CN110362789A - A kind of adaptive sound masking system and method based on GPR model - Google Patents
A kind of adaptive sound masking system and method based on GPR model Download PDFInfo
- Publication number
- CN110362789A CN110362789A CN201910656438.8A CN201910656438A CN110362789A CN 110362789 A CN110362789 A CN 110362789A CN 201910656438 A CN201910656438 A CN 201910656438A CN 110362789 A CN110362789 A CN 110362789A
- Authority
- CN
- China
- Prior art keywords
- sound
- masking
- satisfaction
- adaptive
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000000873 masking effect Effects 0.000 title claims abstract description 158
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 41
- 238000000034 method Methods 0.000 title claims abstract description 37
- 241001342895 Chorus Species 0.000 claims abstract description 46
- HAORKNGNJCEJBX-UHFFFAOYSA-N cyprodinil Chemical compound N=1C(C)=CC(C2CC2)=NC=1NC1=CC=CC=C1 HAORKNGNJCEJBX-UHFFFAOYSA-N 0.000 claims abstract description 43
- 239000003795 chemical substances by application Substances 0.000 claims abstract description 33
- 238000012216 screening Methods 0.000 claims abstract description 27
- 230000008569 process Effects 0.000 claims abstract description 12
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 8
- 241000238631 Hexapoda Species 0.000 claims abstract description 7
- 238000013459 approach Methods 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 17
- 230000007613 environmental effect Effects 0.000 claims description 16
- 230000007246 mechanism Effects 0.000 claims description 13
- 230000015654 memory Effects 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 9
- 230000006054 immunological memory Effects 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000000513 principal component analysis Methods 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 30
- 238000013473 artificial intelligence Methods 0.000 abstract description 9
- 230000009467 reduction Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 210000000987 immune system Anatomy 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 230000007233 immunological mechanism Effects 0.000 description 1
- 238000010348 incorporation Methods 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 238000012917 library technology Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 238000009394 selective breeding Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Theoretical Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Soundproofing, Sound Blocking, And Sound Damping (AREA)
Abstract
The present invention relates to a kind of adaptive sound masking system and method based on GPR model, system includes by various intermittent insects tweeting sounds, the masking sound database that the natural phonations such as birds tweeting sound and water flow sound are constituted, the sound satisfaction agent model modeled by Gaussian process regression algorithm and the adaptive masking sound established based on database and agent model screen subsystem, its function is for different middle low frequency ambient noises, based on unified sound satisfaction agent model, optimal masking sound is filtered out in the database, it obtains corresponding group of chorus simultaneously and its is satisfied with angle value, the system reaches good sound masking effect to various Middle and low frequency noise environment.Compared with prior art, artificial intelligence approach is applied to sound masking technology for the first time by the present invention, and construct a complete adaptive sound masking system, improve the feasibility and universality of sound masking technology, and masking sound screening time is saved, is reduced in traditional sound masking technology by the cost manually adjusted.
Description
Technical field
The present invention relates to environmental noise abatement technical field, covered at the sound more particularly, to a kind of adaptive based on GPR model
Cover system and method.
Background technique
With urbanization high speed development, pollution from environmental noise problem is increasingly by public attention.Traditional noise reduction technology is concentrated
It handles in the sound insulation of sound source noise reduction, the sound absorption of route of transmission process and noise recipient's environment, reduces to a certain extent
Urban environment noise sound level improves the living environment of resident.But due to traditional noise reduction measure traffic noise and fixation set
Limitation and Middle and low frequency noise in standby noise abatement are not easy the characteristic to decay, and urban environment noise Chang Yizhong characteristics of low-frequency is
It is main, especially Concentrated city occur sound level it is up to standard but still the problem of disturb residents.The residential quarter innerland along to major urban arterial highway
And noise carries out tracking and monitoring discovery in urban public lawns region, the sound level in such region is horizontal to maintain 55-60dBA mostly
Steady-state level, however the public is in such environment, and satisfaction is still relatively low, and the higher region satisfaction of sound level is then lower.
Recent study discovery, the public not merely depend on sound pressure level, the sound of pleasant to the satisfaction of acoustic environment
Even if sound pressure level is higher, people also can be improved to the satisfaction of acoustic environment.In fact, this is psychologic acoustics in environmental acoustics
Using.Studies have found that the underwater sound (such as fountain, flowing water) and chirm etc. help to improve traffic noise bring worry.
