and reflection effect of the head and torso of the laboratory simulated human body on the sound signal. In addition, it is worth emphasizing that the sound playback device in this invention is a head-mounted open sound playback device, which is convenient for personalized measurements of users or simulated human bodies.

[0084] In one feasible implementation, after the user puts on the sound playback device, the sound playback device automatically performs personalized measurements on the user after powering on to obtain the first occlusion transfer function corresponding to the user; when performing HRTF (Head Related Transfer Function) measurement, a new personalized measurement for a laboratory simulated human body is added to obtain the occlusion transfer function between the left and right ears of the simulated human body when the sound playback device is worn on the head of the simulated human body.

[0085] Step S20: Calculate the correlation function between the second occlusion transmission function and the target transmission function corresponding to the pre-acquired virtual sound source angle;

[0086] Calculate the correlation function between the second occlusion transfer function and the transfer function corresponding to the pre-acquired virtual sound source angle (hereinafter referred to as the target transfer function for distinction);

[0087] It should be noted that the HRTF database is typically generated in a laboratory using measurements on a simulated human body. The measurement process involves placing a speaker at a certain distance in any direction on the simulated human body and playing specific audio. Recordings are taken from microphones at the left and right ears of the simulated human body, and the transfer functions from the speaker to the left and right microphones are calculated, denoted as H_l and H_r. Measurements are then taken by placing a speaker in each direction to obtain H_l and H_r for each corresponding angle.

[0088] In one feasible implementation, the target transfer functions corresponding to the angle of the virtual sound source to be obtained in advance include the left transfer function H_l and the right transfer function H_r. H_r2l is correlated with H_l and H_l2r is correlated with H_r. The reason for their occurrence includes the same human body structure occlusion and reflection. This correlation can be obtained by a large number of different simulated human body measurements and machine learning. The correlation functions are denoted as Rel_l and Rel_r.

[0089] Step S30: Compensate the target transmission function according to the first occlusion transmission function, the second occlusion transmission function and the related function to obtain the personalized transmission function corresponding to the user, wherein the personalized transmission function is used to simulate the auditory effect corresponding to the angle of the virtual sound source signal to be played.

[0090] The target transmission function is compensated based on the first occlusion transmission function, the second occlusion transmission function, and the related function to obtain a transmission function suitable for the user (hereinafter referred to as the personalized transmission function for distinction). The personalized transmission function is used to simulate the auditory effect corresponding to the angle of the virtual sound source signal to be played.

[0091] In one feasible implementation, the target transmission functions H_L and H_R are compensated based on U_r2l and U_l2r in the first occlusion transmission function, H_r2l and H_l2r in the second occlusion transmission function, and related functions Rel_l and Rel_r, to obtain the user's personalized transmission functions U_L and U_R.

[0092] Further, in one feasible embodiment, step S30 includes:

[0093] Step S301: The ratio of the first occlusion transfer function to the second occlusion transfer function is used as the occlusion transfer change rate between the user and the simulated human body;

[0094] The ratio between the first occlusion transfer function between the user's left and right ears and the second occlusion transfer function between the simulated human's left and right ears is calculated, and this ratio is determined as the rate of change of occlusion transfer between the user and the simulated human.

[0095] In one feasible implementation, the impulse response or occlusion transfer function of the sound signal from the user's right ear position to the left ear position is denoted as U_r2l, the impulse response or occlusion transfer function of the sound signal from the user's left ear position to the right ear position is denoted as U_l2r, the impulse response or occlusion transfer function of the sound signal from the right ear position to the left ear position of the simulated human in the laboratory is denoted as H_r2l, and the impulse response or occlusion transfer function of the sound signal from the left ear position to the right ear position of the simulated human in the laboratory is denoted as H_l2r. Then, the rate of change of occlusion transfer between the user's left ear position and the simulated human's left ear position is calculated as U_r2l / H_r2l, and the rate of change of occlusion transfer between the user's right ear position and the simulated human's right ear position is calculated as U_l2r / H_l2r.

