Real-time sound field testing method, device, equipment and storage medium

Through the combination of micro acoustic sensor network and deep learning model, real-time monitoring and control of the sound field is achieved, the accuracy of sound field data acquisition and prediction in the prior art is solved, and the efficiency and stability of sound field management are improved.

CN118149960BActive Publication Date: 2024-11-26SHENZHEN S TRACK SCIENCE TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410007701.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-01-03
Publication Date
2024-11-26
Estimated Expiration
2044-01-03

AI Technical Summary

Technical Problem

Existing sound field monitoring and control technologies have limitations in acquiring accurate data, making accurate predictions, and implementing effective controls. They fail to make full use of advanced data analysis and machine learning technologies, resulting in a decrease in the accuracy and reliability of analysis.

Method used

Acquire the acoustic field environment parameters and acoustic field energy data through a preset micro acoustic sensor network, use a volume analyzer for real-time detection, combine deep learning models for sound field state analysis and persistence prediction, and design adaptive control strategies to achieve real-time monitoring of the sound field and control.

Benefits of technology

It improves the accuracy of sound field monitoring and the effectiveness of control, ensures real-time and accurate analysis and prediction of sound field states, promptly detects abnormal changes and takes countermeasures, improves the efficiency of sound field management and reduces the problems and losses caused by abnormalities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118149960B_ABST
    Figure CN118149960B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of data processing, and discloses a real-time sound field testing method, device, equipment and storage medium. The real-time sound field testing method comprises: obtaining the environmental parameters of the sound field through a preset micro-acoustic sensor network; performing real-time detection of the sound field energy of the sound field through a preset volume analyzer to obtain sound field energy data; performing sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information; the present invention greatly improves the control accuracy and efficiency of the sound field through real-time and accurate data acquisition, in-depth state analysis, persistence prediction, stability calculation, and efficient abnormal alarm and adaptive control.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a real-time sound field testing method, device, equipment and storage medium. Background Art

[0002] The sound field is the sound distribution area in a specific environment. Real-time monitoring and precise control of the sound field have important applications in fields such as noise control, indoor acoustic optimization, and improvement of the acoustics of concert venues. Traditional sound field monitoring and control are mainly performed manually and rely on fixed algorithms, which have certain shortcomings when dealing with complex environmental parameters and sound field energy information.

[0003] Existing sound field monitoring systems are not sophisticated enough in acquiring environmental parameters and sound field energy data, which limits the accuracy of analysis. Secondly, these systems usually rely only on experience and fixed rules when predicting the persistence of the sound field and measuring the stability of the condition, and fail to make full use of advanced data analysis and machine learning technology, resulting in a decrease in the accuracy and reliability of the prediction results.

[0004] Therefore, the existing sound field monitoring and control technology has obvious limitations in obtaining accurate data, making accurate predictions, and implementing effective control, and needs to be improved and perfected. Summary of the invention

[0005] The present invention provides a real-time sound field testing method, device, equipment and storage medium, which are used to solve the technical problem of how to improve the accuracy of real-time monitoring of the sound field and the effectiveness of control.

[0006] A first aspect of the present invention provides a real-time sound field testing method, the real-time sound field testing method comprising:

[0007] Acquiring environmental parameters of the sound field through a preset micro-acoustic sensor network; wherein the micro-acoustic sensor network includes a plurality of environmental sensing sensors and a volume sensor;

[0008] Performing real-time detection of the sound field energy of the sound field through a preset volume analyzer to obtain sound field energy data;

[0009] Performing sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information;

[0010] Acquiring multidimensional feature data of the sound field, and selecting a corresponding sound field persistence prediction model from a database according to the multidimensional feature data; wherein the sound field persistence prediction model is a trained deep learning model;

[0011] Inputting the sound field state information into a sound field persistence prediction model to predict the persistence data of the sound field, and calculating the stability state of the sound field according to the persistence data of the sound field to obtain the stability state data of the sound field;

[0012] An abnormal warning analysis is performed on the stability status data of the sound field to obtain abnormal warning data, and an adaptive control strategy is designed for the sound field based on the abnormal warning data to obtain a target control strategy for the sound field.

[0013] Optionally, in a first implementation of the first aspect of the present invention, after the step of performing real-time detection of sound field energy on the sound field by a preset volume analyzer to obtain sound field energy data, the method further comprises:

[0014] The sound field energy is monitored in real time through the volume analyzer and continuously compared with the set sound field energy threshold to evaluate the energy status of the current sound field;

[0015] When it is monitored that the energy state of the current sound field drops to or is lower than the set sound field energy threshold, the real-time adjustment module is activated to adjust the energy state of the sound field to the set standard range;

[0016] Monitor environmental parameters in real time, and compare the monitored environmental parameters with the preset range to obtain environmental parameters that deviate from the preset range;

[0017] A corresponding control strategy is adopted to adaptively control the acquired environmental parameters that deviate from the preset range.

