VR Spatial Sound Effect Generation Method Based on Scene Dynamic Matching

Abstract

This invention discloses a VR spatial sound effect generation method based on scene dynamic matching, belonging to the field of spatial sound effect technology. It involves real-time acquisition of scene and user data to generate feature vectors, outputting initial sound source parameters through a pre-trained model, and then combining these with user state correction to obtain a sound source parameter package. Based on this, basic sound effects are mapped and synthesized in real time to output matched spatial sound effects. Continuous monitoring of data changes and dynamic triggering of parameter updates and sound effect adjustments achieves real-time dynamic matching between sound effects and the VR scene. This invention solves the problem of sound effects being disconnected from the scene through real-time scene matching and acoustic parameter calculation; it enhances the realism of sound source localization by combining a physical model and binaural cue correction; it eliminates the need for a scene-customized sound effect library by relying on a pre-trained generalization model, reducing development costs; and it uses incremental adjustment to achieve a smooth transition, avoiding auditory abruptness and fatigue, balancing immersion and comfort.

Description

Technical Field

[0001] This invention relates to the field of spatial sound effects technology, specifically a VR spatial sound effect generation method based on scene dynamic matching. Background Technology

[0002] In virtual reality (VR) systems, spatial sound effect generation is a core element in creating an immersive auditory experience. Currently, mainstream methods mainly rely on pre-loading fixed sound effect libraries or real-time adjustments based on user head movements. Essentially, these methods still statically or simply bind the sound source to the scene in terms of coordinates or orientation, failing to establish a dynamic and continuous correlation model based on the physical characteristics of the scene where the sound source is located. This results in the sound propagation effect not evolving realistically and continuously with changes in the scene, ultimately limiting the realism and immersive experience of the sound effect simulation. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a VR spatial sound effect generation method based on scene dynamic matching, which solves the problem that current VR spatial sound effect generation technologies lack dynamic real-time correlation between sound sources and scene physical characteristics, making it difficult to realistically simulate the propagation effect of sound in changing environments.

[0004] To achieve the above objectives, this invention employs the following technical solution: a VR spatial sound effect generation method based on scene dynamic matching, comprising the following specific steps: S1: Real-time acquisition of VR scene feature data and user state data, time alignment of the two types of data, and generation of scene feature vectors based on VR scene feature data; S2: Inputting the scene feature vectors into a pre-trained scene-to-sound source parameter mapping model, calculating initial sound source parameters, and correcting the initial sound source parameters in conjunction with user state data, outputting a corrected sound source parameter package, which includes reverberation parameters, attenuation parameters, and propagation path parameters; S3: Performing parameter mapping and synthesis processing on basic sound effects based on the sound source parameter package, generating spatial sound effects matching the current scene in real time, and outputting them through an audio device; S4: Continuously monitoring changes in VR scene feature data and user state data, triggering updates to the sound source parameter package, thereby changing the spatial sound effects and achieving dynamic matching between sound effects and the scene.

[0005] Furthermore, the pre-training method of the scene-to-sound source parameter mapping model is as follows: First, a deep neural network consisting of an input layer, a hidden layer, and an output layer is constructed, with the layers connected by adjustable weight matrices and bias vectors; then, the VR simulation scene is encoded into historical feature vectors, and the corresponding real sound source parameters are obtained through acoustic simulation to form a training dataset that pairs feature vectors with acoustic parameters; next, end-to-end supervised training is performed using this dataset, with the predicted value calculated through forward propagation, and the mean squared error used as the loss function and the Adam optimizer driving back propagation to iteratively optimize all weight matrices and bias vectors; finally, training is completed when the predicted value error converges to a threshold.

[0006] Furthermore, the initial sound source parameters are calculated as follows: the scene feature vector is input into the scene-to-sound source parameter mapping model, and forward propagation calculation is performed through the weight matrix and bias vector determined within the model. Linear transformation and nonlinear activation are performed layer by layer, and finally the corresponding initial sound source parameters are directly mapped and generated at the output layer.