Therefore, researchers start to consider to improve using sound masking technology the public to the satisfaction of acoustic environment locating for it, this method tool
There are good application prospect and the market demand.Currently, the environmental noise abatement technology about sound masking method also rests on starting
In the stage, there is not been reported for the related patents of the sound masking method of system.Currently the only more complete sound masking method is environment
In " a method of based on sound masking improve acoustic environment satisfaction " of research in 2018, major function is needle for research institute
To the individual validity feature of environmental background noise, artificial selection has the natural phonation of identical validity feature, then passes through manual iteration
The effective characteristic value of correspondence for adjusting the natural phonation obtains optimal masking sound.This method does not determine that the unification of each background sound is effective
It is unified dependent on validity feature parameter sound satisfaction model not establish out one for feature, also, for optimal masking sound
Selection and adjustment also rest on stage of man-made chamber.In fact, sound is satisfied since the characteristic parameter for influencing satisfaction is numerous
Spending model is complicated higher-dimension, a nonlinear system, and researchers are difficult to pick out the inherent physical mechanism of the complication system,
Therefore, the maximum difficult point of traditional sound masking technology is that the foundation of sound satisfaction mechanism model.
Current manual's intellectual technology has been widely used in internet, industry, agricultural, traffic etc. every field,
Obtain superior effect.In recent years, the highest attention of researcher is also gradually caused in environmental area.It is led in environmental noise abatement
Domain, researchers begin trying artificial intelligence technology being applied to conventional noise control method one after another.In fact, utilizing artificial intelligence
Big data technology in energy technology can be very good to solve the problems, such as that mechanism characteristic is difficult to obtain in conventional noise technology.Therefore,
Artificial intelligence technology is applied to sound masking method, utilizes database, agent model and immunological memory in artificial intelligence approach
Mechanism establishes the adaptive sound masking system of a raising acoustic environment satisfaction with universality, can be by conventional method one
The sound masking technology being directly difficult to realize is built into a complete feasible sound masking system, promoted sound masking technology feasibility and
Practicability, and avoid the cost of labor of sound masking technology and reduce time consumption.
Summary of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide a kind of based on GPR model
Adaptive sound masking system and method, the system combined data library technology and agent model method, are able to ascend sound masking technology
Feasibility and universality, and reduce the technology implementation artificial and time cost.
Innovative point of the invention is that the validity feature of acoustic environment satisfaction has been determined, acquires enough natural phonation sample buildings
Masking sound database is covered the memory mechanism introducing in immune system using the approximate pervasive sound satisfaction model of agent model
Cover sound screening subsystem.It can be directly based upon satisfaction agent model and search for optimal masking sound in masking sound database,
The memory cell for generating the masking respondent behavior simultaneously enables directly shelter to background sound is repeated system next time.The invention
The feasibility and practicability of sound masking technology are effectively improved, and saves to degree produced by manual iteration's masking sound greatly
Raw a large amount of cost and time consumption.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of adaptive sound masking system based on GPR model, the system act on behalf of mould by masking sound database, sound satisfaction
Type and adaptive masking sound screening subsystem are established, and the masking sound database is piped by a large amount of intermittent insects
Sound, birds tweeting sound, water flow sound and other a variety of natural phonations are constituted, and the sound satisfaction agent model is returned by Gaussian process and calculated
Method modeling is constituted, and the adaptive masking sound screening subsystem is acted on behalf of based on the masking sound database and the sound satisfaction
Model, and introduce immunological memory Mechanism establishing and form.
The present invention also provides a kind of building sides for a kind of adaptive sound masking system based on GPR model
Method, method includes the following steps:
Step 1: acquisition background sound sample and natural phonation sample simultaneously establish masking sound database using the sample data of acquisition；
Step 2: utilizing GPR model regression combination sound satisfaction agent model；
Step 3: constructing adaptive masking sound screening subsystem on masking sound database based on sound satisfaction agent model.
Further, the step 1 include it is following step by step:
Step 11: acquisition different scenes, the environmental background sound sample in season and period are carried on the back by subjective assessment
The essential characteristic of scape noise, and the validity feature parameter of influence acoustic environment satisfaction is obtained using Principal Component Analysis；
Step 12: according to validity feature parameter acquire intermittent insects tweeting sound, birds tweeting sound, water flow sound and other
The data of a variety of natural phonations establish masking sound data using the natural phonation data and its corresponding validity feature parameter value of acquisition
Library.