[0018] Optionally, in a second implementation of the first aspect of the present invention, the training process of the sound field persistence prediction model includes:

[0019] Acquire a first data sample and a second data sample in a sound field data sample set; wherein the first data sample represents the environmental parameter characteristics of the sound field identified in real time, and the second data sample represents the screened historical change record characteristics of the sound field;

[0020] The first data sample and the second data sample are converted and integrated through a preset sound field interaction model to obtain first data; wherein the sound field interaction model is obtained based on a trained recursive neural network model;

[0021] Performing feature analysis on the first data through an initial sound field persistence prediction model, outputting a corresponding acoustic characteristic region, and determining a corresponding sound field state input vector based on a value of the acoustic characteristic region;

[0022] Calculate the classification loss function value of the sound field persistence prediction model according to the acoustic characteristic area and the specific acoustic property area;

[0023] Based on the similarity score between the sound field state input vector and the sound field optimization state simulation vector, and the difference score between the sound field state input vector and the sound field abnormal state simulation vector, the difference with the preset contrast loss function is calculated to obtain the contrast loss function value of the optimization state score and the abnormal state score; wherein the sound field optimization state simulation vector is an ideal state sample of the sound field state input vector, and the sound field abnormal state simulation vector is an abnormal state sample of the sound field state input vector;

[0024] The trained sound field persistence prediction model is obtained by combining the classification loss function value, the contrast loss function value of the optimization state score, and the contrast loss function value of the abnormal state score and adjusting the model parameters of the sound field persistence prediction model.

[0025] Optionally, in a third implementation of the first aspect of the present invention, after the step of obtaining the target control strategy of the sound field, the method further comprises:

[0026] Acquire the sound field state information in the sound field target control strategy, and extract feature information of the sound field state information, wherein the feature information includes spectrum features of the sound field signal; wherein the spectrum features include pitch features, intensity features, timbre features, duration features, and sound field layout features;

[0027] Based on the pitch feature, generating a corresponding first phoneme using a preset first parsing rule;

[0028] Based on the sound intensity feature, generating a corresponding second phoneme using a preset second parsing rule;

[0029] Based on the timbre feature, generating a corresponding third phoneme using a preset third parsing rule;

[0030] According to the duration feature, using a preset fourth parsing rule, a corresponding fourth phoneme is generated;

[0031] Based on the sound field layout features, determining the phoneme combination rules; wherein the database stores the corresponding relationship between the sound field layout features and the phoneme combination rules;

[0032] Combining the various phonemes based on a determined phoneme combination rule to obtain a target phoneme element combination;

[0033] According to a preset compression coding algorithm, the obtained target phoneme element combination is compressed and encoded to obtain a target binary byte sequence, and the target binary byte sequence is stored in a distributed storage system.

[0034] Optionally, in a fourth implementation of the first aspect of the present invention, the environmental parameters of the sound field include at least: sound pressure, sound speed, background noise, humidity, temperature, and atmospheric pressure.

[0035] A second aspect of the present invention provides a real-time sound field testing device, the real-time sound field testing device comprising:

[0036] An acquisition module, used to acquire the environmental parameters of the sound field through a preset micro-acoustic sensor network; wherein the micro-acoustic sensor network includes a plurality of environmental sensing sensors and a volume sensor;

[0037] A detection module, used for performing real-time detection of sound field energy on the sound field through a preset volume analyzer to obtain sound field energy data;

[0038] An analysis module, used for performing sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information;

[0039] A selection module, used to obtain multi-dimensional feature data of the sound field, and select a corresponding sound field persistence prediction model from a database according to the multi-dimensional feature data; wherein the sound field persistence prediction model is a trained deep learning model;

[0040] A prediction module, used for inputting the sound field state information into a sound field persistence prediction model, predicting and obtaining persistence data of the sound field, and calculating the stability state of the sound field according to the persistence data of the sound field to obtain stability state data of the sound field;

[0041] The design module is used to perform abnormal warning analysis on the stability status data of the sound field to obtain abnormal warning data, and based on the abnormal warning data, perform adaptive control strategy design on the sound field to obtain a target control strategy for the sound field.

[0042] The third aspect of the present invention provides a real-time sound field testing device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the real-time sound field testing device executes the above-mentioned real-time sound field testing method.

[0043] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned real-time sound field testing method.

[0044] The technical solution provided by the present invention has the following beneficial effects: the present invention provides a real-time sound field testing method, device, equipment and storage medium, which obtains the environmental parameters of the sound field through a preset micro-acoustic sensor network; performs real-time detection of the sound field energy of the sound field through a preset volume analyzer to obtain sound field energy data; performs sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information; obtains multi-dimensional feature data of the sound field, and selects a corresponding sound field persistence prediction model from a database according to the multi-dimensional feature data; inputs the sound field state information into the sound field persistence prediction model to predict the persistence data of the sound field, calculates the stability status of the sound field according to the persistence data of the sound field, and obtains the stability status data of the sound field; performs abnormal warning analysis on the stability status data of the sound field to obtain abnormal warning data, and designs an adaptive control strategy for the sound field based on the abnormal warning data to obtain a target control strategy for the sound field. The present invention uses a preset micro-acoustic sensor network and a volume analyzer, and this system can obtain the environmental parameters and energy data of the sound field in real time and accurately, ensuring the accuracy of subsequent analysis and prediction. Based on the collected environmental parameters and sound field energy data, this system can perform in-depth sound field state analysis, obtain the state information of the sound field in time, and help understand the current state and possible changes of the sound field. In addition, the stability data of the sound field is analyzed for abnormal warnings, and an effective adaptive control strategy is designed. When abnormal changes in the sound field state are found, an alarm can be issued in time, and countermeasures can be taken. Not only does it improve the efficiency of sound field management, but it also effectively reduces the problems and losses caused by abnormal sound field states. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A schematic diagram of an embodiment of a real-time sound field testing method in an embodiment of the present invention;