[0007] Furthermore, the sound source parameter package is obtained by performing real-time physical mapping and binaural cue correction on the reverberation time, attenuation coefficient, and propagation path parameters output by the scene and sound source parameter mapping model using VR scene feature data and user state data, thereby generating the sound source parameter package.

[0008] Furthermore, the real-time physical mapping and binaural cue correction method is as follows: user state data includes the user's real-time position and head movement data; VR scene feature data includes static features and dynamic features. Static features include scene spatial structure parameters and scene material types; dynamic features include dynamic element state information and user interaction data within the scene. Based on the scene spatial structure parameters and scene material types of the static features and the dynamic element state information and user interaction data within the scene of the dynamic features, combined with the user's real-time position, the reverberation time, attenuation coefficient, and propagation path parameters output by the scene and sound source parameter mapping model are corrected in real-time based on a physical acoustic model to achieve physical mapping between sound source propagation attributes and dynamic scenes. Simultaneously, based on head movement data, by calculating the relative orientation of the sound source, the gain difference, time difference, and spectral filtering parameters applied to the left and right channels are dynamically calculated and applied to generate accurate binaural auditory cues.

[0009] Furthermore, the specific method of real-time correction based on the physical acoustic model is as follows: based on the user's real-time location, the state information of dynamic elements in the scene, the scene spatial structure parameters, and the scene material type, the acoustic propagation law is applied to perform real-time calculations, dynamically updating the distance and obstacle attenuation coefficient in the attenuation parameters, the reverberation time and early reflected sound intensity in the reverberation parameters, and the direct sound and reflected sound path ratio in the propagation path parameters, thereby correcting the initial sound source parameters to an accurate set of parameters that conform to the physical characteristics of the current scene.

[0010] Furthermore, the method of applying acoustic propagation laws for real-time calculation is as follows: based on the real-time distance between the user and the sound source, the distance attenuation coefficient is calculated and updated using an inverse square law model; combined with the acoustic absorption properties corresponding to the scene material type, the energy attenuation of multiple reflections of the sound wave in the scene structure is calculated, thereby determining the reverberation time and the intensity of early reflected sound; at the same time, based on the real-time geometric relationship between the sound source, the user, and scene obstacles, the obstruction status of the direct sound path and the main reflection path are determined, and then the weight ratio of each path in the propagation path parameters is updated.

[0011] Furthermore, the method of parameter mapping and synthesis processing of basic sound effects based on the sound source parameter package is as follows: a preset basic sound effect is called as the processing object, and real-time digital signal processing is performed on the called basic sound effect according to the reverberation parameter, attenuation parameter and propagation path parameter in the sound source parameter package; wherein, the reverberation parameter mapping is to add digital reverberation with corresponding time and reflection characteristics to the sound effect, the attenuation parameter mapping is to dynamically scale the sound effect amplitude based on distance and obstacles, and the propagation path parameter mapping is to allocate and adjust the gain, delay and phase difference of the sound effect among multiple output channels.

[0012] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: 1. By constructing a framework for scene feature capture and dynamic parameter calculation, it can respond instantly to changes in the position, state, and spatial structure of dynamic elements in the virtual scene; within milliseconds, it adjusts the reverberation time and attenuation coefficient of sound effects to parameters that conform to the acoustic characteristics of tunnels, solving the problem of switching delay and scene disconnect caused by fixed binding of sound effects in traditional preloading solutions, and ensuring that sound effects are always synchronized with the dynamically evolving visual environment.

[0013] 2. By combining scene and sound source parameter mapping models with real-time correction based on physical acoustic models and binaural cues, the generated sound effects can accurately simulate the propagation, reflection, and attenuation of sound waves in specific material or obstacle environments, allowing users to clearly determine the location and distance of the sound source through hearing and obtain an immersive and realistic experience.

