CN106384598A - Noise quality determination method and device - Google Patents
Noise quality determination method and device Download PDFInfo
- Publication number
- CN106384598A CN106384598A CN201610694818.7A CN201610694818A CN106384598A CN 106384598 A CN106384598 A CN 106384598A CN 201610694818 A CN201610694818 A CN 201610694818A CN 106384598 A CN106384598 A CN 106384598A
- Authority
- CN
- China
- Prior art keywords
- noise
- parameter
- weighted value
- characteristic parameter
- type
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
- G10L25/30—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique using neural networks
Abstract
The embodiment of the invention provides a noise quality determination method and device. The noise quality determination method comprises the steps of acquiring a plurality of target characteristic parameters of noise to be recognized; acquiring weight values corresponding to the target characteristic parameters; determining the noise quality of the noise to be recognized according to the plurality of target characteristic parameters and the weight values of the target characteristic parameters. The method and device provided by the embodiment of the invention are used for improving the noise quality determination accuracy.
Description
Technical field
The present embodiments relate to technical field of acoustics, more particularly, to a kind of noise qualities determine method and device.
Background technology
With scientific and technical continuous development, the requirement to product (such as air-conditioning, refrigerator etc.) noise qualities for the user is more next
Higher.
At present, the noise qualities of product have become the important indicator that user buys product, therefore, before launch,
Typically require the noise qualities determining product.In the prior art, usual startup optimization product, obtains product in running
Produce the decibel value of noise, and noise qualities are determined according to the decibel value of noise.However, decibel value identical difference noise on human
The impact of body comfort level is different, and therefore, in the prior art, the accuracy of the noise qualities determining according only to decibel value is poor.
Content of the invention
The embodiment of the present invention provides a kind of noise qualities to determine method and device, determines the accurate of noise qualities for improving
Property.
In a first aspect, the embodiment of the present invention provides a kind of noise qualities to determine method, including：
Obtain multiple target characteristic parameters of noise to be identified；
Obtain the corresponding weighted value of each described target characteristic parameter；
According to the weighted value of the plurality of target characteristic parameter and each described target characteristic parameter, determine described to be identified
The noise qualities of noise.
In a kind of possible embodiment, described obtain the corresponding weighted value of each described target characteristic parameter, including：
Obtain the target component type of each described target characteristic parameter；
The corresponding weighted value of each described target component type, wherein, described presetting database is obtained in presetting database
Include multiple parameters type and the corresponding weighted value of each described parameter type.
In alternatively possible embodiment, presetting database obtains the corresponding power of each described target component type
Before weight values, also include：
Obtain multiple characteristic parameters and the corresponding user's evaluating of each described sample noise of sample noise；
Multiple characteristic parameters according to described sample noise and the corresponding user's evaluating of each described sample noise, raw
Become the corresponding weighted value of parameter type of each described characteristic parameter；
The corresponding weighted value of each described parameter type is stored in described presetting database.
In alternatively possible embodiment, described multiple characteristic parameters according to described sample noise and each described
Sample noise corresponding user evaluating, generates the corresponding weighted value of parameter type of each described characteristic parameter, including：
According to the characteristic parameter of each described sample noise, determine the corresponding characteristic vector of each described sample noise；
By neutral net to each described characteristic vector and the corresponding user's evaluating of each described characteristic vector at
Reason, obtains the corresponding weighted value of parameter type of each described characteristic parameter.
In alternatively possible embodiment, described neutral net includes BP neural network, RBF neural, broad sense
At least one in recurrent neural networks.
In alternatively possible embodiment, described according to the plurality of target characteristic parameter and each described target is special
Levy the weighted value of parameter, determine the noise qualities of described noise to be identified, including：
According to equation below one, determine noise qualities Q of described noise to be identified：
Wherein, described n is the number of target characteristic parameter, described λiFor the weighted value of i-th target characteristic parameter, described
kiFor i-th target characteristic parameter.
In alternatively possible embodiment, the described multiple characteristic parameters obtaining noise to be identified, including：
Obtain multiple parameter preset types；
According to each described parameter preset type, obtain described noise to be identified, each described parameter preset type corresponding
Characteristic parameter.
In alternatively possible embodiment, the type of described target characteristic parameter includes loudness, sharpness, coarse
At least one in degree, shake degree.
Second aspect, the embodiment of the present invention provides a kind of noise qualities to determine device, including：
First acquisition module, for obtaining multiple target characteristic parameters of noise to be identified；
Second acquisition module, for obtaining the corresponding weighted value of each described target characteristic parameter；
Determining module, for the weighted value according to the plurality of target characteristic parameter and each described target characteristic parameter, really
The noise qualities of fixed described noise to be identified.