Further, the validity feature parameter in the step 11 refers to the main feature for influencing background sound satisfaction, they
The warp such as basic physical features A sound level, characteristic frequency sound pressure level, loudness, roughness, acutance, wow flutter and tone degree by noise
Principal component analysis is crossed to obtain.
Further, the step 2 include it is following step by step:
Step 21: uniform sampling approach being taken to acquire enough masking sound samples in masking sound database；
Step 22: combining the environmental background sound sample of acquisition with masking sound sample to obtain a group chorus sample, and obtain group
The validity feature parameter value of chorus；
Step 23: satisfaction subjective assessment test being carried out to all groups of choruses in group chorus sample and is obtained corresponding full
Meaning degree observation；
Step 24: using in group chorus sample all groups of choruses and corresponding satisfaction observation as training sample；
Step 25: consideration group chorus satisfaction agent model, and the agent model is modeled using Gaussian process regression algorithm.
Further, the agent model in the step 24, describes formula are as follows:
In formula, chorus sample is organizedIndicate the d being made of group chorus effective feature volume dimension input variable,Indicate that potential (true) extent function, y () indicate to correspond to group chorus sampleSatisfaction observation function, with
Machine deviation variables ε indicates that true sound is satisfied with angle value and is influenced by Different Individual subjective differences.
Further, the Gaussian process regression algorithm in the step 25, predictor formula are as follows:
In formula,Respectively indicate any combination soundSatisfaction predicted value and prediction variance,Degree of being satisfied with observation vector, K*,mn、KmnAnd K**Equal representative function covariance
Matrix,Indicate deviation covariance matrix, ΙmnIndicate mn grades of unit matrixs.
Further, the step 3 include it is following step by step:
Step 31: identifying the background sound newly to arrive and activate memory cell simultaneously；
Step 32: if being to repeat background sound to new background sound recognition result, in the masking sound of adaptive sound masking system
Directly optimal masking sound is remembered in matching in database；
Step 33: if being completely new background sound to new background sound recognition result, the predictor formula of group chorus satisfaction is utilized,
The group chorus satisfaction value of the background sound Yu masking sound database is calculated one by one, and screening obtains maximum and is satisfied with angle value in the database
Corresponding optimal masking sound, and memory cell is generated simultaneously；
Step 04: exporting the optimal masking sound and control intervention is carried out to this background sound.
Compared with prior art, the invention has the following advantages that
One, based on the sound masking technology of masking effect: the thinking of masking effect in psychologic acoustics is introduced noise by the present invention
Control field proposes a kind of sound masking technology for the purpose of improving crowd's subjective feeling.It is difficult to decrease in ambient sound level, and
When the acoustic environment satisfaction of crowd is lower, do not need to consider further that how to do one's utmost to reduce noise, but by the way that effective masking sound is added
Crowd is promoted to the satisfaction of acoustic environment, consider the masking benefit of psychologic acoustics, from the subjective feeling angle of people, utilized
Sound masking technology improves the public to the satisfaction of acoustic environment locating for it.
Two, the sound masking system with universality: the present invention utilizes the artificial intelligence such as database, agent model and artificial immunity
Energy technology establishes a complete adaptive masking sound system, effectively improves the feasibility and universality of sound masking technology,
Improve the application prospect of sound masking technology.The application of artificial intelligence approach, the sound masking that conventional method is difficult to realize always
Technology is built into a complete feasible sound masking system, wherein the application of database has established number for masking sound automatic screening
According to basis；The application of Gaussian process regression algorithm realizes masking sound system high-precision, efficient screening.
Three, organize chorus satisfaction model: the present invention has unified the validity feature of background sound, and is covered according to validity feature foundation
Cover sound database, further, a group chorus satisfaction model proposed based on validity feature parameter, can the simplification sound of big degree cover
Cover the complexity of system modelling, wherein validity feature database represents meaning first is that having according to varying environment, season, period
Environmental background noise sample determined influence acoustic environment validity feature；Second is that being established based on validity feature enough, suitable
Masking sound database for low frequency background sound in difference.It the foundation for being configured to sound satisfaction agent model of the database and covers
The automatic screening for covering sound provides data basis.