[0096] Step S302: Multiply the occlusion transmission change rate by the correlation function to obtain the calibration function;

[0097] Multiplying the aforementioned occlusion transmission change rate and the correlation function yields a calibration function used to compensate for the target transmission function.

[0098] In one feasible implementation, the calibration function for the user's left ear is calculated as (U_r2l / H_r2l)*Rel_l, and the calibration function for the user's right ear is calculated as (U_l2r / H_l2r)*Rel_r.

[0099] Step S303: Based on the calibration function, perform personalized compensation on the target transmission function to obtain the personalized transmission function corresponding to the user.

[0100] The target transfer function is compensated based on the calibration function to obtain the personalized transfer function corresponding to the user.

[0101] In one feasible implementation, the HRTF after compensation for user personalized features is denoted as U_l and U_r. The formula for calculating the left ear transmission function U_l in the personalized transmission function is: U_l=H_l*(U_r2l / H_r2l)*Rel_l; the formula for calculating the right ear transmission function U_r in the personalized transmission function is: U_r=H_r*(U_l2r / H_l2r)*Rel_r.

[0102] It should be noted that H_l and H_r are the left and right ear transmission transfer functions obtained from laboratory measurements simulating a human body. U_r²l / H_r²l (and similarly U_l²r / H_l²r) represent the change in occlusion transmission between the left and right ears of an actual user compared to laboratory measurements of a simulated human body. Multiplying this by the Rel_l (Rel_r) correlation function yields the calibration function for the HRTF, which, when applied to H_l (H_r), results in the calibrated U_l and U_r.

[0103] Furthermore, in one feasible embodiment, after step S30, the method further includes:

[0104] Step A10: Perform frequency domain filtering on the audio source signal to be played based on the personalized transfer function to obtain the audio signal to be played.

[0105] Step A20: Play the audio signal to be played through the speaker on the sound playback device.

[0106] The pre-acquired audio source signal to be played is frequency domain filtered based on the compensated personalized transfer function to obtain the filtered audio signal to be played, and then the audio signal to be played is played through the speaker on the sound playback device.

[0107] It should be noted that the speaker on the aforementioned sound playback device can be the left speaker (close to the user's left ear) or the right speaker (close to the user's right ear).

[0108] In one feasible implementation, the audio source signal to be played is processed by the calibrated HRTF functions U_l and U_r respectively to obtain the audio signals of the left and right channels, and the audio data of the left and right channels are mapped to the left and right speakers for playback.

[0109] In another feasible implementation, such as Figure 3As shown, the application flow of the virtual sound source direction method of the present invention is as follows: First, step 001: After the user wears and starts the sound playback device, the sound playback device automatically performs personalized measurements on the user; Step 002: Based on the measurement data in step 001, personalized compensation is performed on the HRTF to obtain a compensated HRTF suitable for the user; Step 003: The compensated HRTF is used to process the simulated sound source data; Step 004: The processed sound signal is played through the left and right speakers of the sound playback device. Thus, personalized compensation is achieved when the user uses the sound playback device.

[0110] In this embodiment, the sound source direction virtualization method of the present invention obtains a first occlusion transmission function obtained by measuring a user and a second occlusion transmission function obtained by measuring a simulated human body. The first occlusion transmission function is the transmission function between the user's left and right ears, and the second occlusion transmission function is the transmission function between the simulated human body's left and right ears. A correlation function is calculated between the second occlusion transmission function and a pre-obtained target transmission function corresponding to the angle of the sound source to be virtualized. The target transmission function is compensated based on the first occlusion transmission function, the second occlusion transmission function, and the correlation function to obtain a personalized transmission function suitable for the user. This personalized transmission function is used to virtually project the sound source signal to be played. The auditory effect corresponding to the angle of the virtual sound source is determined; the ratio between the first occlusion transfer function between the user's left and right ears and the second occlusion transfer function between the simulated human's left and right ears is calculated, and this ratio is determined as the occlusion transfer change rate between the user and the simulated human. The occlusion transfer change rate is multiplied by the relevant function to obtain a calibration function for compensating the target transfer function. The target transfer function is then compensated based on the calibration function to obtain the personalized transfer function corresponding to the user. The pre-acquired audio source signal to be played is subjected to frequency domain filtering based on the compensated personalized transfer function to obtain the filtered audio signal to be played. The audio signal to be played is then played through the speaker on the sound playback device.