[0014] 3. By training machine learning models through a vast array of diverse scenarios, they can automatically adapt to VR scenes with different materials and spatial structures, without the need for manual pre-production and binding of a large fixed sound effects library for each specific scene, thus achieving a dual improvement in development efficiency and experience quality.

[0015] 4. When dynamically updating sound effects, an incremental adjustment strategy based on change monitoring is adopted instead of regenerating the sound effects at breakpoints. This means that when the user's head turns or scene elements move, the system only performs a slight smooth transition on parameters such as channel balance and volume of the sound effects. According to tests, this method can reduce the frequency of sound effect stuttering or abrupt changes perceived by users during long-term VR experiences by more than an order of magnitude, effectively reducing auditory fatigue and ensuring comfort and smoothness of use while pursuing high realism.

[0016] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0017] Figure 1 This is a flowchart of the VR spatial sound effect generation method of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus.

[0020] Example 1: like Figure 1 As shown, this embodiment of the invention provides a VR spatial sound effect generation method based on scene dynamic matching, including the following specific steps: S1: Real-time acquisition of VR scene feature data and user status data, time alignment of the two types of data, and generation of scene feature vectors based on VR scene feature data; User status data acquisition includes: obtaining the user's real-time position through the GPS and inertial navigation fusion positioning module built into the VR headset; acquiring head motion data, i.e., rotation angle and angular velocity, through the nine-axis sensors built into the headset, such as gyroscopes, accelerometers, and magnetometers; VR scene feature data acquisition includes: scanning the scene through the ToF depth sensor integrated into the VR device to obtain scene spatial structure parameters; identifying scene material types through a multispectral imaging sensor combined with reflectivity analysis algorithms; tracking dynamic elements in the scene, i.e., the state information of moving objects, such as position, speed, and trajectory, through an RGB-D camera combined with a real-time target detection model, such as YOLOv5; and acquiring user interaction data, i.e., operation actions and operation force, through the inertial measurement unit of the VR controller. Time alignment is achieved by stamping all sensor data streams with a unified timestamp based on a hardware clock synchronization protocol, such as PTP, ensuring the time consistency of the data. The core principle of generating scene feature vectors is: by standardizing and numerically encoding the collected scene spatial structure parameters, scene material type, position, speed, trajectory operation actions and operation force through feature encoding, and then concatenating and reducing the dimensionality of all the independently encoded values ​​according to the preset dimensional order to finally form a dense numerical vector of fixed dimensions. This vector can comprehensively and structurally represent the static attributes and dynamic state of the VR scene at the current moment, providing high-dimensional feature inputs that can be directly calculated for subsequent machine learning models. S2: Input the scene feature vector into the pre-trained scene and sound source parameter mapping model to calculate the initial sound source parameters, and correct the initial sound source parameters in combination with user state data, and output the corrected sound source parameter package, which includes reverberation parameters, attenuation parameters and propagation path parameters. S3: Based on the sound source parameter package, perform parameter mapping and synthesis processing on the basic sound effects to generate spatial sound effects that match the current scene in real time, and output them through audio devices; S4: Continuously monitor changes in VR scene feature data and user state data. This involves performing inter-frame difference calculations on the currently collected VR scene feature data and user state data at fixed intervals, such as every 0.02 seconds. By comparing the data with the corresponding data from the previous moment, the magnitude of the change is quantified. Key monitoring indicators include: the rate of change of scene spatial structure, the change in the user's head rotation angle, and the change in the speed of dynamic elements. Various change thresholds are preset, such as structural changes greater than 5%, head rotation greater than 10°, and speed changes greater than 0.2m / s. When any indicator exceeds the corresponding threshold, it is determined that a significant scene or state change has occurred. This triggers the update of the sound source parameter package, thereby changing the spatial sound effects and achieving dynamic matching between sound effects and the scene.