Four, GPR (Gaussian process recurrence) model recurrence sound satisfaction model: the present invention is based on agent model thoughts, utilize
GPR model regression combination sound satisfaction model, has effectively recognized high latitude, the non-linear complexity that mechanism model is difficult to
System.Using GPR algorithm regression combination sound satisfaction agent model, a group predictor formula for chorus satisfaction is derived, for masking
Sound screening subsystem provides high-precision Filtering system.Also, individual subjective bias is considered in GPR model to sound satisfaction
The influence of value further improves the precision and reliability of satisfaction degree estimation.
Five, the sound masking system based on immunologic mechanism: the present invention is devised based on masking sound database, group chorus satisfaction
The masking sound of agent model screens subsystem, and introduces immunological memory mechanism in screening subsystem, and subsystem can be enhanced
The adaptivity of screening promotes masking sound matching efficiency, saves screening time, is on the one hand embodied in a group chorus satisfaction degree estimation
Formula is Filtering system, and the automatic screening carried out in masking sound database realizes the Optimum Matching of background sound；Another aspect body
Immunological memory mechanism is introduced into screening subsystem now, the direct matching to background sound is repeated is realized, further enhances masking
Sound screens the adaptivity of subsystem, improves sound masking screening system efficiency.
Detailed description of the invention
Fig. 1 is the construction method flow chart of the adaptive sound masking system of the invention based on GPR model；
Fig. 2 is the principle organization chart of adaptive sound masking system of the invention；
Fig. 3 is the masking sound screening process figure of adaptive sound masking system of the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiment is a part of the embodiments of the present invention, rather than whole embodiments.Based on this hair
Embodiment in bright, those of ordinary skill in the art's every other reality obtained without making creative work
Example is applied, all should belong to the scope of protection of the invention.
Some researches show that when environmental background noise is lower than 70dB (A), it is satisfied that sound masking technology can be improved acoustic environment
Degree, sound level should take noise reduction measure to be handled first when being higher than 70dB (A).The background that the present invention is directed to lower than 70dB (A) is made an uproar
Sound provides a kind of intelligent sound masking technology for improving acoustic environment satisfaction.Its basic principle is based in psychologic acoustics
Masking effect thought, choose the physics and psychoacoustic characteristics of suitable masking sound adjustment background sound, and then improve crowd couple
The satisfaction of acoustic environment.
Since different background sounds has different a physical features and Perception Features, and the feature of each natural phonation is also phase not to the utmost
Together.It determines that ambient noise influences the uniform characteristics of satisfaction, masking sound is targetedly added based on these features, can be adjusted
The characteristic value of background sound, and then play the effect of adjustment acoustic environment satisfaction.It is high in order in the case where reducing cost as far as possible
Effect filters out the corresponding optimal masking sound of background sound, and the present invention devises the adaptive sound masking system with high-precision and universality
System, as shown in Figure 1, firstly, determining the validity feature of ambient noise, root by subjective assessment and Principal Component Analysis
Masking sound database is established according to validity feature；Then, GPR algorithm modeling sound satisfaction agent model is utilized；Finally, being based on GPR
Model prediction formula constructs adaptive masking sound screening subsystem on masking sound database.Steps are as follows for specific method:
Firstly, determining the validity feature of environmental background noise.Studies have shown that determining the principal element of acoustic environment satisfaction is
Sound pressure level and loudness, it is contemplated that masking sound is added can improve acoustic environment satisfaction in the suitably increased situation of loudness, therefore
It needs to be determined that influence of other parameters to acoustic environment satisfaction.Specifically, varying environment, season, period are acquired with representative meaning
M environmental background noise sample of justice carries out subjective assessment.Experiment is played back using multichannel loudspeaker, and reduction as far as possible is practical
The sound field of environment obtains crowd to the satisfaction evaluation of noise sample as a result, really using the semantic differential analysis of multistage odd number scope
It is fixed the basic physical features of sound satisfaction: A sound levelCharacteristic frequency sound pressure levelLoudnessRoughness R, acutance S,
Wow flutter F and tone degree K.Dimensionality reduction is carried out to above-mentioned main feature parameter using Principal Component Analysis, obtains influencing acoustic environment
A validity feature parameter of d (≤7) of satisfaction:
Then, masking sound database is constructed.The validity feature parameter tieed up according to dAcquire intermittent insects
A large amount of natural phonation data (or various natural phonation databases are downloaded on the net) such as tweeting sound, birds tweeting sound and water flow sound.Into
One step expands data volume by adjusting the value of each validity feature in natural phonation sample, constructs a masking sound databaseThe database is made of effective parameter of above-mentioned natural phonation, and includes nature
Sound data volume is N.