[0111] Thus, this embodiment of the invention obtains a first occlusion transmission function between the user's ears and a second occlusion transmission function between the simulated human ears by measuring the user and the simulated human ears respectively. In addition, it calculates the correlation function between the second occlusion transmission function and the target transmission function corresponding to the pre-acquired virtual sound source angle. Finally, it uses the first occlusion transmission function, the second occlusion transmission function and the correlation function to perform personalized compensation on the target transmission function, thereby obtaining a personalized transmission function that can virtualize the auditory effect corresponding to the virtual sound source angle of the sound source signal to be played. In other words, by using the personalized transmission function after personalized compensation to virtualize the sound source direction, the adaptability between the sound playback device and the user and the accuracy of virtual sound source direction are improved.

[0112] Furthermore, based on the first embodiment of the sound source direction virtual method of the present invention described above, a second embodiment of the sound source direction virtual method of the present invention is proposed.

[0113] In this embodiment, the sound playback device includes a first speaker, a first microphone, and a second microphone. The first speaker and the first microphone are installed at a first position on the sound playback device, and the second microphone is installed at a second position on the sound playback device. The first position and the second position are either the left ear position or the right ear position of the sound playback device. Step S10 includes:

[0114] Step S101: While measuring the user, the preset audio signal played by the first speaker is recorded through the first microphone and the second microphone to obtain the first sound signal recorded by the first microphone and the second sound signal recorded by the second microphone.

[0115] When measuring a user, a preset audio signal played by a first speaker is recorded through a first microphone and a second microphone to obtain the sound signal recorded by the first microphone (hereinafter referred to as the first sound signal for distinction) and the sound signal recorded by the second microphone (hereinafter referred to as the second sound signal for distinction).

[0116] Step S102: Calculate the first sound pressure level corresponding to the first sound signal and the second sound pressure level corresponding to the second sound signal;

[0117] Calculate the sound pressure level of the first sound signal (hereinafter referred to as the first sound pressure level for distinction) and the sound pressure level of the second sound signal (hereinafter referred to as the second sound pressure level for distinction).

[0118] In one feasible implementation, the sound pressure level is calculated using the formula Lp = 20log10(P / P0), where Lp represents the sound pressure level, P represents the sound pressure, and P0 represents the reference sound pressure, which is a preset constant. The first sound pressure level P1 of the first sound signal and the second sound pressure level P2 of the second sound signal are then calculated.

[0119] Step S103: The ratio of the second sound pressure level to the first sound pressure level is used as the first occlusion transmission function corresponding to the user.

[0120] The ratio of the second sound pressure level to the first sound pressure level is used as the first obstruction transmission function corresponding to the user.

[0121] In one feasible implementation, the aforementioned first occlusion transmission function includes a left ear occlusion transmission function U_r2l and a right ear occlusion transmission function U_l2r. A preset audio signal is played through the right speaker at the right ear position of the sound playback device, and the sound signal is recorded through the left microphone at the left ear position and the right microphone at the right ear position of the sound playback device. The left sound signal recorded by the left microphone and the right sound signal recorded by the right microphone are obtained. Then, the left sound pressure level of the left sound signal and the right sound pressure level of the right sound signal are calculated respectively. The ratio of the left sound pressure level to the right sound pressure level is used as the left ear occlusion transmission function U_r2l for the sound signal from the user's right ear position to the left ear position. Similarly, the right ear occlusion transmission function U_l2r for the sound signal from the user's left ear position to the right ear position can be determined.

[0122] Furthermore, in a feasible embodiment, step S10 above further includes:

[0123] Step S104: While measuring the simulated human body, the preset audio signal played by the first speaker is recorded through the first microphone and the second microphone to obtain the third sound signal recorded by the first microphone and the fourth sound signal recorded by the second microphone.