[0021] Example 2 differs from Example 1 in that: The pre-training method for the scene-to-sound source parameter mapping model is as follows: First, a deep neural network consisting of an input layer, hidden layers, and an output layer is constructed. The layers are connected by adjustable weight matrices and bias vectors, thus building a multilayer perceptron with nonlinear fitting capabilities. By adjusting the weights and biases of the interlayer connections, the complex mapping relationship between scene features and acoustic parameters is learned. The significance lies in establishing a general parameter prediction model that can replace human experience, automatically deriving initial acoustic parameters conforming to physical laws for any given scene features, laying the foundation for dynamic matching. Next, VR simulation scenes are encoded into historical feature vectors, and corresponding real sound source parameters are obtained through acoustic simulation, forming a training dataset that pairs feature vectors with acoustic parameters. This involves using a large number of known VR scene samples to encode the spatial structure, materials, and dynamic elements of each scene into feature vectors. Simultaneously, high-precision offline acoustic simulation software, such as a geometry-based acoustics simulator, is used to calculate the actual reverberation time and attenuation coefficient for this scene, thus forming paired samples. The significance of this is to construct a large-scale training set with broad coverage and accurate data, providing sufficient and reliable learning for the model. The dataset is used to ensure its generalization ability. Then, end-to-end supervised training is performed on this dataset. Forward propagation calculates predicted values, and backpropagation is driven by the Adam optimizer using mean squared error as the loss function. All weight matrices and bias vectors are iteratively optimized. During training, the model predicts sound source parameters based on the input feature vector, calculates the mean squared error between the predicted values ​​and the simulated true values, and uses the Adam optimization algorithm to adjust all weights and biases of the network based on the error gradient. Through multiple iterations, the prediction error is continuously reduced. The significance of this is that, through a data-driven approach, the model automatically learns and internalizes the inherent correlation between scene features and acoustic parameters, ultimately obtaining a high-precision prediction model. Training is completed when the prediction error converges to a threshold. A preset error threshold is set, such as a mean squared error below 0.01. When the model's prediction error on the validation set no longer decreases significantly and remains stable below this threshold, the model is considered sufficiently trained, and iteration stops. This ensures that the trained model has sufficient prediction accuracy and stability, providing a reliable guarantee for subsequent real-time computation.

[0022] The initial sound source parameters are calculated as follows: The scene feature vector is input into the scene-to-sound source parameter mapping model. Using the pre-determined weight matrix and bias vector within the model, forward propagation calculations are performed, with linear transformations and nonlinear activations applied layer by layer. Finally, the corresponding initial sound source parameters are directly mapped and generated at the output layer. Alternatively, the real-time generated scene feature vector is input into a pre-trained static neural network model. Based on its training-determined and unchanging weights and bias parameters, the model performs linear transformations of matrix multiplication and vector addition on the input data, following the order from the input layer to the hidden layer and then to the output layer. This is followed by ReLU nonlinear activation processing to abstract and transform features layer by layer. Ultimately, the neuron values ​​in the output layer directly correspond to the predicted initial acoustic parameters, such as reverberation time or attenuation coefficient. The significance lies in using a model with learned knowledge to achieve rapid inference from arbitrary new scene features to reasonable acoustic parameters. This set of initial parameters reflects the inherent physical acoustic characteristics of the scene, providing a high-precision computational starting point for subsequent real-time correction based on user status.

[0023] The method for obtaining the sound source parameter package is as follows: By using VR scene feature data and user state data, the reverberation time, attenuation coefficient, and propagation path parameters output by the scene and sound source parameter mapping model are physically mapped in real time and corrected with binaural cues to generate a sound source parameter package. In other words, through a two-stage strategy that combines model prediction baseline and real-time state correction, a complete set of parameters is generated that reflects the general acoustic characteristics of the scene and accurately adapts to the current instantaneous user state and dynamic environmental details. This provides a precise data-driven core for achieving highly realistic and personalized spatial sound effect dynamic synthesis.