[0046] Figure 2 FIG. 4 is a schematic diagram of an embodiment of a real-time sound field testing device in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] Embodiments of the present invention provide a real-time sound field test method, device, equipment and storage medium. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0048] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 An embodiment of the real-time sound field testing method in the embodiment of the present invention includes:

[0049] Step 101: Acquire the environmental parameters of the sound field through a preset micro-acoustic sensor network; wherein the micro-acoustic sensor network includes a plurality of environmental sensing sensors and a volume sensor;

[0050] It is understandable that the execution subject of the present invention may be a real-time sound field test device, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0051] Specifically, in order to implement step 101, the following is a specific embodiment:

[0052] Deployment of a micro acoustic sensor network: In the acoustic field, a plurality of environmental sensing sensors and volume sensors are pre-installed. These sensors can be selected and placed according to the characteristics of the sound signal and the requirements of the environmental parameters. For example, the environmental sensing sensors may include temperature sensors, humidity sensors, gas sensors, etc., for detecting environmental parameters such as temperature, humidity, and gas composition of the acoustic field. The volume sensor can measure the volume level of the acoustic field.

[0053] Environmental parameter collection: Use a network of micro-acoustic sensors to collect environmental parameters in the sound field. The sensors monitor and collect the values ​​of environmental parameters in real time based on their specific measurement principles and working methods. For example, environmental sensing sensors can sense physical quantities such as temperature, humidity, and gas concentration and convert the measurement results into digital signals. Volume sensors can measure the sound level in the sound field and convert it into corresponding electrical signals.

[0054] Signal processing and data extraction: Data is extracted and analyzed from the raw signals collected by the sensors using signal processing algorithms and techniques. Specific processing methods may include filtering, amplification, sampling, and conversion operations to obtain accurate and reliable environmental parameter data. For example, the data of environmental sensing sensors is filtered and corrected to eliminate noise and improve measurement accuracy. The signal of the volume sensor is amplified and linearized to achieve a wider range of measurements and accuracy.

[0055] Step 102: Real-time detection of sound field energy is performed on the sound field by a preset volume analyzer to obtain sound field energy data;

[0056] Specifically, in order to implement step 102, the following is a specific embodiment:

[0057] Pre-installed volume analyzer: A volume analyzer is pre-installed in the sound field. A volume analyzer is a professional audio measurement and analysis device that can be used to detect and analyze characteristics such as the volume level and energy of a sound field. The preset volume analyzer can be a digital audio processor (DSP), a spectrum analyzer, or a sound level meter.

[0058] Real-time detection and acquisition: The sound field is detected and acquired in real time through a preset volume analyzer. The volume analyzer can obtain the audio signal in the sound field by using a microphone or other signal input device, sample and calculate parameters such as volume level and sound field energy. The volume analyzer can achieve sound field energy measurement and analysis with different accuracy and resolution according to the selected measurement mode and parameters.

[0059] Data processing and analysis: Use professional audio processing and data analysis technology to process and analyze the collected sound field energy data. For the output data of the volume analyzer, digital signal processing algorithms can be applied to filter, amplify, smooth or denoise operations to obtain clear and accurate sound field energy data for further analysis and processing.

[0060] Real-time feedback and control: Real-time sound field energy data is used to achieve feedback and control of the sound synthesis process. For example, the volume and intensity of the sound can be adjusted according to the sound field energy, or the speed and rhythm of speech synthesis can be adjusted to achieve a more accurate and personalized sound effect. At the same time, sound warnings and safety controls can also be performed according to changes in sound field energy to ensure safety and stability in the sound field.

[0061] Step 103: performing sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information;

[0062] Specifically, in order to implement step 103, the following is a specific embodiment:

[0063] Data integration and preprocessing: Integrate the environmental parameters and energy data of the sound field, and preprocess and clean the data. The integration of environmental parameters and energy data of the sound field can be carried out through a database system, such as using data management technologies such as database management system (DBMS) or relational database management system (RDBMS). Preprocessing and cleaning include filtering, deduplication, completion or correction of data to ensure the accuracy and reliability of the data.

[0064] Establishing a sound field state model: Establishing a state model of the sound field based on the integrated and preprocessed data. The state model can be modeled based on techniques such as statistics, machine learning, or deep learning. For example, a classification and recognition model of the sound field state can be established based on environmental parameters of the sound field and sound field energy data using methods such as ridge regression analysis, support vector machine (SVM), or neural network.