[0124] When measuring a simulated human body, a preset audio signal played by a first speaker is recorded through a first microphone and a second microphone to obtain the sound signal recorded by the first microphone (hereinafter referred to as the third sound signal for distinction) and the sound signal recorded by the second microphone (hereinafter referred to as the fourth sound signal for distinction).

[0125] Step S105: Calculate the third sound pressure level corresponding to the third sound signal and the fourth sound pressure level corresponding to the fourth sound signal;

[0126] Calculate the sound pressure level of the third sound signal (hereinafter referred to as the third sound pressure level for distinction) and the sound pressure level of the fourth sound signal (hereinafter referred to as the fourth sound pressure level for distinction).

[0127] In one feasible implementation, the third sound pressure level P3 of the third sound signal and the second sound pressure level P4 of the fourth sound signal are calculated according to the formula for calculating sound pressure level.

[0128] Step S106: The ratio of the fourth sound pressure level to the third sound pressure level is used as the second occlusion transmission function corresponding to the simulated human body.

[0129] The ratio of the fourth sound pressure level to the third sound pressure level is used as the second obstruction transmission function corresponding to the user.

[0130] In one feasible implementation, the aforementioned second occlusion transfer function includes a left ear occlusion transfer function H_r2l and a right ear occlusion transfer function H_l2r. A preset audio signal is played through the right speaker at the right ear position of the sound playback device, and the sound signal is recorded through the left microphone at the left ear position and the right microphone at the right ear position of the sound playback device. The left sound signal recorded by the left microphone and the right sound signal recorded by the right microphone are obtained. Then, the left sound pressure level of the left sound signal and the right sound pressure level of the right sound signal are calculated respectively. The ratio of the left sound pressure level to the right sound pressure level is used as the left ear occlusion transfer function H_r2l for simulating the sound signal from the right ear position to the left ear position of the human body. Similarly, the right ear occlusion transfer function H_l2r for simulating the sound signal from the left ear position to the right ear position of the human body can be determined.

[0131] In this embodiment, when measuring a user, the virtual sound source direction method of the present invention records a preset audio signal played by a first speaker through a first microphone and a second microphone to obtain a first sound signal recorded by the first microphone and a second sound signal recorded by the second microphone. The method then calculates the first sound pressure level of the first sound signal and the second sound pressure level of the second sound signal, and uses the ratio of the second sound pressure level to the first sound pressure level as the first occlusion transmission function corresponding to the user. When measuring a simulated human body, the method records a preset audio signal played by a first speaker through a first microphone and a second microphone to obtain a third sound signal recorded by the first microphone and a fourth sound signal recorded by the second microphone. The method then calculates the third sound pressure level of the third sound signal and the fourth sound pressure level of the fourth sound signal, and uses the ratio of the fourth sound pressure level to the third sound pressure level as the second occlusion transmission function corresponding to the user.

[0132] Thus, the present invention utilizes the accessories provided with the sound playback device to perform personalized measurements on the user and the simulated human body, obtaining personalized feature data to measure the occlusion and reflection effect of the user's or simulated human body's head and torso on the sound signal, and then performs personalized compensation for HRTF based on this personalized feature data.

[0133] Furthermore, based on the first and / or second embodiments of the virtual sound source direction method of the present invention described above, a third embodiment of the virtual sound source direction method of the present invention is proposed.

[0134] In this embodiment, step S20 includes:

[0135] Step S201: Obtain the target transfer function corresponding to the angle of the virtual sound source from the head-related transfer function database based on the angle of the virtual sound source to be virtualized;

[0136] Based on the current angle of the virtual sound source, the target transfer function corresponding to the angle of the virtual sound source is obtained from the database of de novo related transfer functions.