[0024] The real-time physical mapping and binaural cue correction methods are as follows: User status data includes real-time user location and head movement data. VR scene feature data includes static and dynamic features. Static features include scene spatial structure parameters and scene material types, while dynamic features include the state information of dynamic elements within the scene and user interaction data. Based on the scene spatial structure parameters and scene material types of the static features, and the state information of dynamic elements within the scene and user interaction data of the dynamic features, combined with the user's real-time location, the reverberation time, attenuation coefficient, and propagation path parameters output by the scene-to-sound source parameter mapping model are corrected in real time based on a physical acoustic model, realizing the physical mapping of sound source propagation attributes to the dynamic scene. Simultaneously, based on head movement data, by calculating the relative orientation of the sound source, the gain difference, time difference, and spectral filtering parameters applied to the left and right channels are dynamically calculated and applied to generate accurate binaural auditory cues. This ensures the consistency between the auditory and visual scenes from a physical perspective, which is the foundation for the realism of immersion.

[0025] The specific method for real-time correction based on the physical acoustic model is as follows: Based on the user's real-time location, the state information of dynamic elements in the scene, the scene's spatial structure parameters, and the scene's material type, the acoustic propagation laws are applied for real-time calculations. The distance and obstacle attenuation coefficients in the attenuation parameters, the reverberation time and early reflected sound intensity in the reverberation parameters, and the direct sound and reflected sound path ratio in the propagation path parameters are dynamically updated. This corrects the initial sound source parameters to a precise set of parameters that conforms to the physical characteristics of the current scene. Through lightweight real-time acoustic simulation, the sound effect parameters can continuously and synchronously respond to all dynamic changes in the VR scene, such as user movement, object displacement, and state changes. This solves the core problem of the disconnect between sound effects and scene dynamics in traditional solutions and achieves frame-level synchronization between sound effects and the three-dimensional virtual environment.

[0026] The method for real-time calculation using the laws of acoustic propagation is as follows: Based on the real-time distance between the user and the sound source, the distance attenuation coefficient is calculated and updated using an inverse square law model, achieving a realistic simulation of the physical effect of sound attenuation with distance at extremely low computational cost. Combining the acoustic absorption properties corresponding to the scene material type, the energy attenuation of sound waves after multiple reflections in the scene structure is calculated, thereby determining the reverberation time and the intensity of early reflected sound. Simultaneously, based on the real-time geometric relationship between the sound source, the user, and scene obstacles, the occlusion status of the direct sound path and the main reflection path are determined, and the weight ratio of each path in the propagation path parameters is updated. This achieves real-time modeling of the sound propagation path, enabling the simulation of complex acoustic phenomena such as obstacle occlusion, reflection, and diffraction, greatly improving the completeness and realism of sound propagation simulation in complex scenes.

[0027] The method for parameter mapping and synthesis of basic sound effects based on sound source parameter packages is as follows: The system calls upon preset base sound effects as the processing object and performs real-time digital signal processing on the called base sound effects based on the reverberation parameters, attenuation parameters, and propagation path parameters in the sound source parameter package. Specifically, the reverberation parameter is mapped to add digital reverberation with corresponding time and reflection characteristics to the sound effect, the attenuation parameter is mapped to dynamically scale the sound effect amplitude based on distance and obstacles, and the propagation path parameter is mapped to adjust the gain, delay, and phase difference of the sound effect across multiple output channels. Through multi-channel rendering or binaural rendering technology, the parameters containing orientation information are converted into specific electrical signals that drive the speakers or headphones, and finally, a three-dimensional sound image is accurately reconstructed in the user's auditory perception, completing the final transformation from data parameters to an immersive auditory experience.