[0065] Sound field state analysis and classification: Use the established state model to analyze and classify the state of the sound field. According to the environmental parameters of the sound field and the sound field energy data, the sound field is divided into different states, such as quiet state, noisy state, high temperature state or low temperature state. At the same time, the sound field state can also be identified and classified, such as classifying a multi-person chat sound field, a single-person speaking sound field or a reverberant sound field.

[0066] Sound field state information extraction: Extract the sound field state information based on the state model of the sound field and the sound field state analysis results. The sound field state information may include parameters such as temperature, humidity, gas composition, sound intensity, sound frequency, etc. of the sound field, as well as the state label and classification of the sound field.

[0067] Step 104: acquiring multidimensional feature data of the sound field, and selecting a corresponding sound field persistence prediction model from a database according to the multidimensional feature data; wherein the sound field persistence prediction model is a trained deep learning model;

[0068] Specifically, in order to implement step 104, the following is a specific embodiment:

[0069] Acquisition of multi-dimensional feature data: Acquisition of multi-dimensional feature data from the sound field. The multi-dimensional feature data may include feature parameters of multiple dimensions such as sound frequency, volume level, speech speed, etc. The specific implementation method may include collecting sound signals through a microphone and extracting sound features using a professional digital signal processing algorithm, such as using Fourier transform to obtain frequency features, using an audio compression algorithm to obtain volume features, etc.

[0070] Sound field persistence prediction models in database: Establish and maintain a database containing multiple trained deep learning models for sound field persistence prediction. Deep learning models can be trained using techniques such as neural networks to learn and predict the persistence characteristics of sound fields in different environments. Each deep learning model corresponds to a different sound field and can be selected and matched according to the characteristics of the sound field.

[0071] Multi-dimensional feature data matching: According to the acquired multi-dimensional feature data, the corresponding sound field persistence prediction model is selected from the database. By inputting the multi-dimensional feature data into multiple deep learning models in the database, the prediction ability of the model is used to discriminate and match the persistence characteristics of the sound field. The model that best matches the input feature data is selected as the sound field persistence prediction model.

[0072] Application of sound field persistence prediction model: Apply the selected sound field persistence prediction model to the sound field persistence prediction task. Use the pre-trained deep learning model to predict and estimate the persistence characteristics of the sound field. The prediction results can include information such as sound attenuation, echo time, noise characteristics, etc., which can be used for subsequent sound synthesis and sound field environment adjustment.

[0073] Step 105: input the sound field state information into a sound field persistence prediction model to predict the persistence data of the sound field, and calculate the stability status of the sound field according to the persistence data of the sound field to obtain the stability status data of the sound field;

[0074] Specifically, in order to implement step 105, the following is a specific embodiment:

[0075] Sound field state information input: The sound field state information is used as input to the selected sound field persistence prediction model. The sound field state information may include environmental parameters of the sound field, sound field energy data, and classification and recognition results of the sound field state. The input process may pass the sound field state information to the sound field persistence prediction model for processing through a data transmission interface.

[0076] Sound field persistence prediction: Use the selected sound field persistence prediction model to predict the persistence data of the sound field. The selected deep learning model has been trained to accurately predict the persistence characteristics of the sound field. By inputting the sound field state information, the prediction model will output the corresponding sound field persistence data, including the duration of the sound, attenuation characteristics, echo time, etc.

[0077] Sound field stability calculation: Based on the persistence data of the sound field, the stability of the sound field is calculated. The stability calculation may include the evaluation of the persistence, changeability and echo difference of the sound field. Based on the predicted sound field persistence data, the stability index of the sound field can be calculated, such as the average duration of the sound field, the change of the noise level, etc.

[0078] Acoustic field stability status data acquisition: Acoustic field stability status data is acquired based on the acoustic field stability measurement results. The stability status data may include the continuity index, stability score, sound attenuation trend, etc. of the acoustic field. These data may be further used in applications such as sound synthesis, acoustic field environment adjustment, and acoustic processing to achieve precise control and optimization of the acoustic field.

[0079] Step 106: Perform abnormal warning analysis on the stability status data of the sound field to obtain abnormal warning data, and perform adaptive control strategy design on the sound field based on the abnormal warning data to obtain a target control strategy for the sound field.

[0080] Specifically, in order to implement step 106, the following is a specific embodiment:

[0081] Abnormal warning analysis of stability status data: abnormal warning analysis is performed on the stability status data of the sound field. Abnormal warning analysis can be achieved through statistical methods, machine learning algorithms and other means. First, by establishing a baseline model of a normal sound field, the stability status data is compared and analyzed to detect whether there are abnormal conditions. For example, statistical indicators such as mean value, variance, etc. are used to analyze the stability status data of the sound field. When certain indicators exceed the preset threshold, an abnormal warning signal will be issued.