[0137] It should be noted that each transfer function in HFRF is defined as the ratio between the sound pressure p generated by a plane wave at or near a specific point in or near the ear canal in the left ear canal (pL) and the reference (p1). The target transfer functions are H_l and H_r, corresponding to the angle of the virtual sound source, pre-stored in the HRTF database. Specifically, the HRTF database is typically generated in a laboratory using measurements on a simulated human body. The measurement process involves placing a speaker at a certain distance in any direction on the simulated human body and playing specific audio. Recordings are taken from microphones at the left and right ears of the simulated human body, and the transfer functions from the speaker to the left and right microphones are calculated, denoted as H_l and H_r. Measurements are taken by placing a speaker in each direction to obtain H_l and H_r for each angle. Traditionally, the reference selected is the sound pressure Pn generated by a plane wave at the right side of the head, only in the presence of a listener. In the frequency domain, the formula for calculating HRTF is: H_l = PL / Pn, H_r = PR / Pn, where l specifies the left ear and r specifies the right ear, and P is the sound pressure level in the frequency domain.

[0138] Step S202: Extract the fifth sound pressure level corresponding to each of the two ears of the simulated human body from the target transfer function;

[0139] Extract the fifth sound pressure level corresponding to each of the two ears of the simulated human body from the target transfer function;

[0140] In one feasible implementation, the target transfer function includes H_l corresponding to the left ear of the simulated human body and H_r corresponding to the right ear of the simulated human body, and the fifth sound pressure level includes the sound pressure level P5 of the sound signals recorded by the left and right microphones of the sound playback device.

[0141] Step S203: Input the fifth sound pressure level and the fourth sound pressure level into the pre-trained target deep learning model to generate a correlation function between the second occlusion transfer function and the target transfer function.

[0142] The fifth sound pressure level and the fourth sound pressure level in the second occlusion transfer function are input into the pre-trained target deep learning model, which then generates a correlation function between the second occlusion transfer function and the target transfer function based on the input.

[0143] It should be noted that the target deep learning model mentioned above is a trained deep learning model. This invention does not limit the type of deep learning model, which may include, but is not limited to, neural network models.

[0144] In one feasible implementation, since H_r2l and H_l, as well as H_l2r and H_r, all contain the same human body structure occlusion and reflection, the correlation between H_r2l and H_l, and between H_l2r and H_r, is calculated by the target deep learning model to generate correlation functions, which are denoted as Rel_l and Rel_r, respectively.

[0145] Furthermore, in one feasible embodiment, prior to step S203, the method further includes:

[0146] Step B10: Extract the sixth sound pressure level corresponding to the second microphone in the preset occlusion transmission function and the seventh sound pressure level corresponding to the second microphone in the preset transmission function from the preset occlusion transmission function of the known relevant function and the preset transmission function corresponding to the virtual angle to be simulated;

[0147] From the preset occlusion transmission function of the known relevant function and the preset transmission function corresponding to the virtual angle, extract the sound pressure level (hereinafter referred to as the sixth sound pressure level) corresponding to the second microphone in the preset occlusion transmission function and the sound pressure level (hereinafter referred to as the seventh sound pressure level) corresponding to the second microphone in the preset transmission function.

[0148] In one feasible implementation, multiple different simulated human bodies are pre-established, and the occlusion transfer function and transfer function corresponding to each simulated human body are measured. The sixth sound pressure level P6 of the sound signal recorded by the left microphone of the sound playback device is extracted from the occlusion transfer function, and the seventh sound pressure level P7 of the sound signal recorded by the left microphone for each virtual angle is extracted from the preset transfer function. It should be noted that P6 is the sound pressure level of the preset audio signal played by the right speaker recorded by the left microphone, and P7 is the sound pressure level of the preset audio signal played by the sound source at any angle recorded by the left microphone. It can be understood that when the head of the simulated human body is imagined as a circle, the left microphone is located at a point on the circle, denoted as the left position point, and the right speaker is located at a point symmetrical to the left position point, denoted as the right position point. Then P7 is the sound pressure level distributed on the arc between the two points, and P6 is the sound pressure level at the position point of the right speaker. Theoretically, P6 is the smallest value among all sound pressure levels on the circle, and a functional relationship of a circle with the line connecting the left position point and the right position point as its diameter can be determined based on P7 and P6.

[0149] Step B20: Establish a simulated human training set based on the sixth sound pressure level and the seventh sound pressure level;

[0150] A simulated human training set was established based on the sixth and seventh sound pressure levels.