Thereafter, GPR model regression combination sound satisfaction model is utilized.Using uniform sampling, n are acquired in masking data library and is covered
Cover sound sampleWith m background sound sampleIt is combined, obtains mn group chorus
By subjective assessment, the satisfaction observation y of group chorus is obtained1,y2,…ymn.Since these observations are by different
The influence of body subjective differences, thus, it is supposed that y1,y2,…ymnIt is 0 by a mean value, variance isRandom deviation ε influence.
Due to the actual mechanism model of sound satisfactionIt is difficult to obtain, be thought based on agent model
Think, utilizes approximate potential (true) satisfaction model of Data Modeling Method.Utilize mn group chorus sampleThe agent model of training satisfaction.By analyzing mn training sampleData distribution trend, it is known that potential satisfaction modelIt is a high latitude
Degree, nonlinear complicated function, therefore, using GPR algorithm modeling agent model.It can derive any Combination nova sound samplePrediction model, it meets mean value and isVariance isGaussian Profile:
Finally, establishing adaptive masking sound screening subsystem.Using the group chorus prediction model of formula (1), to arbitrarily newly arriving
The background sound comeIn masking sound databaseMiddle carry out automatic screening, calculates a group maximum value for chorus satisfaction
This point is the background soundOptimal masking soundFurther, immunological memory mechanism is introduced in screening subsystem, every
Memory cell is generated while primary masking sound successful match, automatic activation note interim to repeat background sound next time
Recall cell, reaches immediately directly matching.
It is illustrated in figure 2 the principle organization chart of adaptive sound masking system of the invention, establishes system concrete principle
Steps are as follows:
1, background sound sample, natural phonation sample are acquired, masking sound database is established
1) background sound sample is acquiredAcquire different scenes, Various Seasonal and
Different periods have the m environmental background sound sample (each sample duration is identical) for representing meaning, pass through subjective assessment analysis of experiments
The feature of ambient noise out: A sound levelCharacteristic frequency sound pressure levelAnd loudnessRoughness R, acutance S, wow flutter F with
And tone degree K.Further, using Principal Component Analysis, obtain the main feature for influencing different background sound satisfaction, determine d (≤
7) a main feature parameter, is denoted as:Defining this d main feature amount is the effective of influence acoustic environment satisfaction
Characteristic parameter；
2) masking sound database is constructedIt is adopted according to validity feature parameter
Collect a large amount of natural phonation data such as intermittent insects tweeting sound, birds tweeting sound and water flow sound (or by adjusting natural phonation
The value enlarged sample amount of each validity feature in sample).It is built using the N number of natural phonation data and its corresponding validity feature value of acquisition
Vertical masking sound database
2, agent model approximation set chorus satisfaction model is utilized
1) uniform sampling: using uniform sampling method in masking sound databaseMiddle acquisition n masking
Sound sample
2) acquisition group chorus sampleBy m background sound of acquisition point
It is not combined with n masking sound sample, obtains mn group chorus sample and corresponding mn group validity feature parameter value
3) subjective assessment: to mn group chorusSatisfaction subjective assessment is carried out, obtains mn
Satisfaction observation y1,y2,…ymn, each observation yk(k=1 ..., mn) it indicates really to be satisfied with angle valueWith subjectivity individual
Deviation εkSum.
4) training group chorus satisfaction agent model: consider agent modelWhereinIt is the d dimension input variable being made of the effective feature volume of group chorus,Indicate potential (true
It is real) extent function, y () indicates the satisfaction observation function for corresponding to group chorus sample, its shadow by a random deviation ε
It rings.Here ε expression sound is satisfied with angle value to be influenced by Different Individual subjective differences, it is assumed that ε meets Gaussian Profile:
5) GPR is modeled: due to a group chorus satisfaction modelIt is a high-dimensional Continuous Nonlinear complication system,
Utilize GPR model foundation agent model, it is assumed that mn is potential (true) to be satisfied with angle valueObey a Gauss mistake
Journey:Wherein,It indicatesMean function,Indicate sample
WithBetween covariance function, i, j=1 ..., mn.For simplified model, zero-mean gaussian process priori is considered, evenTherefore, mn extent function valueMeet Gaussian process priori:In turn, mn observation Y=[y1,…,ymn]TMeet Gaussian Profile
Finally, any one Combination nova sound sample can be derivedPrediction model, it is satisfied with angle valueMeet equal
Value isVariance isGaussian Profile:
In formula,Respectively indicate any combination soundSatisfaction predicted value and prediction variance,Degree of being satisfied with observation vector, K*,mn、KmnAnd K**Equal representative function association side
Poor matrix,Indicate deviation covariance matrix, ΙmnIndicate mn grades of unit matrixs.