[0028] The preferred embodiments disclosed in this invention are merely illustrative examples of feasible implementation methods and are not intended to exhaustively cover all technical details of the invention, nor do they constitute a limitation on the scope of protection of this invention. In practical applications, those skilled in the art can make appropriate adjustments, combinations, or substitutions to the methods or systems described in these embodiments based on specific production conditions, equipment configurations, and process requirements, without departing from the core concept of this invention. For example, the acquisition method, data processing algorithm, control threshold, or specific implementation form of the execution unit can all be reasonably modified according to the actual situation.

[0029] Furthermore, the technical concepts disclosed in this invention have universal extensibility and adaptability. They are not only applicable to the specific scenarios described in the embodiments, but can also be applied in similar technical fields or related industrial processes through analogy, transplantation, or improvement. Any technical solution formed by making logically equivalent substitutions, reasonable adjustments to the order of steps, or recombination of module functions based on the principles, ideas, or framework disclosed in this specification should be considered to fall within the spirit and scope of this invention.

[0030] It should be further clarified that the specific descriptions and drawings in the patent documents are for the purpose of assisting in understanding the present invention only, and their details should not be interpreted as limitations on the claims. The true scope of protection of the present invention should be determined by the content of the claims recorded in the authorized text, and should cover all equivalent technical solutions that comply with the provisions of patent law under these claims. Any implementation method that has the same or similar function and achieves similar effects through reasonable changes in technical means under the guidance of the concept of the present invention falls within the scope of protection sought by the present invention.

[0031] Therefore, the descriptions in this specification are merely illustrative. Any adjustments to the implementation methods, equivalent substitutions of technical features, or further applications based on the concept of this invention, as long as they do not depart from the overall technical idea described in this invention, should be included within the scope of protection of this invention. For example, specific selection and optimization of sensor types, data fusion algorithms, neural network structures, or audio processing engines; reasonable changes to feature encoding methods, parameter correction processes, or dynamic monitoring thresholds; application of the method to augmented reality (AR), mixed reality (MR), or other interactive audiovisual simulation systems; or combination with emerging technologies such as cloud rendering and AIGC (AI Generated Content) to achieve more efficient scene analysis and sound effect generation—all of these, as long as their core remains the same—involving real-time capture of dynamic scene features and mapping and matching with acoustic parameters to achieve adaptive generation and adjustment of spatial sound effects, should be considered within the scope of the technical concept proposed in this invention. We encourage those skilled in the art to innovate and optimize based on a full understanding of the core of this invention, combined with specific practices, to jointly promote the progress and development of related technologies.

Claims

1. A VR spatial sound effect generation method based on scene dynamic matching, characterized in that: The specific steps include the following: S1: Real-time collection of VR scene feature data and user status data, time alignment of the two types of data, and generation of scene feature vectors based on VR scene feature data; S2: Input the scene feature vector into the pre-trained scene and sound source parameter mapping model to calculate the initial sound source parameters, and correct the initial sound source parameters in combination with user state data, and output the corrected sound source parameter package, which includes reverberation parameters, attenuation parameters and propagation path parameters. S3: Based on the sound source parameter package, perform parameter mapping and synthesis processing on the basic sound effects to generate spatial sound effects that match the current scene in real time, and output them through audio devices; S4: Continuously monitor changes in VR scene feature data and user status data, and trigger updates to the sound source parameter package, thereby changing the spatial sound effects and achieving dynamic matching between sound effects and the scene.

2. The VR spatial sound effect generation method based on scene dynamic matching according to claim 1, characterized in that: The pre-training method for the scene-to-sound source parameter mapping model is as follows: First, a deep neural network consisting of an input layer, hidden layers, and an output layer is constructed, with adjustable weight matrices and bias vectors connecting the layers. Then, the VR simulation scene is encoded into historical feature vectors, and the corresponding real sound source parameters are obtained through acoustic simulation, forming a training dataset with paired feature vectors and acoustic parameters. Next, end-to-end supervised training is performed using this dataset. The predicted values ​​are calculated through forward propagation, and backpropagation is driven by the mean squared error loss function and the Adam optimizer to iteratively optimize all weight matrices and bias vectors. Finally, training is completed when the prediction error converges to a threshold.

3. The VR spatial sound effect generation method based on scene dynamic matching according to claim 2, characterized in that: The initial sound source parameters are calculated as follows: The scene feature vector is input into the scene-to-sound source parameter mapping model. Through the weight matrix and bias vector determined within the model, forward propagation calculation is performed, and linear transformation and nonlinear activation are carried out layer by layer. Finally, the corresponding initial sound source parameters are directly mapped and generated at the output layer.

4. The VR spatial sound effect generation method based on scene dynamic matching according to claim 3, characterized in that: The method for obtaining the sound source parameter package is as follows: By using VR scene feature data and user status data, the reverberation time, attenuation coefficient, and propagation path parameters output by the scene and sound source parameter mapping model are physically mapped in real time and corrected with binaural cues to generate a sound source parameter package.

5. The VR spatial sound effect generation method based on scene dynamic matching according to claim 4, characterized in that: The real-time physical mapping and binaural cue correction method is as follows: User status data includes real-time user location and head movement data. VR scene feature data includes static and dynamic features. Static features include scene spatial structure parameters and scene material types, while dynamic features include the status information of dynamic elements within the scene and user interaction data. Based on the scene spatial structure parameters and scene material types of the static features, and the status information of dynamic elements within the scene and user interaction data of the dynamic features, combined with the user's real-time location, the reverberation time, attenuation coefficient, and propagation path parameters output by the scene-to-sound source parameter mapping model are corrected in real time based on a physical acoustic model, realizing the physical mapping of sound source propagation attributes to the dynamic scene. Simultaneously, based on head movement data, by calculating the relative orientation of the sound source, the gain difference, time difference, and spectral filtering parameters applied to the left and right channels are dynamically calculated and applied to generate accurate binaural auditory cues.

6. The VR spatial sound effect generation method based on scene dynamic matching according to claim 5, characterized in that: The specific method for real-time correction based on the physical acoustic model is as follows: Based on the user's real-time location, the status information of dynamic elements in the scene, the scene's spatial structure parameters, and the scene's material type, the acoustic propagation laws are applied to perform real-time calculations, dynamically updating the distance and obstacle attenuation coefficients in the attenuation parameters, the reverberation time and early reflected sound intensity in the reverberation parameters, and the direct sound and reflected sound path ratio in the propagation path parameters, thereby correcting the initial sound source parameters to a precise set of parameters that conform to the physical characteristics of the current scene.

7. The VR spatial sound effect generation method based on scene dynamic matching according to claim 6, characterized in that: The method for real-time calculation using the laws of acoustic propagation is as follows: Based on the real-time distance between the user and the sound source, the distance attenuation coefficient is calculated and updated using an inverse square law model. Combining the acoustic absorption properties corresponding to the scene material type, the energy attenuation of multiple reflections of the sound wave in the scene structure is calculated to determine the reverberation time and the intensity of early reflected sound. At the same time, based on the real-time geometric relationship between the sound source, the user, and scene obstacles, the occlusion status of the direct sound path and the main reflection path are determined, and the weight ratio of each path in the propagation path parameters is updated.

8. The VR spatial sound effect generation method based on scene dynamic matching according to claim 7, characterized in that: The method of parameter mapping and synthesis of basic sound effects based on sound source parameter packages is as follows: The system calls a preset base sound effect as the processing object and performs real-time digital signal processing on the base sound effect according to the reverb parameters, attenuation parameters and propagation path parameters in the sound source parameter package. The reverb parameters are mapped to add digital reverb with corresponding time and reflection characteristics to the sound effect, the attenuation parameters are mapped to dynamically scale the sound effect amplitude based on distance and obstacles, and the propagation path parameters are mapped to adjust the gain, delay and phase difference of the sound effect among multiple output channels.

Images