[0151] In one feasible implementation, the P6 and P7 corresponding to multiple different simulated human bodies are stored in the simulated human body training set.

[0152] Step B30: Train the pre-constructed initial deep learning model based on the simulated human training set to obtain the target deep learning model.

[0153] The target deep learning model is obtained by training a pre-built initial deep learning model using a simulated human training set.

[0154] In one feasible implementation, a pre-constructed initial deep learning model is trained using training data from a simulated human training set until the loss function corresponding to the model meets preset requirements, thereby determining the target deep learning model.

[0155] In this embodiment, the method for simulating the direction of a sound source according to the present invention obtains the target transfer function corresponding to the current angle of the sound source to be simulated from a database of head-related transfer functions; extracts the fifth sound pressure level corresponding to each ear of the simulated human body from the target transfer function; extracts the sixth sound pressure level corresponding to the second microphone in the preset occlusion transfer function and the preset transfer function corresponding to the angle to be simulated from the preset occlusion transfer function of the known correlation function and the preset transfer function corresponding to the angle to be simulated; establishes a simulated human body training set based on the sixth and seventh sound pressure levels; trains a pre-constructed initial deep learning model based on the simulated human body training set to obtain a target deep learning model; inputs the fifth sound pressure level and the fourth sound pressure level in the second occlusion transfer function into the pre-trained target deep learning model, and the target deep learning model generates a correlation function between the second occlusion transfer function and the target transfer function according to the input.

[0156] Thus, this invention uses machine learning to determine the occlusion transfer function and the correlation function between the transfer functions corresponding to the simulated human body, and performs personalized compensation for HRTF based on the correlation function.

[0157] Furthermore, embodiments of the present invention also provide a virtual sound source direction device, which is applied to a sound playback device.

[0158] Please refer to Figure 4 , Figure 4 This is a functional module diagram of an embodiment of the virtual sound source direction device of the present invention, as shown below. Figure 4 As shown, the virtual sound source direction device of the present invention includes:

[0159] The measurement module 10 is used to acquire a first occlusion transmission function and a second occlusion transmission function, wherein the first occlusion transmission function is the transmission function between the user's ears obtained by measuring the user, and the second occlusion transmission function is the transmission function between the ears of the simulated human body obtained by measuring the simulated human body.

[0160] The correlation function module 20 is used to calculate the correlation function between the second occlusion transmission function and the target transmission function corresponding to the pre-acquired virtual sound source angle;

[0161] The personalized transmission function module 30 is used to compensate the target transmission function according to the first occlusion transmission function, the second occlusion transmission function and the related function to obtain the personalized transmission function corresponding to the user, wherein the personalized transmission function is used to simulate the auditory effect corresponding to the angle of the sound source signal to be played.

[0162] Furthermore, the sound playback device includes a first speaker, a first microphone, and a second microphone. The first speaker and the first microphone are installed at a first position on the sound playback device, and the second microphone is installed at a second position on the sound playback device. One of the first position and the second position is the left ear position of the sound playback device, and the other is the right ear position of the sound playback device. The measurement module 10 includes:

[0163] The user measurement unit is used to record a preset audio signal played by a first speaker through the first microphone and the second microphone when measuring a user, and to obtain a first sound signal recorded by the first microphone and a second sound signal recorded by the second microphone.

[0164] The first sound pressure level calculation unit is used to calculate the first sound pressure level corresponding to the first sound signal and the second sound pressure level corresponding to the second sound signal.

[0165] The first occlusion transmission function unit is used to take the ratio of the second sound pressure level to the first sound pressure level as the first occlusion transmission function corresponding to the user.

[0166] Furthermore, the measurement module 10 also includes:

[0167] The simulated human body measurement unit is used to record the preset audio signal played by the first speaker through the first microphone and the second microphone when measuring a simulated human body, and to obtain the third sound signal recorded by the first microphone and the fourth sound signal recorded by the second microphone.

[0168] The second sound pressure level calculation unit is used to calculate the third sound pressure level corresponding to the third sound signal and the fourth sound pressure level corresponding to the fourth sound signal.

[0169] The second occlusion transmission function unit is used to take the ratio of the fourth sound pressure level to the third sound pressure level as the second occlusion transmission function corresponding to the simulated human body.

[0170] Furthermore, the personalized transfer function module 30 includes:

[0171] The occlusion transmission rate of change unit is used to take the ratio of the first occlusion transmission function and the second occlusion transmission function as the occlusion transmission rate of change between the user and the simulated human body.

[0172] The calibration function unit is used to multiply the occlusion transmission change rate by the correlation function to obtain the calibration function;

[0173] The personalized transfer function unit is used to compensate the target transfer function based on the calibration function to obtain the personalized transfer function corresponding to the user.

[0174] Furthermore, the virtual sound source direction device of the present invention also includes:

[0175] The frequency domain filtering module is used to perform frequency domain filtering on the audio source signal to be played based on the personalized transfer function to obtain the audio signal to be played.

[0176] The playback module is used to play the audio signal to be played through the speaker on the sound playback device.

[0177] Furthermore, the relevant function module 20 includes:

[0178] The target transfer function acquisition unit is used to acquire the target transfer function corresponding to the angle of the virtual sound source from the head-related transfer function database based on the angle of the virtual sound source to be acquired.

[0179] The fifth sound pressure level extraction unit is used to extract the fifth sound pressure level corresponding to each of the two ears of the simulated human body from the target transfer function;

[0180] The correlation function unit is used to input the fifth sound pressure level and the fourth sound pressure level into a pre-trained target deep learning model to generate a correlation function between the second occlusion transfer function and the target transfer function.

[0181] Furthermore, the virtual sound source direction device of the present invention also includes:

[0182] The sound pressure level extraction module is used to extract the sixth sound pressure level corresponding to the second microphone in the preset occlusion transmission function and the seventh sound pressure level corresponding to the second microphone in the preset transmission function from the preset occlusion transmission function corresponding to the preset correlation function and the preset transmission function corresponding to the virtual angle to be simulated;

[0183] The training set establishment module is used to establish a simulated human training set based on the sixth sound pressure level and the seventh sound pressure level;

[0184] The model training module is used to train the pre-constructed initial deep learning model based on the simulated human training set to obtain the target deep learning model.

[0185] The present invention also provides a computer storage medium storing a sound source direction virtual program, wherein when the sound source direction virtual program is executed by a processor, the sound source direction virtual program method as described in any of the above embodiments is implemented.

[0186] The specific embodiments of the computer storage medium of the present invention are basically the same as the embodiments of the sound source direction virtual program method of the present invention described above, and will not be repeated here.

[0187] The present invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the virtual sound source direction method of the present invention as described in any of the above embodiments, which will not be elaborated here.

[0188] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0189] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0190] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (such as TWS earphones, etc.) to execute the methods described in the various embodiments of the present invention.

[0191] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for simulating the direction of a sound source, characterized in that, The sound source direction virtualization method is applied to a sound playback device, and the sound source direction virtualization method includes the following steps: Obtain a first occlusion transmission function and a second occlusion transmission function, wherein the first occlusion transmission function is the transmission function between the user's ears obtained by measuring the user, and the second occlusion transmission function is the transmission function between the ears of the simulated human body obtained by measuring the simulated human body; Calculate the correlation function between the second occlusion transfer function and the target transfer function corresponding to the pre-acquired virtual sound source angle; The ratio of the first occlusion transfer function to the second occlusion transfer function is used as the occlusion transfer change rate between the user and the simulated human body. The calibration function is obtained by multiplying the rate of change of occlusion transmission by the correlation function; The target transmission function is compensated based on the calibration function to obtain the personalized transmission function corresponding to the user. The personalized transmission function is used to simulate the auditory effect corresponding to the angle of the sound source signal to be played.

2. The virtual sound source direction method as described in claim 1, characterized in that, The sound playback device includes a first speaker, a first microphone, and a second microphone. The first speaker and the first microphone are installed at a first position of the sound playback device, and the second microphone is installed at a second position of the sound playback device. One of the first position and the second position is the left ear position of the sound playback device, and the other is the right ear position of the sound playback device. The steps to obtain the first occlusion transfer function include: When measuring a user, a preset audio signal played by the first speaker is recorded through the first microphone and the second microphone to obtain a first sound signal recorded by the first microphone and a second sound signal recorded by the second microphone; Calculate the first sound pressure level corresponding to the first sound signal and the second sound pressure level corresponding to the second sound signal; The ratio of the second sound pressure level to the first sound pressure level is used as the first occlusion transmission function corresponding to the user.

3. The virtual sound source direction method as described in claim 2, characterized in that, The steps to obtain the second occlusion transfer function include: When measuring a simulated human body, the preset audio signal played by the first speaker is recorded through the first microphone and the second microphone to obtain the third sound signal recorded by the first microphone and the fourth sound signal recorded by the second microphone; Calculate the third sound pressure level corresponding to the third sound signal and the fourth sound pressure level corresponding to the fourth sound signal; The ratio of the fourth sound pressure level to the third sound pressure level is used as the second occlusion transmission function corresponding to the simulated human body.

4. The virtual sound source direction method as described in claim 1, characterized in that, After the step of compensating the target transfer function based on the calibration function to obtain the personalized transfer function corresponding to the user, the method further includes: Based on the personalized transfer function, the audio source signal to be played is subjected to frequency domain filtering to obtain the audio signal to be played. The audio signal to be played is played through the speaker on the sound playback device.

5. The virtual sound source direction method as described in claim 3, characterized in that, The step of calculating the correlation function between the second occlusion transfer function and the target transfer function corresponding to the pre-acquired virtual sound source angle includes: Based on the angle of the virtual sound source, the target transfer function corresponding to the angle of the virtual sound source is obtained from the de novo related transfer function database; Extract the fifth sound pressure level corresponding to each of the two ears of the simulated human body from the target transfer function; The fifth sound pressure level and the fourth sound pressure level are input into a pre-trained target deep learning model to generate a correlation function between the second occlusion transfer function and the target transfer function.

6. The virtual sound source direction method as described in claim 5, characterized in that, Before the step of inputting the fifth sound pressure level and the fourth sound pressure level into the pre-trained target deep learning model to generate the correlation function between the second occlusion transfer function and the target transfer function, the method further includes: Extract the sixth sound pressure level corresponding to the second microphone in the preset occlusion transmission function and the seventh sound pressure level corresponding to the second microphone in the preset transmission function from the preset occlusion transmission function and the preset transmission function corresponding to the virtual angle; A simulated human training set is established based on the sixth and seventh sound pressure levels; The target deep learning model is obtained by training the pre-constructed initial deep learning model based on the simulated human training set.

7. A virtual device for the direction of a sound source, characterized in that, The sound source direction virtual device is applied to a sound playback device, and the sound source direction virtual device includes: The measurement module is used to acquire a first occlusion transmission function and a second occlusion transmission function, wherein the first occlusion transmission function is the transmission function between the user's ears obtained by measuring the user, and the second occlusion transmission function is the transmission function between the ears of the simulated human body obtained by measuring the simulated human body. The correlation function module is used to calculate the correlation function between the second occlusion transmission function and the target transmission function corresponding to the pre-acquired virtual sound source angle; A personalized transmission function module is used to take the ratio of the first occlusion transmission function and the second occlusion transmission function as the occlusion transmission change rate between the user and the simulated human body; multiply the occlusion transmission change rate by the relevant function to obtain a calibration function; and compensate the target transmission function based on the calibration function to obtain the personalized transmission function corresponding to the user, wherein the personalized transmission function is used to simulate the auditory effect corresponding to the angle of the sound source signal to be played.

8. A terminal device, characterized in that, The terminal device includes: a memory, a processor, and a sound source direction virtual program stored in the memory and executable on the processor. When the sound source direction virtual program is executed by the processor, it implements the steps of the sound source direction virtual method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a virtual program for the direction of a sound source, which, when executed by a processor, implements the steps of the virtual method for the direction of a sound source as described in any one of claims 1 to 6.

Images