Here, covariance matrix K*,mn、KmnAnd K**By covariance function valueIt constitutes, tool
Body expression formula is as follows:
Here using square exponential function as covariance function, i.e.,Calculation formula it is as follows:
3, it establishes the adaptive masking sound based on immunological memory mechanism and screens subsystem
Adaptive masking sound screening: background sound of newly arriving is identifiedActivate memory cell, judgementWhether completely new background sound:
If repeating background sound, masking sound is remembered in directly matching；If completely new background sound, calculated separately using predictor formula (1)With
N number of masking sound in databaseGroup chorus satisfactionIt searches maximum and is satisfied with angle value
Corresponding masking sound isOptimal masking soundAnd generate memory cell.Masking sound filtering algorithm flow chart such as Fig. 3.
It further concludes, obtains the noise control method using the adaptive sound masking system based on GPR model, this method
The following steps are included:
Step 01: handling background sound of newly arriving, obtain its effective parameter value；
Step 02: the background sound input adaptive masking sound being screened into subsystem, background sound identify and while being swashed
Memory cell living；
Step 03: if being to repeat background sound to the background sound recognition result, directly matching is remembered in masking sound database
Recall optimal masking sound；
Step 04: if being completely new background sound to the background sound recognition result, the predictor formula of group chorus satisfaction is utilized,
Calculate one by one the background sound and data can in library N number of masking sound group chorus satisfaction value, screened in masking sound database
It is satisfied with the corresponding optimal masking sound of angle value to maximum, and generates memory cell simultaneously；
Step 05: exporting the optimal masking sound and control intervention is carried out to background sound.
Embodiment 1:
Along the crossing and arterial street of heavy traffic, for the traffic noise of low, middle and high frequency section, pass through data processing
Determine the effective parameter of its noiseThen, it is screened, is being sheltered using adaptive masking sound system
Sound databaseIn match optimal masking soundCorresponding optimal set chorus is obtained simultaneously
In this context, using optimal masking soundTo the traffic noiseIntervened, experiences the crowd in the environment
The combined sound of traffic sound and masking soundThe subjective satisfaction of crowd in the environment is improved, reduces crowd because of traffic
The impression of worry caused by noise.
Embodiment 2:
In the outdoor platform of rail traffic, the wheel-rail noise generated for train in/out station?
It is screened in adaptive masking sound system, matches optimal masking soundAnd combinations thereof soundIn this platform, use
The masking soundIt is rightIntervened, obtains the optimal set chorus under the environmentKeep the frequency spectrum of ambient sound more balanced,
The interference that the crowd of reducing is subject to, cooperation station in launch corresponding landscape poster or video, allow the public seemingly stay nature it
In, it is worried to alleviate noise bring.
The present invention is based on the related laws of psychologic acoustics, improve living environment from the subjective feeling angle of people, are promoted
Crowd has agreed with people-oriented theory to the satisfaction of acoustic environment；Using artificial intelligence the relevant technologies, the sense of noise subjectivity is proposed
The agent model received constructs the adaptive sound masking system of wisdom, realizes intelligentized sound masking technology, by artificial intelligence
Energy technological incorporation is in field of noise control.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (8)
1. a kind of adaptive sound masking system based on GPR model, which is characterized in that the system is full by masking sound database, sound
Meaning degree agent model and adaptive masking sound screening subsystem are built-up, and the masking sound database is by intermittent insects
Tweeting sound, birds tweeting sound, water flow sound and a variety of natural phonations are constituted, and the sound satisfaction agent model is returned by Gaussian process and calculated
Method modeling is constituted, and the adaptive masking sound screening subsystem is acted on behalf of based on the masking sound database and the sound satisfaction
Model, and introduce immunological memory mechanism construction and form.
2. it is a kind of for a kind of construction method of the adaptive sound masking system based on GPR model as described in claim 1,
It is characterized in that, method includes the following steps:
Step 1: acquisition background sound sample and natural phonation sample simultaneously establish masking sound database using the sample data of acquisition；
Step 2: utilizing GPR model regression combination sound satisfaction agent model；
Step 3: constructing adaptive masking sound screening subsystem on masking sound database based on sound satisfaction agent model.