[0082] Abnormal warning data acquisition: According to the results of abnormal warning analysis, abnormal warning data is obtained. Abnormal warning data may include abnormal identification, abnormal type, abnormal degree and other information. According to the results of the analysis, abnormal conditions of sound field stability can be identified, and abnormal warning data can be used as the basis for subsequent control strategy design.

[0083] Adaptive control strategy design: Based on the abnormal warning data, an adaptive control strategy is designed for the sound field. The adaptive control strategy can adopt corresponding adjustment strategies according to different situations of the abnormal warning data to achieve control and adjustment of the sound field. For example, when the abnormality is marked as the sound is too sharp, a related tone adjustment algorithm can be designed according to the abnormality type to make the sound softer. For another example, when the degree of abnormality is high, a more stringent control strategy can be adopted to restore the stability of the sound field.

[0084] Acquisition of sound field target control strategy: Based on the results of the adaptive control strategy design, the target control strategy of the sound field is obtained. The target control strategy can include the adjustment of control parameters, optimization of sound processing algorithms, etc. Based on the abnormal warning data and the adaptive control strategy, the target control strategy of the sound field can be formulated for different abnormal situations to achieve the stability and optimization of the sound field.

[0085] In the embodiment of the present invention, the beneficial effect is: the embodiment of the present invention can achieve accurate prediction and estimation of the persistence characteristics of the sound field by acquiring multi-dimensional feature data and selecting a trained deep learning model from a database. It can provide prediction results that are more in line with the actual sound field characteristics for subsequent sound processing and synthesis, thereby improving the realism and fidelity of the sound synthesis. The embodiment of the present invention can detect abnormal conditions of the sound field by performing abnormal warning analysis on the sound field stability status data, and perform adaptive control strategy design based on the warning data to achieve correction and optimization of the sound field. In this way, the stability, adaptability and quality of the sound synthesis of the sound field can be improved, thereby meeting the needs of different application scenarios for sound field control.

[0086] Another embodiment of the real-time sound field testing method in the embodiment of the present invention includes:

[0087] After the step of performing real-time detection of the sound field energy of the sound field by a preset volume analyzer to obtain the sound field energy data, the method further comprises:

[0088] The sound field energy is monitored in real time through the volume analyzer and continuously compared with the set sound field energy threshold to evaluate the energy status of the current sound field;

[0089] When it is monitored that the energy state of the current sound field drops to or is lower than the set sound field energy threshold, the real-time adjustment module is activated to adjust the energy state of the sound field to the set standard range;

[0090] Monitor environmental parameters in real time, and compare the monitored environmental parameters with the preset range to obtain environmental parameters that deviate from the preset range;

[0091] A corresponding control strategy is adopted to adaptively control the acquired environmental parameters that deviate from the preset range.

[0092] Specifically, in a specific embodiment of the present invention, a method for detecting the sound field energy in real time by a preset volume analyzer and adjusting the sound field energy according to the detection result is described, comprising the following steps:

[0093] Real-time monitoring of sound field energy through volume analyzer:

[0094] Use professional audio processing equipment, such as microphones and sound acquisition cards, to monitor the sound intensity in the sound field in real time.

[0095] Continuously compare with the set sound field energy threshold:

[0096] The real-time monitored sound field energy data is continuously compared with the pre-set sound field energy threshold to evaluate the energy state of the current sound field.

[0097] Start the real-time adjustment module to adjust the energy state of the sound field:

[0098] When it is monitored that the energy state of the current sound field drops to or below the set sound field energy threshold, the real-time adjustment module is triggered, which can adjust the energy state of the sound field in real time to keep it within the set standard range.

[0099] Real-time monitoring of environmental parameters:

[0100] At the same time, the system also monitors environmental parameters related to the sound field in real time, such as noise level, temperature, humidity, etc.

[0101] Compare the monitored environmental parameters with the preset range:

[0102] The monitored environmental parameters are compared with the preset parameter range to determine whether there are environmental parameters that deviate from the preset range.

[0103] Take corresponding control strategies:

[0104] Once it is detected that the environmental parameters deviate from the preset range, the system will adopt corresponding control strategies to maintain the appropriate state of the sound field.

[0105] Adaptive control of environmental parameters that deviate from the preset range:

[0106] This includes a series of measures such as adjusting the gain of sound equipment, noise reduction processing, temperature control, etc. to adapt to different environmental conditions.

[0107] In the embodiments of the present invention, the beneficial effects are as follows: the embodiments of the present invention can realize real-time monitoring and adaptive adjustment of sound field energy, ensure that the sound field maintains an appropriate energy state under various environmental conditions, and improve the sound collection quality and user experience. The embodiments of the present invention are applicable to various practical scenarios, such as conference rooms, audio recording studios, etc.

[0108] Another embodiment of the real-time sound field testing method in the embodiment of the present invention includes:

[0109] The training process of the sound field persistence prediction model includes:

[0110] Acquire a first data sample and a second data sample in a sound field data sample set; wherein the first data sample represents the environmental parameter characteristics of the sound field identified in real time, and the second data sample represents the screened historical change record characteristics of the sound field;

[0111] The first data sample and the second data sample are converted and integrated through a preset sound field interaction model to obtain first data; wherein the sound field interaction model is obtained based on a trained recursive neural network model;

[0112] Performing feature analysis on the first data through an initial sound field persistence prediction model, outputting a corresponding acoustic characteristic region, and determining a corresponding sound field state input vector based on a value of the acoustic characteristic region;

[0113] Calculate the classification loss function value of the sound field persistence prediction model according to the acoustic characteristic area and the specific acoustic property area;

[0114] Based on the similarity score between the sound field state input vector and the sound field optimization state simulation vector, and the difference score between the sound field state input vector and the sound field abnormal state simulation vector, the difference with the preset contrast loss function is calculated to obtain the contrast loss function value of the optimization state score and the abnormal state score; wherein the sound field optimization state simulation vector is an ideal state sample of the sound field state input vector, and the sound field abnormal state simulation vector is an abnormal state sample of the sound field state input vector;

[0115] The trained sound field persistence prediction model is obtained by combining the classification loss function value, the contrast loss function value of the optimization state score, and the contrast loss function value of the abnormal state score and adjusting the model parameters of the sound field persistence prediction model.

[0116] Specifically, in a specific embodiment of the present invention, a method for training a sound field persistence prediction model is described, and the specific process includes the following steps:

[0117] Get the first data sample and the second data sample in the sound field data sample set:

[0118] The first data sample captures the environmental parameter characteristics of the current sound field, such as sound pressure level, spectral characteristics and reverberation time.

[0119] The second data samples represent historical sound field data, where the data may include sound field data previously collected under the same or different environmental conditions.

[0120] The data samples are processed through the preset sound field interaction model:

[0121] The sound field interaction model is obtained based on the trained recurrent neural network (RNN) model. The RNN model is used to process time series problems and is used to analyze and predict sound field data.

[0122] Characterization by sound field persistence prediction model:

[0123] The initial sound field persistence prediction model is used to perform feature analysis on the data obtained from the sound field interaction model, and the sound field is further decomposed into various acoustic characteristic areas, such as pitch, intensity, timbre, etc.

[0124] Calculate the classification loss function value of the sound field persistence prediction model:

[0125] This function calculates the loss function value based on the difference between the true sound field and the predicted sound field.

[0126] Calculate the contrast loss function value of the optimized state score and the contrast loss function value of the abnormal state score:

[0127] The calculation is performed using the similarity score and difference score between the sound field optimization state simulation vector and the sound field abnormal state simulation vector and the difference with the preset contrast loss function.

[0128] Adjust the model parameters of the acoustic field persistence prediction model:

[0129] Based on the above-calculated classification loss function value, the contrast loss function value of the optimization state score, and the contrast loss function value of the abnormal state score, the parameters of the sound field persistence prediction model are adjusted through gradient descent or other optimization algorithms.

[0130] In the embodiment of the present invention, the beneficial effect is that after the above series of training steps, a sound field persistence prediction model that can effectively predict the sound field persistence is finally obtained. This model can adapt to various sound field environments, has a strong ability to identify and predict changes in the sound field, and can be widely used in various acoustic fields, such as indoor acoustic design, sound system layout optimization, etc.

[0131] Another embodiment of the real-time sound field testing method in the embodiment of the present invention includes:

[0132] After the step of obtaining the target control strategy of the sound field, the method further comprises:

[0133] Acquire the sound field state information in the sound field target control strategy, and extract feature information of the sound field state information, wherein the feature information includes spectrum features of the sound field signal; wherein the spectrum features include pitch features, intensity features, timbre features, duration features, and sound field layout features;

[0134] Based on the pitch feature, generating a corresponding first phoneme using a preset first parsing rule;

[0135] Based on the sound intensity feature, generating a corresponding second phoneme using a preset second parsing rule;

[0136] Based on the timbre feature, generating a corresponding third phoneme using a preset third parsing rule;

[0137] According to the duration feature, using a preset fourth parsing rule, a corresponding fourth phoneme is generated;

[0138] Based on the sound field layout features, determining the phoneme combination rules; wherein the database stores the corresponding relationship between the sound field layout features and the phoneme combination rules;

[0139] Combining the various phonemes based on a determined phoneme combination rule to obtain a target phoneme element combination;

[0140] According to a preset compression coding algorithm, the obtained target phoneme element combination is compressed and encoded to obtain a target binary byte sequence, and the target binary byte sequence is stored in a distributed storage system.

[0141] Specifically, in a specific embodiment of the present invention, a process of obtaining a target binary byte sequence by controlling a sound field target strategy and generating a corresponding phoneme element combination is introduced. The specific steps are as follows:

[0142] Get the sound field status information and extract feature information:

[0143] Extracting sound field status information mainly includes the spectrum characteristics of the sound field signal, including pitch characteristics, sound intensity characteristics, timbre characteristics, duration characteristics and sound field layout characteristics.

[0144] Generate the corresponding phonemes:

[0145] Based on the extracted feature information and the preset parsing rules, the corresponding phonemes are generated. Specifically:

[0146] The pitch feature is used to generate a first phoneme through a preset first parsing rule.

[0147] The intensity feature is used to generate a second phoneme through a preset second parsing rule.

[0148] The timbre features are used to generate a third phoneme through a preset third parsing rule.

[0149] The duration feature is used to generate a fourth phoneme through a preset fourth parsing rule.

[0150] Determine the phoneme combination rules:

[0151] Based on the sound field layout features, the phoneme combination rules are determined. The corresponding relationship between the sound field layout features and the phoneme combination rules is stored in the database.

[0152] Combine each phoneme to get the target phoneme element combination:

[0153] Based on the determined phoneme combination rules, the individual phonemes are successfully combined into a target phoneme element combination.

[0154] Encode and store the target phoneme element combination:

[0155] According to a preset compression coding algorithm, the obtained target phoneme element combination is compressed and encoded to obtain a target binary byte sequence.

[0156] Finally, this target binary byte sequence is stored in a distributed storage system to provide data support for subsequent use or analysis.

[0157] In the embodiment of the present invention, the beneficial effect is that the embodiment of the present invention can successfully compress a large amount of sound field data without losing too much original information, and ensure the validity and accuracy of the data, which is convenient for subsequent data management and use.

[0158] Another embodiment of the real-time sound field testing method in the embodiment of the present invention includes: the environmental parameters of the sound field include at least: sound pressure, sound speed, background noise, humidity, temperature, and atmospheric pressure.

[0159] The above describes the real-time sound field testing method in the embodiment of the present invention. The following describes the real-time sound field testing device in the embodiment of the present invention. Figure 2 , an embodiment of the real-time sound field testing device in the embodiment of the present invention includes:

[0160] An acquisition module, used to acquire the environmental parameters of the sound field through a preset micro-acoustic sensor network; wherein the micro-acoustic sensor network includes a plurality of environmental sensing sensors and a volume sensor;

[0161] A detection module, used for performing real-time detection of sound field energy on the sound field through a preset volume analyzer to obtain sound field energy data;

[0162] An analysis module, used for performing sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information;

[0163] A selection module, used to obtain multi-dimensional feature data of the sound field, and select a corresponding sound field persistence prediction model from a database according to the multi-dimensional feature data; wherein the sound field persistence prediction model is a trained deep learning model;

[0164] A prediction module, used for inputting the sound field state information into a sound field persistence prediction model, predicting and obtaining persistence data of the sound field, and calculating the stability state of the sound field according to the persistence data of the sound field to obtain stability state data of the sound field;

[0165] The design module is used to perform abnormal warning analysis on the stability status data of the sound field to obtain abnormal warning data, and based on the abnormal warning data, perform adaptive control strategy design on the sound field to obtain a target control strategy for the sound field.

[0166] The present invention also provides a real-time sound field testing device, which includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the real-time sound field testing method in the above embodiments.

[0167] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the real-time sound field testing method.

[0168] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0169] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0170] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A real-time sound field testing method, characterized in that: The following steps are involved: Acquiring environmental parameters of the sound field through a preset micro-acoustic sensor network; wherein the micro-acoustic sensor network includes a plurality of environmental sensing sensors and a volume sensor; Performing real-time detection of the sound field energy of the sound field through a preset volume analyzer to obtain sound field energy data; Performing sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information; Acquire multidimensional feature data of the sound field, and select a corresponding sound field persistence prediction model from a database based on the multidimensional feature data; wherein the sound field persistence prediction model is a trained deep learning model; the training process of the sound field persistence prediction model includes: acquiring a first data sample and a second data sample in a sound field data sample set; wherein the first data sample represents the environmental parameter characteristics of the sound field identified in real time, and the second data sample represents the screened historical change record characteristics of the sound field; converting and integrating the first data sample and the second data sample through a preset sound field interaction model to obtain first data; wherein the sound field interaction model is obtained based on a trained recursive neural network model; performing feature analysis on the first data through an initial sound field persistence prediction model, outputting a corresponding acoustic characteristic area, and based on the acoustic characteristic area The method comprises the following steps: determining the corresponding sound field state input vector according to the acoustic characteristic area and the specific acoustic attribute area, calculating the classification loss function value of the sound field persistence prediction model according to the acoustic characteristic area and the specific acoustic attribute area; calculating the difference with the preset contrast loss function based on the similarity score between the sound field state input vector and the sound field optimization state simulation vector, and the difference score between the sound field state input vector and the sound field abnormal state simulation vector, and obtaining the contrast loss function value of the optimization state score and the abnormal state score; wherein the sound field optimization state simulation vector is an ideal state sample of the sound field state input vector, and the sound field abnormal state simulation vector is an abnormal state sample of the sound field state input vector; combining the classification loss function value, the contrast loss function value of the optimization state score and the contrast loss function value of the abnormal state score, and adjusting the model parameters of the sound field persistence prediction model to obtain a trained sound field persistence prediction model; Inputting the sound field state information into a sound field persistence prediction model to predict the persistence data of the sound field, and measuring the stability state of the sound field according to the persistence data of the sound field to obtain the stability state data of the sound field; An abnormal warning analysis is performed on the stability status data of the sound field to obtain abnormal warning data, and an adaptive control strategy is designed for the sound field based on the abnormal warning data to obtain a target control strategy for the sound field.