In a kind of possible embodiment, described second determining module specifically for：
Obtain the target component type of each described target characteristic parameter；
The corresponding weighted value of each described target component type, wherein, described presetting database is obtained in presetting database
Include multiple parameters type and the corresponding weighted value of each described parameter type.
In alternatively possible embodiment, described device also includes the 3rd acquisition module, generation module and storage mould
Block, wherein,
Described 3rd acquisition module is used for, and obtains each described target ginseng in described second acquisition module in presetting database
Before the corresponding weighted value of several classes of type, obtain multiple characteristic parameters and the corresponding user of each described sample noise of sample noise
Evaluating；
Described generation module is used for, and the multiple characteristic parameters according to described sample noise and each described sample noise correspond to
User's evaluating, generate each described characteristic parameter the corresponding weighted value of parameter type；
Described memory module is used for, and stores the corresponding weighted value of each described parameter type in described presetting database.
In alternatively possible embodiment, described generation module specifically for：
According to the characteristic parameter of each described sample noise, determine the corresponding characteristic vector of each described sample noise；
By neutral net to each described characteristic vector and the corresponding user's evaluating of each described characteristic vector at
Reason, obtains the corresponding weighted value of parameter type of each described characteristic parameter.
In alternatively possible embodiment, described neutral net includes BP neural network, RBF neural, broad sense
At least one in recurrent neural networks.
In alternatively possible embodiment, described determining module specifically for：
According to equation below one, determine noise qualities Q of described noise to be identified：
Wherein, described n is the number of target characteristic parameter, described λiFor the weighted value of i-th target characteristic parameter, described
kiFor i-th target characteristic parameter.
In alternatively possible embodiment, described first acquisition module specifically for：
Obtain multiple parameter preset types；
According to each described parameter preset type, obtain described noise to be identified, each described parameter preset type corresponding
Characteristic parameter.
In alternatively possible embodiment, the type of described target characteristic parameter includes loudness, sharpness, coarse
At least one in degree, shake degree.
Noise qualities shown in the embodiment of the present invention determine method and device, when determining device it needs to be determined that noise to be identified
Noise characteristic when, determine that device obtains multiple target characteristic parameters of noise to be identified and each target characteristic parameter is corresponding
Weighted value, according to the weighted value of multiple target characteristic parameters and each target characteristic parameter, determines the noise product of noise to be identified
Matter.In above process, because the corresponding weighted value of target characteristic parameter is the objective spy determining device according to sample noise
Levy the parameter and user subjectivity to sample noise user's evaluating determine obtain so that target characteristic parameter is corresponding
Weighted value is more accurate, and then makes to determine the weighted value according to multiple target characteristic parameters and each target characteristic parameter for the device,
Determine that the noise qualities obtaining are more accurate, improve the accuracy determining noise qualities.
Brief description
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
Have technology description in required use accompanying drawing be briefly described it should be apparent that, drawings in the following description are these
Some bright embodiments, for those of ordinary skill in the art, without having to pay creative labor, acceptable
Other accompanying drawings are obtained according to these accompanying drawings.
The noise qualities that Fig. 1 provides for the present invention determine the application scenarios schematic diagram of method；
The noise qualities that Fig. 2 provides for the present invention determine the schematic flow sheet of method；
The schematic flow sheet of the determination weighted value method that Fig. 3 provides for the present invention；
The schematic flow sheet of the determination parameter type corresponding weighted value method that Fig. 4 provides for the present invention；
The noise qualities that Fig. 5 provides for the present invention determine the structural representation one of device；
The noise qualities that Fig. 6 provides for the present invention determine the structural representation two of device.
Specific embodiment
Purpose, technical scheme and advantage for making the embodiment of the present invention are clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described it is clear that described embodiment is
The a part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment being obtained under the premise of not making creative work, broadly falls into the scope of protection of the invention.
The noise qualities that Fig. 1 provides for the present invention determine the application scenarios schematic diagram of method, refer to Fig. 1, including noise
Quality determining device 101.This noise qualities determines that device can determine the noise qualities of multiple product, and wherein, this product includes
But it is not limited only to air-conditioning, refrigerator, washing machine, computer etc..When it needs to be determined that noise to be identified noise qualities when, noise qualities are true
Determine the various features parameter that device first obtains noise to be identified, and according to default identification model to various features parameter at
Reason, obtains the noise qualities of noise to be identified.In this application, the noise product of the various features parameter determination noise according to noise
Matter is so that determine that the noise qualities obtaining are more accurate.Below, by specific embodiment, the technical scheme shown in the application is entered
Row describes in detail.
It should be noted that these specific embodiments can be combined with each other, for same or analogous concept below
Or process may repeat no more in certain embodiments.
The noise qualities that Fig. 2 provides for the present invention determine the schematic flow sheet of method, and the executive agent of the method can be
Noise qualities determine device (hereinafter referred determines device).Refer to Fig. 2, the method can include：
S201, multiple target characteristic parameters of acquisition noise to be identified；
S202, the acquisition corresponding weighted value of each target characteristic parameter；
S203, the weighted value according to multiple target characteristic parameters and each target characteristic parameter, determine making an uproar of noise to be identified
Sound quality.
In the embodiment shown in Figure 2, noise qualities can be represented by various features parameter, the type of this feature parameter
Loudness, sharpness, roughness, shake degree etc. can be included；It is, of course, also possible to noise is represented by other types of characteristic parameter
Quality, the present invention is not especially limited to this.
In embodiments of the present invention, determine that device determines that the process of the noise qualities of each noise is identical.Below, with true
As a example determining the process of noise qualities that device determines noise to be identified, the method shown in Fig. 2 embodiment is described in detail.
When determining that device needs the noise qualities obtaining noise to be identified, determine that device obtains the multiple of noise to be identified
Target characteristic parameter.Optionally, determine that device can obtain multiple parameter preset types, and according to each parameter preset type, obtain
Take noise to be identified, the corresponding characteristic parameter of each parameter preset type.Optionally, the plurality of parameter preset type can include
Loudness, sharpness, roughness, shake degree etc..Optionally, determine that device can manually first-class equipment collection be treated by Digital Simulation
The target characteristic parameter of identification noise.
After determining that device acquires the target characteristic parameter of noise to be identified, determine that device obtains each target special
Levy the corresponding weighted value of parameter.The corresponding weighted value of one target characteristic parameter is used for embodying this target characteristic parameter to noise product
The degree of influence of matter, wherein, the weighted value of target characteristic parameter is higher, then this target characteristic parameter power of influence to noise qualities
Degree is bigger.Optionally, the corresponding weighted value of each target characteristic parameter be determine device according to sample noise clarification of objective parameter,
And user's evaluating of the subjectivity to sample noise for the user determines and obtains.
Optionally, the corresponding weighted value of parameters type can be preset in presetting database, accordingly, determine device
The target component type of each target characteristic parameter can be obtained, presetting database obtains the corresponding power of each target component type
Weight values, wherein, presetting database includes multiple parameters type and the corresponding weighted value of each parameter type.Optionally, present count
According in storehouse storage, the corresponding weighted value of each parameter type can determine that device previously generates, the embodiment shown in Fig. 3
In to generate the corresponding weighted value of each parameter type be described in detail, herein no longer to generate weighted value process go to live in the household of one's in-laws on getting married
State.Optionally, in presetting database, the weighted value of storage can be as shown in table 1：
Table 1
| Parameter type | Weighted value |
| Parameter type 1 | λ1 |
| Parameter type 2 | λ2 |
| Parameter type 3 | λ3 |
| Parameter type 4 | λ4 |
| …… | …… |
It should be noted that table 1 simply illustrates storage form in presetting database for the weighted value in exemplary fashion, and
It is not the restriction of the storage form to weighted value.
After determining that device acquires the corresponding weighted value of each target characteristic parameter of noise to be identified, determine device
According to the weighted value of multiple target characteristic parameters and each target characteristic parameter, determine the noise qualities of noise to be identified.Optional
, determine that device can determine noise qualities Q of noise to be identified according to equation below one：
Wherein, n is the number of target characteristic parameter, λiFor the weighted value of i-th target characteristic parameter, kiFor i-th target
Characteristic parameter.
Determine the product with corresponding weighted value for the various features parameter that the noise qualities obtaining are noise according to formula one
Cumulative, accordingly, it is determined that the noise qualities obtaining are concrete numerical value, and this numerical value is higher, then the noise qualities of noise are poorer.
Determine the noise qualities obtaining noise to be identified in determination device, for the ease of the understanding to noise qualities for the user,
Noise qualities can also be sorted out, to determine the grade of noise.For example, the noise qualities in the first preset range are excellent
Good, the noise qualities in the second preset range are medium, and the noise qualities in the second preset range are poor.Optionally, make an uproar
The numerical value of sound quality can be as shown in table 2 with the corresponding relation of noise grade：
Table 2
| The numerical value of noise qualities | Noise grade |
| 1-5 | Excellent |
| 6-10 | Medium |
| More than 10 | Difference |
| …… | …… |
It should be noted that table 2 is simply illustrated in exemplary fashion, and the numerical value of noise qualities is corresponding with noise grade to close
System, is not the restriction of the numerical value to noise qualities and the corresponding relation of noise grade, in actual application, can basis
It is actually needed this corresponding relation of setting.
Below, by specific example, the method shown in Fig. 2 embodiment is described in detail.
Exemplary it is assumed that noise to be identified is noise 1, when determine device it needs to be determined that noise 1 noise qualities when, really
Determine device first obtain multiple target characteristic parameters of noise to be identified and the corresponding weighted value of each target characteristic parameter it is assumed that
The weighted value of multiple target characteristic parameters of noise to be identified and each target characteristic parameter is as shown in table 3：
Table 3
| Parameter type | Characteristic parameter | Weighted value |
| Loudness | 3 | 0.5 |
| Shake degree | 6 | 0.4 |
| Roughness | 7 | 0.6 |
| Sharpness | 5 | 0.4 |
| …… | …… | …… |
Determine the weighted value of multiple target characteristic parameters according to table 3 for the device and each target characteristic parameter, determine
To the noise qualities of noise 1 be：3*0.5+6*0.4+7*0.6+5*0.4=10.1.
Determine that this noise 1 can be defined as excellent according to the corresponding relation of noise qualities and predetermined level by device.
Noise qualities shown in the embodiment of the present invention determine method, when determining device it needs to be determined that the noise of noise to be identified
During characteristic, determine that device obtains multiple target characteristic parameters and the corresponding weighted value of each target characteristic parameter of noise to be identified,
According to the weighted value of multiple target characteristic parameters and each target characteristic parameter, determine the noise qualities of noise to be identified.Above-mentioned
During, due to the corresponding weighted value of target characteristic parameter be determine device according to the clarification of objective parameter of sample noise and
User's evaluating of the subjectivity to sample noise for the user determine obtaining so that the corresponding weighted value of target characteristic parameter more
Accurately, and then so that determining the weighted value according to multiple target characteristic parameters and each target characteristic parameter for the device, determination obtains
Noise qualities are more accurate, improve the accuracy determining noise qualities.
On the basis of embodiment illustrated in fig. 2, obtain each target component type pair in presetting database in determination device
Before the weighted value answered, determine that device can first determine the weighted value of many kinds of parameters type.Below, by the enforcement shown in Fig. 3
Example, is described in detail to the method determining the corresponding weighted value of parameter type.
The schematic flow sheet of the determination weighted value method that Fig. 3 provides for the present invention, refers to Fig. 3, the method can include：
S301, the multiple characteristic parameters obtaining sample noise and each sample noise corresponding user evaluating；
S302, the multiple characteristic parameters according to sample noise and each sample noise corresponding user evaluating, generate each
The corresponding weighted value of parameter type of characteristic parameter；
S303, store the corresponding weighted value of each parameter type in presetting database.
It should be noted that determining that device can periodically execute the method shown in Fig. 3 embodiment, to determine each parameter
The corresponding weighted value of type, accordingly, when determining that device executes the method shown in Fig. 2 embodiment it is only necessary to from preset data
The weighted value of relevant parameter type is obtained in storehouse；Certainly, determine that device can also execute the enforcement shown in Fig. 2 each time
Before example, it is performed both by the embodiment shown in Fig. 3.In actual application, determination device can be set according to actual needs and hold
In the moment of method shown in row Fig. 3 embodiment, the present invention is not especially limited to this.
When determining device it needs to be determined that during the corresponding weighted value of parameter type, determining that device first determines the multiple of sample noise
Characteristic parameter.The noise that this sample noise can be pre-selected for user, in actual application, can be according to actual needs
Determine the number of sample noise.For example, the number of sample noise can be 100,200 etc..Optionally, determine that device can lead to
Manually first-class equipment gathers multiple characteristic parameters of each sample noise to cross Digital Simulation.
Determine that device also obtains user's evaluating to each sample noise for the user.Optionally, subjectivity can be set up in advance
Evaluate team, each of subjective assessment team user can be giveed training, and be existed by the user in subjective assessment team
Under default environment, audition is carried out to sample noise, and given a mark for each sample noise, accordingly, can be according to each user
Marking to each sample noise, determines user's evaluating of each sample noise.Optionally, can by multiple users to
The meansigma methodss of the marking of this noise are defined as user's evaluating of this sample noise；Can also be to multiple users to same sample
The marking of noise is screened, and is effectively given a mark, and determines user's evaluating of sample noise according to effective marking.Need
Illustrate, user's evaluating of sample noise, the present invention in actual application, can also be determined according to alternate manner
This is not especially limited.
After determining that device determination obtains the corresponding weighted value of parameter type of each characteristic parameter, determine device according to sample
Multiple characteristic parameters of this noise and each sample noise corresponding user evaluating, generate the parameter type of each characteristic parameter
Corresponding weighted value.
In the embodiment shown in fig. 3, determine device according to the clarification of objective parameter of sample noise and user to sample
User's evaluating of the subjectivity of noise determines weighted value, so, determine the weighted value that obtains more conform to practical situation so that
Weighted value is more accurate, further increases the accuracy determining noise qualities.
On the basis of embodiment illustrated in fig. 3, optionally, can be by implementation feasible as follows according to sample noise
Multiple characteristic parameters and each sample noise corresponding user evaluating, generate each characteristic parameter parameter type corresponding
Weighted value (S302 in embodiment illustrated in fig. 3), specifically, embodiment shown in Figure 4.
The schematic flow sheet of the determination parameter type corresponding weighted value method that Fig. 4 provides for the present invention, refers to Fig. 4,
The method can include：
S401, the characteristic parameter according to each sample noise, determine the corresponding characteristic vector of each sample noise；
S402, by neutral net, each characteristic vector and each characteristic vector corresponding user evaluating are processed,
Obtain the corresponding weighted value of parameter type of each characteristic parameter.
Neutral net has MPP, distributed information storage, good self-organizing self-learning capability, in god
Through in the running of network, can be to neural network inputs mass data, neutral net can be entered to the mass data of input
Row analysis, to determine the functional relationship between data.
In the embodiment shown in fig. 4, can be corresponding to a large amount of characteristic vector of neural network inputs and each characteristic vector
User's evaluating, neutral net can be according to the characteristic vector of input and each characteristic vector corresponding user evaluating, really
Determine the functional relationship between characteristic vector and user's evaluating, and according to this functional relationship, determine the parameter of each characteristic parameter
The corresponding weighted value of type.Optionally, neutral net can include BP neural network, RBF neural, general regression neural net
At least one in network.
During neutral net is processed to mass data, optionally, can to the parameter of neutral net (for example
Error of sum square, extension constant, number of relay cell etc.) it is adjusted, to obtain accurate weighted value.Need explanation
It is according to existing neutral net, each characteristic vector and each characteristic vector corresponding user evaluating can be processed,
The present invention no longer repeats to processing procedure.
In the embodiment shown in fig. 4, optionally, determine that device can be passed through shown in equation below two with control neural network
Model determine the corresponding weighted value of the parameter type of each characteristic parameter：
Wherein, (an1an2an3an4) be n-th sample noise characteristic vector, λmCorresponding for m-th parameter type
Weighted value, snUser is user's evaluating of n-th sample noise determination.
Determine that model according to formula two for the device control neural network corresponds to each characteristic vector and each characteristic vector
User's evaluating processed, obtain weighted value λ1To λm.It should be noted that the characteristic vector due to inputting and user
Accurate functional relationship is not had, therefore, neutral net determines the weight obtaining according to above-mentioned formula two between evaluating
Value λ1To λmIt is not applied for all of characteristic vector and characteristic vector corresponding user evaluating, weighted value can only be made
λ1To λmIt is applied to characteristic vector and the characteristic vector corresponding user evaluating of majority.
Below, by specific example, the method shown in Fig. 3-Fig. 4 embodiment is described in detail.
Exemplary it is assumed that user determines 100 sample noise, be designated as sample noise 1- sample noise 100 respectively.Again
Assume need obtain the characteristic parameter of each sample noise type include loudness, sharpness, roughness and shake degree it is determined that
Device passes through Digital Simulation, and manually first-class equipment is processed to sample noise 1- sample noise 100, to obtain sample noise 1-
The loudness of sample noise 100, sharpness, roughness and shake degree.Assume to determine the sample noise 1- sample that device acquires
The loudness of noise 100, sharpness, roughness and shake degree are as shown in table 4：
Table 4
| Sample noise | Loudness | Sharpness | Roughness | Shake degree |
| Sample noise 1 | 3 | 6 | 3 | 5 |
| Sample noise 2 | 5 | 5 | 7 | 4 |
| Sample noise 3 | 7 | 4 | 5 | 3 |
| Sample noise 4 | 5 | 2 | 5 | 5 |
| …… | …… | …… | …… | …… |
Determine the characteristic parameter of each sample noise according to table 4 for the device, determine the corresponding feature of each sample noise to
Amount is respectively as shown in table 5：
Table 5
| Sample noise | Characteristic parameter |
| Sample noise 1 | (6,3,5,5) |
| Sample noise 2 | (5,5,7,4) |
| Sample noise 3 | (7,4,5,3) |
| Sample noise 4 | (5,2,5,5) |
| …… | …… |
Audition is carried out to this 100 sample noise by each of subjective assessment team user, and to this 100 samples
Noise is given a mark it is assumed that being given a mark to each sample noise according to user, determines the use of each sample noise obtaining
Family evaluating is as shown in table 6：
Table 6
| Sample noise | User's evaluating |
| Sample noise 1 | 9 |
| Sample noise 2 | 8 |
| Sample noise 3 | 6 |
| Sample noise 4 | 7 |
| …… | …… |
Determine that device is entered to the user's evaluating shown in 100 characteristic vectors shown in table 5 and table 6 by neutral net
Row is processed, and obtains loudness, sharpness, roughness and the corresponding weighted value of shake degree.Optionally, determine that device can be with formula two institute
The model showing is processed, to obtain loudness, sharpness, roughness and the corresponding weighted value of shake degree.Specifically, determine device
Can be with control neural network according to equation below three, the characteristic vector to sample noise 1- sample noise 100 and user evaluate ginseng
Number is processed, and obtains the weighted value λ of loudness1, the weighted value λ of sharpness2, the weighted value λ of roughness3And the weight of shake degree
Value λ4.
Model according to above-mentioned formula three for the neutral net, determines λ1To λ4, so that λ1To λ4For most features to
Amount and characteristic vector corresponding user evaluating.Optionally, neutral net determination obtain loudness, sharpness, roughness and
The corresponding weighted value of shake degree can be as shown in table 7：
Table 7
| Parameter type | Weighted value |
| Loudness | 0.5 |
| Shake degree | 0.4 |
| Roughness | 0.6 |
| Sharpness | 0.4 |
| …… | …… |
The noise qualities that Fig. 5 provides for the present invention determine the structural representation one of device, refer to Fig. 5, and this device is permissible
Including：
First acquisition module 501, for obtaining multiple target characteristic parameters of noise to be identified；
Second acquisition module 502, for obtaining the corresponding weighted value of each described target characteristic parameter；
Determining module 503, for the weight according to the plurality of target characteristic parameter and each described target characteristic parameter
Value, determines the noise qualities of described noise to be identified.
Noise qualities provided in an embodiment of the present invention determine that device can execute the technical side shown in said method embodiment
Case, it realizes principle and beneficial effect is similar to, and is no longer repeated herein.
In a kind of possible embodiment, described second determining module 502 specifically for：
Obtain the target component type of each described target characteristic parameter；
The corresponding weighted value of each described target component type, wherein, described presetting database is obtained in presetting database
Include multiple parameters type and the corresponding weighted value of each described parameter type.
The noise qualities that Fig. 6 provides for the present invention determine the structural representation two of device, on the basis of embodiment illustrated in fig. 5
On, refer to Fig. 6, described device also includes the 3rd acquisition module 504, generation module 505 and memory module 506, wherein,
Described 3rd acquisition module 504 is used for, and obtains each described in presetting database in described second acquisition module 502
Before the corresponding weighted value of target component type, obtain multiple characteristic parameters of sample noise and each described sample noise corresponds to
User's evaluating；
Described generation module 505 is used for, the multiple characteristic parameters according to described sample noise and each described sample noise pair
The user's evaluating answered, generates the corresponding weighted value of parameter type of each described characteristic parameter；
Described memory module 506 is used for, and stores the corresponding weighted value of each described parameter type in described presetting database.
In alternatively possible embodiment, described generation module 505 specifically for：
According to the characteristic parameter of each described sample noise, determine the corresponding characteristic vector of each described sample noise；
By neutral net to each described characteristic vector and the corresponding user's evaluating of each described characteristic vector at
Reason, obtains the corresponding weighted value of parameter type of each described characteristic parameter.
In alternatively possible embodiment, described neutral net includes BP neural network, RBF neural, broad sense
At least one in recurrent neural networks.
In alternatively possible embodiment, described determining module 503 specifically for：
According to equation below one, determine noise qualities Q of described noise to be identified：
Wherein, described n is the number of target characteristic parameter, described λiFor the weighted value of i-th target characteristic parameter, described
kiFor i-th target characteristic parameter.
In alternatively possible embodiment, described first acquisition module 501 specifically for：
Obtain multiple parameter preset types；
According to each described parameter preset type, obtain described noise to be identified, each described parameter preset type corresponding
Characteristic parameter.
In alternatively possible embodiment, the type of described target characteristic parameter includes loudness, sharpness, coarse
At least one in degree, shake degree.
Noise qualities provided in an embodiment of the present invention determine that device can execute the technical side shown in said method embodiment
Case, it realizes principle and beneficial effect is similar to, and is no longer repeated herein.
One of ordinary skill in the art will appreciate that：The all or part of step realizing above-mentioned each method embodiment can be led to
Cross the related hardware of programmed instruction to complete.Aforesaid program can be stored in a computer read/write memory medium.This journey
Sequence upon execution, executes the step including above-mentioned each method embodiment；And aforesaid storage medium includes：ROM, RAM, magnetic disc or
Person's CD etc. is various can be with the medium of store program codes.
Finally it should be noted that：Various embodiments above only in order to technical scheme to be described, is not intended to limit；To the greatest extent
Pipe has been described in detail to the present invention with reference to foregoing embodiments, it will be understood by those within the art that：Its according to
So the technical scheme described in foregoing embodiments can be modified, or wherein some or all of technical characteristic is entered
Row equivalent；And these modifications or replacement, do not make the essence of appropriate technical solution depart from various embodiments of the present invention technology
The scope of scheme.
Claims (10)
1. a kind of noise qualities determine method it is characterised in that including：
Obtain multiple target characteristic parameters of noise to be identified；
Obtain the corresponding weighted value of each described target characteristic parameter；
According to the weighted value of the plurality of target characteristic parameter and each described target characteristic parameter, determine described noise to be identified
Noise qualities.
2. method according to claim 1 is it is characterised in that the corresponding weight of each described target characteristic parameter of described acquisition
Value, including：
Obtain the target component type of each described target characteristic parameter；
Obtain the corresponding weighted value of each described target component type in presetting database, wherein, wrap in described presetting database
Include multiple parameters type and the corresponding weighted value of each described parameter type.
3. method according to claim 2 is it is characterised in that obtain each described target component type in presetting database
Before corresponding weighted value, also include：
Obtain multiple characteristic parameters and the corresponding user's evaluating of each described sample noise of sample noise；
Multiple characteristic parameters according to described sample noise and the corresponding user's evaluating of each described sample noise, generate each
The corresponding weighted value of parameter type of described characteristic parameter；
The corresponding weighted value of each described parameter type is stored in described presetting database.
4. method according to claim 3 it is characterised in that described multiple characteristic parameters according to described sample noise,
And the corresponding user's evaluating of each described sample noise, generate the corresponding weighted value of parameter type of each described characteristic parameter,
Including：
According to the characteristic parameter of each described sample noise, determine the corresponding characteristic vector of each described sample noise；
By neutral net, each described characteristic vector and the corresponding user's evaluating of each described characteristic vector are processed,
Obtain the corresponding weighted value of parameter type of each described characteristic parameter.
5. method according to claim 4 is it is characterised in that described neutral net includes BP neural network, RBF nerve net
At least one in network, generalized regression nerve networks.
6. the method according to any one of claim 1-5 is it is characterised in that described join according to the plurality of target characteristic
Number and the weighted value of each described target characteristic parameter, determine the noise qualities of described noise to be identified, including：
According to equation below one, determine noise qualities Q of described noise to be identified：
Wherein, described n is the number of target characteristic parameter, described λiFor the weighted value of i-th target characteristic parameter, described kiFor
I-th target characteristic parameter.
7. the method according to any one of claim 1-5 is it is characterised in that multiple features of described acquisition noise to be identified
Parameter, including：
Obtain multiple parameter preset types；
According to each described parameter preset type, obtain described noise to be identified, the corresponding feature of each described parameter preset type
Parameter.
8. the method according to any one of claim 1-5 is it is characterised in that the type of described target characteristic parameter includes ringing
At least one in degree, sharpness, roughness, shake degree.
9. a kind of noise qualities determine device it is characterised in that including：
First acquisition module, for obtaining multiple target characteristic parameters of noise to be identified；
Second acquisition module, for obtaining the corresponding weighted value of each described target characteristic parameter；
Determining module, for the weighted value according to the plurality of target characteristic parameter and each described target characteristic parameter, determines institute
State the noise qualities of noise to be identified.
10. device according to claim 9 it is characterised in that described second determining module specifically for：
Obtain the target component type of each described target characteristic parameter；
Obtain the corresponding weighted value of each described target component type in presetting database, wherein, wrap in described presetting database
Include multiple parameters type and the corresponding weighted value of each described parameter type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610694818.7A CN106384598A (en) | 2016-08-18 | 2016-08-18 | Noise quality determination method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610694818.7A CN106384598A (en) | 2016-08-18 | 2016-08-18 | Noise quality determination method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106384598A true CN106384598A (en) | 2017-02-08 |
Family
ID=57916827
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610694818.7A Pending CN106384598A (en) | 2016-08-18 | 2016-08-18 | Noise quality determination method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106384598A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109141623A (en) * | 2018-08-14 | 2019-01-04 | 中车青岛四方机车车辆股份有限公司 | A kind of evaluation method and device of train in-vehicle sound quality |
| CN112530455A (en) * | 2020-11-24 | 2021-03-19 | 东风汽车集团有限公司 | Automobile door closing sound quality evaluation method and evaluation system based on MFCC |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103471709A (en) * | 2013-09-17 | 2013-12-25 | 吉林大学 | Method for predicting noise quality of noise inside passenger vehicle |
| CN103743435A (en) * | 2013-12-23 | 2014-04-23 | 广西科技大学 | Multi-sensor data fusion method |
| CN105021275A (en) * | 2015-07-15 | 2015-11-04 | 吉林大学 | Time domain dynamic characteristic analysis-based intra-vehicle sound quality prediction method |
| CN105136280A (en) * | 2015-09-18 | 2015-12-09 | 华晨汽车集团控股有限公司 | System and method for testing single noise quality in multi-source noise environment |
-
2016
- 2016-08-18 CN CN201610694818.7A patent/CN106384598A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103471709A (en) * | 2013-09-17 | 2013-12-25 | 吉林大学 | Method for predicting noise quality of noise inside passenger vehicle |
| CN103743435A (en) * | 2013-12-23 | 2014-04-23 | 广西科技大学 | Multi-sensor data fusion method |
| CN105021275A (en) * | 2015-07-15 | 2015-11-04 | 吉林大学 | Time domain dynamic characteristic analysis-based intra-vehicle sound quality prediction method |
| CN105136280A (en) * | 2015-09-18 | 2015-12-09 | 华晨汽车集团控股有限公司 | System and method for testing single noise quality in multi-source noise environment |
Non-Patent Citations (1)
| Title |
|---|
| 高印寒 等: "汽车声品质的GA-BP神经网络预测与权重分析", 《光学精密工程》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109141623A (en) * | 2018-08-14 | 2019-01-04 | 中车青岛四方机车车辆股份有限公司 | A kind of evaluation method and device of train in-vehicle sound quality |
| CN112530455A (en) * | 2020-11-24 | 2021-03-19 | 东风汽车集团有限公司 | Automobile door closing sound quality evaluation method and evaluation system based on MFCC |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107797931B (en) | Software quality evaluation method and system based on secondary evaluation | |
| DE102012102770B9 (en) | System and method for error isolation and error mitigation based on network modeling | |
| CN110717671B (en) | Method and device for determining contribution degree of participants | |
| CN106022892A (en) | Credit scoring model update method and credit scoring model update system | |
| CN111174370A (en) | Fault detection method and device, storage medium and electronic device | |
| CN106612511B (en) | Wireless network throughput evaluation method and device based on support vector machine | |
| CN109426655B (en) | Data analysis method and device, electronic equipment and computer readable storage medium | |
| CN110751175A (en) | Method and device for optimizing loss function, computer equipment and storage medium | |
| CN111143143A (en) | Performance test method and device | |
| CN110969600A (en) | Product defect detection method and device, electronic equipment and storage medium | |
| CN105933328A (en) | Method and device for processing user access behaviors | |
| CN111860568B (en) | Method and device for balanced distribution of data samples and storage medium | |
| CN111222553B (en) | Training data processing method and device of machine learning model and computer equipment | |
| CN106384598A (en) | Noise quality determination method and device | |
| Dini et al. | Water distribution network quality model calibration: a case study–Ahar | |
| CN106407659A (en) | Air quality index (AQI) predicting method and device | |
| CN103679269A (en) | Method and device for selecting classifier sample based on active learning | |
| CN107292320A (en) | System and its index optimization method and device | |
| CN107391363A (en) | Test resource scheduling method and device and electronic equipment | |
| CN111428125A (en) | Sorting method and device, electronic equipment and readable storage medium | |
| CN108280299A (en) | truss structure sensitivity analysis method | |
| CN110889635B (en) | Method for performing emergency drilling on food safety event processing | |
| DE112021001418T5 (en) | PERFORMANCE EVENTS TROUBLESHOOTING SYSTEM | |
| CN112434839B (en) | Distribution transformer heavy overload risk prediction method and electronic equipment | |
| CN109583649A (en) | The detection method and device of information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170208 |
|
| RJ01 | Rejection of invention patent application after publication |