3. a kind of construction method of adaptive sound masking system based on GPR model according to claim 2, feature exist
In, the step 1 include it is following step by step:
Step 11: acquisition different scenes, the environmental background sound sample in season and period, and obtained using Principal Component Analysis multiple
Influence the validity feature parameter of acoustic environment satisfaction；
Step 12: intermittent insects tweeting sound, birds tweeting sound, water flow sound being acquired based on validity feature parameter and other are a variety of
The data of natural phonation establish masking sound database using the data and its corresponding validity feature parameter of acquisition.
4. a kind of construction method of adaptive sound masking system based on GPR model according to claim 3, feature exist
In the validity feature parameter in the step 11 is the main feature for influencing background sound satisfaction, especially by noise object
Reason feature: A sound level, characteristic frequency sound pressure level, loudness, roughness, acutance, wow flutter and tone degree carry out principal component analysis and obtain
?.
5. a kind of construction method of adaptive sound masking system based on GPR model according to claim 2, feature exist
In, the step 2 include it is following step by step:
Step 21: uniform sampling approach being taken to acquire masking sound sample in masking sound database；
Step 22: combining environmental background sound sample with masking sound sample to obtain group chorus sample and corresponding multiple groups validity feature
Parameter value；
Step 23: satisfaction subjective assessment test being carried out to all groups of choruses in group chorus sample and obtains corresponding satisfaction
Observation；
Step 24: using in group chorus sample all groups of chorus validity feature values and its corresponding satisfaction observation as training
Sample；
Step 25: consideration group chorus satisfaction agent model, and model the sound satisfaction using Gaussian process regression algorithm and act on behalf of
Model.
6. a kind of construction method of adaptive sound masking system based on GPR model according to claim 5, feature exist
In agent model in the step 24 describes formula are as follows:
In formula, chorus sample is organizedIndicate the d being made of group chorus effective feature volume dimension input variable,
Indicate that potential (true) extent function, y () indicate to correspond to group chorus sampleSatisfaction observation function, random deviation
ε indicates that true sound is satisfied with angle value and is influenced by Different Individual subjective differences.
7. a kind of construction method of adaptive sound masking system based on GPR model according to claim 5, feature exist
In Gaussian process regression model in the step 25 derives satisfaction degree estimation formula are as follows:
In formula,Respectively indicate any combination soundSatisfaction predicted value and prediction variance,Degree of being satisfied with observation vector, K*,mn、KmnAnd K**Equal representative function association side
Poor matrix,Indicate deviation covariance matrix, ΙmnIndicate mn grades of unit matrixs.
8. a kind of construction method of adaptive sound masking system based on GPR model according to claim 2, feature exist
In, which is characterized in that the step 3 include it is following step by step:
Step 31: adaptive masking sound screening subsystem identifies any new background sound and activates memory cell simultaneously；
Step 32: if being to repeat background sound to new background sound recognition result, in the masking sound data of adaptively masking sound system
Directly optimal masking sound is remembered in matching in library；
Step 33: if being completely new background sound to new background sound recognition result, using the predictor formula of group chorus satisfaction, one by one
The group chorus satisfaction value of the background sound Yu masking sound database is calculated, screening obtains maximum and is satisfied with angle value correspondence in the database
Optimal masking sound, and generate memory cell simultaneously；
Step 34: exporting the optimal masking sound and control intervention is carried out to new background sound.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910656438.8A CN110362789B (en) | 2019-07-19 | 2019-07-19 | GPR model-based adaptive sound masking system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910656438.8A CN110362789B (en) | 2019-07-19 | 2019-07-19 | GPR model-based adaptive sound masking system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110362789A true CN110362789A (en) | 2019-10-22 |
| CN110362789B CN110362789B (en) | 2024-02-02 |
Family
ID=68221159
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910656438.8A Active CN110362789B (en) | 2019-07-19 | 2019-07-19 | GPR model-based adaptive sound masking system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110362789B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112932489A (en) * | 2020-12-30 | 2021-06-11 | 广东电网有限责任公司电力科学研究院 | Transformer substation noise subjective annoyance degree evaluation model establishing method and model establishing system |