2. The sound field testing method according to claim 1, characterized in that: After the step of performing real-time detection of the sound field energy of the sound field by a preset volume analyzer to obtain the sound field energy data, the method further comprises: The sound field energy is monitored in real time through the volume analyzer and continuously compared with the set sound field energy threshold to evaluate the energy status of the current sound field; When it is monitored that the energy state of the current sound field drops to or is lower than the set sound field energy threshold, the real-time adjustment module is activated to adjust the energy state of the sound field to the set standard range; Monitor environmental parameters in real time, and compare the monitored environmental parameters with the preset range to obtain environmental parameters that deviate from the preset range; A corresponding control strategy is adopted to adaptively control the acquired environmental parameters that deviate from the preset range.

3. The sound field testing method according to claim 1, characterized in that: After the step of obtaining the target control strategy of the sound field, the method further comprises: Acquire the sound field state information in the sound field target control strategy, and extract feature information of the sound field state information, wherein the feature information includes spectrum features of the sound field signal; wherein the spectrum features include pitch features, intensity features, timbre features, duration features, and sound field layout features; Based on the pitch feature, generating a corresponding first phoneme using a preset first parsing rule; Based on the sound intensity feature, generating a corresponding second phoneme using a preset second parsing rule; Based on the timbre feature, generating a corresponding third phoneme using a preset third parsing rule; According to the duration feature, using a preset fourth parsing rule, a corresponding fourth phoneme is generated; Based on the sound field layout features, determining the phoneme combination rules; wherein the database stores the corresponding relationship between the sound field layout features and the phoneme combination rules; Combining the various phonemes based on a determined phoneme combination rule to obtain a target phoneme element combination; According to a preset compression coding algorithm, the obtained target phoneme element combination is compressed and encoded to obtain a target binary byte sequence, and the target binary byte sequence is stored in a distributed storage system.

4. The sound field testing method according to claim 1, characterized in that: The environmental parameters of the sound field include at least: sound pressure, sound speed, background noise, humidity, temperature, and atmospheric pressure.

5. A real-time sound field testing device, characterized in that: The real-time sound field testing device comprises: An acquisition module, used to acquire the environmental parameters of the sound field through a preset micro-acoustic sensor network; wherein the micro-acoustic sensor network includes a plurality of environmental sensing sensors and a volume sensor; A detection module, used for performing real-time detection of sound field energy on the sound field through a preset volume analyzer to obtain sound field energy data; An analysis module, used for performing sound field state analysis on the sound field according to the environmental parameters of the sound field and the sound field energy data to obtain sound field state information; A selection module is used to obtain multi-dimensional feature data of the sound field, and select a corresponding sound field persistence prediction model from a database according to the multi-dimensional feature data; wherein the sound field persistence prediction model is a trained deep learning model; the training process of the sound field persistence prediction model includes: obtaining a first data sample and a second data sample in a sound field data sample set; wherein the first data sample represents the environmental parameter characteristics of the sound field identified in real time, and the second data sample represents the screened sound field historical change record characteristics; the first data sample and the second data sample are converted and integrated through a preset sound field interaction model to obtain first data; wherein the sound field interaction model is obtained based on a trained recursive neural network model; the first data is feature analyzed through an initial sound field persistence prediction model, the corresponding acoustic characteristic area is output, and based on the acoustic characteristics The value of the area determines the corresponding sound field state input vector; according to the acoustic characteristic area and the specific acoustic attribute area, the classification loss function value of the sound field persistence prediction model is calculated; based on the similarity score between the sound field state input vector and the sound field optimization state simulation vector, and the difference score between the sound field state input vector and the sound field abnormal state simulation vector, the difference with the preset contrast loss function is calculated to obtain the contrast loss function value of the optimization state score and the abnormal state score; wherein the sound field optimization state simulation vector is an ideal state sample of the sound field state input vector, and the sound field abnormal state simulation vector is an abnormal state sample of the sound field state input vector; combining the classification loss function value, the contrast loss function value of the optimization state score and the contrast loss function value of the abnormal state score, by adjusting the model parameters of the sound field persistence prediction model, a trained sound field persistence prediction model is obtained; A prediction module, used for inputting the sound field state information into a sound field persistence prediction model, predicting and obtaining persistence data of the sound field, and calculating the stability state of the sound field according to the persistence data of the sound field to obtain stability state data of the sound field; The design module is used to perform abnormal warning analysis on the stability status data of the sound field to obtain abnormal warning data, and based on the abnormal warning data, perform adaptive control strategy design on the sound field to obtain a target control strategy for the sound field.

6. A real-time sound field testing device, characterized in that: The real-time sound field testing device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the real-time sound field testing device to execute the real-time sound field testing method according to any one of claims 1-4.

7. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the real-time sound field testing method according to any one of claims 1 to 4 is implemented.