| WO2023211385A1 (en) * | 2022-04-27 | 2023-11-02 | Nanyang Technological University | Soundscape augmentation system and method of forming the same |
| EP4365890A1 (en) | 2022-11-07 | 2024-05-08 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Adaptive harmonic speech masking sound generation apparatus and method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030103632A1 (en) * | 2001-12-03 | 2003-06-05 | Rafik Goubran | Adaptive sound masking system and method |
| CN106933146A (en) * | 2017-03-14 | 2017-07-07 | 吉林大学 | Electrocar pedestrian's caution sound method for designing, caution sound control system and method |
| CN108357445A (en) * | 2018-03-20 | 2018-08-03 | 吉林大学 | Car masking sound quality self-adapting control system and method |
| CN109238448A (en) * | 2018-09-17 | 2019-01-18 | 上海市环境科学研究院 | A method of acoustic environment satisfaction is improved based on sound masking |
-
2019
- 2019-07-19 CN CN201910656438.8A patent/CN110362789B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030103632A1 (en) * | 2001-12-03 | 2003-06-05 | Rafik Goubran | Adaptive sound masking system and method |
| CN106933146A (en) * | 2017-03-14 | 2017-07-07 | 吉林大学 | Electrocar pedestrian's caution sound method for designing, caution sound control system and method |
| CN108357445A (en) * | 2018-03-20 | 2018-08-03 | 吉林大学 | Car masking sound quality self-adapting control system and method |
| CN109238448A (en) * | 2018-09-17 | 2019-01-18 | 上海市环境科学研究院 | A method of acoustic environment satisfaction is improved based on sound masking |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112932489A (en) * | 2020-12-30 | 2021-06-11 | 广东电网有限责任公司电力科学研究院 | Transformer substation noise subjective annoyance degree evaluation model establishing method and model establishing system |
| WO2023211385A1 (en) * | 2022-04-27 | 2023-11-02 | Nanyang Technological University | Soundscape augmentation system and method of forming the same |
| EP4365890A1 (en) | 2022-11-07 | 2024-05-08 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Adaptive harmonic speech masking sound generation apparatus and method |
| WO2024099913A1 (en) | 2022-11-07 | 2024-05-16 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Device and method for adaptive, harmonic voice masking sound generation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110362789B (en) | 2024-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rychtáriková et al. | Soundscape categorization on the basis of objective acoustical parameters | |
| CN110362789A (en) | A kind of adaptive sound masking system and method based on GPR model | |
| Kang | Urban sound environment | |
| Botteldooren et al. | Understanding urban and natural soundscapes | |
| CN109238448A (en) | A method of acoustic environment satisfaction is improved based on sound masking | |
| CN105788592A (en) | Audio classification method and apparatus thereof | |
| CN103077267A (en) | Parameter sound source modeling method based on improved BP (Back Propagation) neural network | |
| Hong et al. | Cognitive persistence of soundscape in urban parks | |
| CN105159066B (en) | A kind of intelligent music Room regulation and control method and regulation device | |
| CN103377656A (en) | Five-tone analysis method for audio frequency file, player, and electronic equipment | |
| Lavandier et al. | Urban soundscape maps modelled with geo-referenced data | |
| CN110333484B (en) | Indoor area level positioning method based on environmental background sound perception and analysis | |
| Martins et al. | Built heritage research and education | |
| Yue et al. | A visualized soundscape prediction model for design processes in urban parks | |
| Till | Songs of the Stones: an investigation into the musical history and culture of Stonehenge | |
| Hao | Effects of urban morphology on urban sound environment from the perspective of masking effects | |
| CN113327624B (en) | Method for intelligent monitoring of environmental noise by adopting end-to-end time domain sound source separation system | |
| Maag et al. | The sonic public realm-chances for improving the sound quality of the everyday city | |
| Polli | Atmospherics/Weatherworks: A Multi-Channel Storm Sonification Project. | |
| Xinchen et al. | Modeling Long-Term Perceived Soundscape in Urban Parks: A Case Study of Three Urban Parks in Vancouver, Canada | |
| CN107371122A (en) | The implementation method of auxiliary positioning based on electronic equipment behavior pattern | |
| Liwosz | Petroglyphs and p uha: how multisensory experiences evidence landscape agency | |
| Yan et al. | Study on the Spatio-temporal Characteristics of Soundscape in Famous Historical and Cultural Towns: a Case Study of Labrang Town, Xiahe County, China | |
| Badino | Acoustic performance-based design: exploration of the effects of different geometries and acoustical properties of an urban façade on the mitigation of chatting noise in a street canyon | |
| Yu et al. | Using ANN to study sound preference evaluation in urban open spaces |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |