CN110530653B - Subjective evaluation method for automobile sound quality - Google Patents
Subjective evaluation method for automobile sound quality Download PDFInfo
- Publication number
- CN110530653B CN110530653B CN201910805315.6A CN201910805315A CN110530653B CN 110530653 B CN110530653 B CN 110530653B CN 201910805315 A CN201910805315 A CN 201910805315A CN 110530653 B CN110530653 B CN 110530653B
- Authority
- CN
- China
- Prior art keywords
- sound samples
- group
- sound
- grade
- samples
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
Abstract
The invention discloses a method for subjective evaluation of sound quality and data quantization thereof, which comprises the steps of dividing sound samples into a plurality of groups, obtaining corresponding subjective evaluation grade scores after mutual comparison of the sound samples in the groups based on a grade evaluation method, then obtaining a clustering subset for a linear conversion model in each group through clustering of evaluators in each group, calculating the grade scores of the sound samples in each group and establishing a linear conversion model to obtain a subjective evaluation result. The method is suitable for subjective evaluation tests with a large number of sound samples and low evaluation experience of evaluators, can obtain accurate subjective evaluation results in a short time, and provides reasonable and effective data support for subsequent sound quality analysis.
Description
Technical Field
The invention relates to NVH performance of an automobile, in particular to a subjective evaluation method for automobile sound quality.
Background
With the popularization of automobile products and the rapid development of technologies, consumers are increasingly pursuing automobile quality. Sound quality, one of the important factors in the texture of automobile products, includes human subjective perception of sound in the automobile, and has received high attention from researchers and automobile manufacturers. The sound quality subjective evaluation experiment directly reflects the intuitive feeling of people on certain subjective evaluation attributes (such as satisfaction, annoyance and the like) of sound, provides targets for subsequent sound quality design and modification, and is an important part for embodying sound quality significance.
At present, four evaluation methods are mainly adopted in a sound quality subjective evaluation experiment: simple ranking, rank scoring, semantic subdivision, and pairwise comparison. The simple ordering method is that after the evaluator listens to all the sound samples, the sound samples are ordered. Taking the subjective satisfaction degree of the sound as an example of the subjective evaluation attribute, the closer the ranking number is, the higher the subjective satisfaction degree of the evaluator on the sound sample is.
The grade evaluation method is characterized in that a certain subjective evaluation attribute of sound is divided into a plurality of grades, after an evaluator listens to a single sound sample, the corresponding grade is selected according to subjective feeling, and common grades are divided into 1-10 grades. Referring to table 1, the subjective satisfaction of sound was taken as a subjective evaluation attribute.
TABLE 1 example of 10-level ranking of subjective satisfaction of sound
The semantic subdivision method is to carry out semantic subdivision on a pair of subjective antisense words for describing sound, and after an evaluator listens to a single sound sample, corresponding semantics are selected according to subjective feeling, wherein common semantic subdivision grades comprise 5 grades, 7 grades, 9 grades and the like. See table 2 for an example of "dissatisfied-satisfied" as the subjective evaluation attribute.
TABLE 2 "unsatisfied-satisfied" 7-level semantic subdivision example
The pairwise comparison method compares the sound samples pairwise, and referring to table 3, after the two samples are listened by an evaluator, the corresponding sound sample is selected according to the subjective evaluation attribute.
TABLE 3 paired comparative example of subjective satisfaction of sound
From the above description, it can be seen that the simple sorting method is convenient and fast, is suitable for rough comparison among sound samples, but cannot obtain good and bad scale information among sound samples, and the total number of sound samples that can be evaluated at the same time is limited, and generally does not exceed 6 sound samples. The grade grading method and the semantic subdivision method can obtain the good and bad scale information among the sound samples, are convenient and quick, and are mainly suitable for experiential evaluators. If the evaluation experience of the evaluator is not rich (such as common consumers), the paired comparison method can obtain more accurate evaluation results, but the evaluation time is exponentially increased due to the increase of the sound samples, so that the paired comparison method is suitable for no more than 16 sound samples. If the number of the sound samples is too large and the evaluation experience of an evaluator is not rich, the four evaluation methods have certain limitations.
Disclosure of Invention
The invention aims to provide a subjective evaluation method for automobile sound quality, which is suitable for subjective evaluation tests with a large number of sound samples and poor evaluation experience of evaluators, can obtain accurate subjective evaluation results in a short time, and provides reasonable and effective data support for subsequent sound quality analysis.
The subjective evaluation method of the automobile sound quality comprises the following steps.
Dividing M sound samples into N groups, wherein each group comprises 4-5 sound samples, two associated sound samples are arranged in each group, space transformation is carried out on the groups and the groups through linear models of the two associated sound samples, the M sound samples are converted into the same space, and K evaluators respectively compare the sound samples in the groups with each other by a grade grading method to obtain grade scores of the sound samples in the groups.
And step two, adopting a system clustering method to cluster evaluators with similar subjective feelings of the sound samples in each group into one group.
Step three, determining a cluster set with the highest contact ratio of the evaluators in the N groups, and setting the number of cluster subsets of the evaluators in the nth group as QnThen there are Q sound samples in the Nth group of space1×Q2...×Qn...×QNA seed level score vector;
the determination of the cluster set with the highest contact ratio specifically comprises the following steps:
in the formula, V*Is a cluster set with the highest contact ratio of evaluators in the N groups, (V)*)nIs a V*The cluster subset used for establishing the linear transformation model in the nth group is V1×Q2...×Qn...×QNA set of clusters of one of (a) or (b),
is the q-th group in the n-th groupnA cluster subset of class evaluators, K is an evaluator number in the cluster set V, and K belongs to {1, 2., K }, count (K) is a statistic of the number of K occurrences in the cluster set V, N*In order to evaluate the lower threshold value of the human overlap times,
to be provided withIn the clustering set V, the number of coincidence is not less than N*Total number of evaluators.
Step four, acquiring a clustering subset (V) used for linear model conversion establishment in the nth group*)nCalculating the grade scores of all the sound samples in the nth group by using an averaging method to obtain grade score vectors s of all the sound samples in the nth groupn。
Step five, grade fraction vectors s of all sound samples in the nth groupnAnd converting the linear model of the grade scores of the two associated sound samples into the space of the Nth group to obtain the subjective evaluation result of the automobile sound quality.
Further, N in the third step*The calculation formula of (A) is as follows:
in the formula, N*E {1, 2.., N }, σ is a lower threshold of the qualification rate of the evaluator, and the value σ is 0.7.
Further, the correlated sound samples in the first step are at a medium sound quality level in each group, and there is a significant sound quality difference between different correlated sound samples.
Further, the linear model conversion in the fifth step is specifically as follows: the nth group and the nth group are respectively provided with two related sound samples l and m according to a calculation formula
N belongs to {1,2,. and N } for conversion; in the formula, SlnAnd SmnRespectively the grade scores obtained by the sound samples l and m after all the sound samples in the nth group are mutually compared, SlNAnd SmNRespectively the grade scores, s, obtained by the sound samples l and m after all the sound samples in the Nth group are compared with each othernThe rank score vectors for all sound samples in the nth set,
and (4) grade score vectors of all sound samples in the nth group space.
Further, the grade scoring method in the first step is to score a certain subjective evaluation attribute of the sound sample into a plurality of grades, and after the evaluator listens to a single sound sample, the evaluator selects a corresponding grade score according to subjective feeling.
The method comprises the steps of dividing sound samples into a plurality of groups, obtaining corresponding subjective evaluation grade scores after the sound samples in the groups are mutually compared based on a grade grading method, then obtaining a clustering subset used for a linear conversion model in each group through clustering of evaluators in each group, calculating the grade scores of the sound samples in each group, and establishing the linear conversion model to obtain a subjective evaluation result. The method is suitable for subjective evaluation experiments with a large number of sound samples and poor evaluation experience of evaluators, can obtain accurate subjective evaluation results in a short time, and provides reasonable and effective data support for objective quantification of subsequent sound quality, establishment of subjective and objective analysis models and the like.
Drawings
FIG. 1 is a schematic diagram of the packet level comparison method of the present invention;
FIG. 2 is a schematic diagram of evaluator clustering based on sound sample rating scores within groups according to the present invention;
FIG. 3 is a schematic diagram of an evaluation experiment of subjective satisfaction of accelerated sound based on a grade comparison method in an embodiment of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and the specific embodiments.
The embodiment I provides a subjective evaluation method for automobile sound quality, which comprises the following steps:
dividing M sound samples into N groups, wherein each group comprises 4-5 sound samples, the subjective evaluation of the sound samples in the group adopts a grade grading method, and different grade grades of the sound samples in the group are given according to a certain subjective evaluation attribute of sound after the sound samples in the group are compared with each other. In order to obtain the comparison grade of the sound samples among the groups, two associated sound samples are arranged in each group, the group and the group are subjected to space transformation through a linear model of the two associated sound samples, and the M sound samples are converted into the same space. The associated sound samples are at a medium sound quality level within each group, avoiding selecting sound samples having a large difference in sound quality from other sound samples as associated sound samples. And the different correlated sound samples have obvious sound quality difference, and the resolution and the accuracy of the space linear conversion model are improved.
Referring to fig. 1, the sound samples i, j in the 1 st group and the sound samples l, M in the nth group are both correlated sound samples, where i, j, l, M ∈ {1, 2. si1And sj1Respectively, the grade scores obtained for sound samples i and j after all the sound samples in group 1 are compared with each other. siNAnd sjNRespectively, the grade scores obtained by the sound samples i and j after all the sound samples in the Nth group are compared with each other. sln,smn,slN,smNBy analogy, N ∈ {1, 2.
The level scores of all the sound samples in the 1 st group are converted into the space of the Nth group through a linear model of the level scores of the associated sound samples i and j, and the conversion formula is as follows:
in the formula, s1The rank score vectors for all sound samples in group 1,
the level score vectors in the nth set of space are all sound samples in the 1 st set.
The level scores of all the sound samples in the nth group are transformed to the space of the nth group by a linear model correlating the level scores of the sound samples l and m, with a transformation similar to that of group 1. Finally, the comparison of the M sound samples in the Nth group is realized.
Step two, adopting a system clustering method to cluster evaluators with similar subjective feelings of the sound samples in each group into one group, referring to fig. 2, according to the 1 st groupThe similarity degree of the subjective feelings of the sound samples is divided into Q by K evaluators1Class QnThe number of subsets of rater clusters in the nth group. (V)1)1Is a cluster subset of the 1 st subjective raters in group 1, (V)qn)nIs the q-th group in the n-th groupnAnd clustering subsets of the subjective appraisers, and the rest in the same way. Within group 1, q1∈{1,2,...,Q1}; within group n, qn∈{1,2,...,Qn}。
Step three, determining a cluster set with the highest contact ratio of the evaluators in the N groups, and setting the number of cluster subsets of the evaluators in the nth group as QnThen there are Q sound samples in the Nth group of space1×Q2...×Qn...×QNA rank score vector is seeded.
The determination of the cluster set with the highest contact ratio specifically comprises the following steps:
in the formula, V*Is a cluster set with the highest contact ratio of evaluators in the N groups, (V)*)nIs a V*The cluster subset used for establishing the linear transformation model in the nth group is V1×Q2...×Qn...×QNA set of clusters of one of (a) or (b),
is the q-th group in the n-th groupnA cluster subset of class evaluators, K is an evaluator number in the cluster set V, and K belongs to {1, 2., K }, count (K) is a statistic of the number of K occurrences in the cluster set V, N*In order to evaluate the lower threshold value of the human overlap times,
for counting the number of coincidence times in the cluster set V is not less than N*Total number of evaluators.
Said N is*Is calculated byComprises the following steps:
in the formula, N*E {1, 2.., N }, σ is a lower threshold of the qualification rate of the evaluator, and the value σ is 0.7.
Step four, acquiring a clustering subset (V) used for linear model conversion establishment in the nth group*)nCalculating the grade scores of all the sound samples in the nth group by using an averaging method to obtain grade score vectors s of all the sound samples in the nth groupn。
Step five, grade fraction vectors s of all sound samples in the nth groupnConverting to the Nth group of spaces through a linear model of the grade scores of two associated sound samples, specifically: the nth group and the nth group are respectively provided with two related sound samples l and m according to a calculation formula
N belongs to {1,2,. and N } for conversion; in the formula, SlnAnd SmnRespectively the grade scores obtained by the sound samples l and m after all the sound samples in the nth group are mutually compared, SlNAnd SmNRespectively the grade scores, s, obtained by the sound samples l and m after all the sound samples in the Nth group are mutually comparednRank score vectors, s, for all sound samples in the nth set* nAnd (4) grade score vectors of all sound samples in the nth group space.
And calculating grade scores of the M sound samples in the Nth group of spaces to obtain a subjective evaluation result of the automobile sound quality, and finishing the subjective evaluation of the automobile sound quality.
In the second embodiment, the subjective satisfaction of evaluating 10 accelerated sound samples is taken as a specific embodiment, a subjective evaluation experiment is designed by adopting a grouping grade comparison method, and a subjective evaluation result is obtained based on the data quantification method, wherein the specific implementation process is as follows.
Step one, adopting a subjective evaluation experiment designed by a grouping grade comparison method. Referring to fig. 3, 10 accelerated sound samples are divided into 3 groups, wherein the 1 st group contains sound samples 1,2, 3, 4, and 5, the 2 nd group contains sound samples 6, 7, 8, 9, and 10, and the 3 rd group contains sound samples 2, 4, 8, and 9. The associated sound samples of group 1 and group 3 are 2 and 4; the associated sound samples for group 2 and group 3 are 8 and 9. 60 drivers engaged in the non-automobile NVH industry participate in the evaluation experiment of the subjective satisfaction degree of the accelerated sound, the sound samples in each group are subjectively evaluated by adopting a grade grading method, and the corresponding subjective satisfaction degree of the sound samples in each group is obtained after the evaluation personnel are mutually compared.
Secondly, clustering the subjective satisfaction degrees of the accelerated sound samples in each group based on the clustering of the evaluators of the subjective satisfaction degrees of the accelerated sound samples in each group, and performing clustering analysis on the subjective satisfaction degrees of the accelerated sound of 60 evaluators by adopting a system clustering method, wherein the evaluators have similar subjective satisfaction degrees on the accelerated sound samples in each group and can be classified into one group; meanwhile, the clustering subset with the number of evaluators not more than 10 is removed, and the clustering results of the evaluators in the 1 st, 2 nd and 3 rd groups are shown in table 4.
TABLE 4 clustering results of evaluators in evaluation test of accelerated sound subjective satisfaction
From Table 4, Q1,Q2,Q3＝2。
Step three, determining clustering subsets used for the linear transformation model in each group, wherein Q is used for determining the clustering subsets1,Q2,Q3As can be seen, there are 8 subjective satisfaction vectors in the 3 rd set of space for 10 accelerated sound samples. In the subjective evaluation test of 10 accelerated sound samples, N is 3.
According to
Can be obtained, N*＝3。
Using calculation formulas
The total number of evaluators overlapping 3 times in 8 cluster sets V was counted, and the results are shown in table 5.
Table 58 Total number of panelists in the cluster set, which were overlapped 3 times
Determined from Table 5 (V)1)1,(V2)2,(V3)3Respectively, a subset of clusters for the linear transformation model within each group.
Step four, after determining clustering subsets used for linear conversion models in each group, calculating subjective satisfaction of sound samples in 3 groups by using an averaging method to obtain s1,s2,s3。
Step five, based on
Completion of s1,s2,s3To
Finally, the subjective satisfaction of 10 accelerated sound samples in the 3 rd group space is obtained, and the results are shown in table 6.
To verify the rationality of the subjective satisfaction obtained with this method, the subjective satisfaction of 10 samples of accelerated sounds obtained using a pairwise comparison method was used, and the results are shown in table 6.
TABLE 6 comparison of subjective evaluation results based on group rank comparison and pair comparison
The correlation of the subjective satisfaction degrees obtained by the two methods is as high as 0.980, and the correctness of the subjective evaluation method is verified. When the pair comparison method is adopted to finish the evaluation experiment of the subjective satisfaction degree of 10 accelerated sounds, a single person needs to consume 1.5 hours, and when the grouping grade comparison method is adopted, a single person only needs to consume 0.5 hours, so that the time of the subjective evaluation experiment is greatly reduced. In conclusion, the subjective evaluation method is suitable for subjective evaluation experiments with a large number of sound samples and poor evaluation experience of evaluators, and can obtain accurate subjective evaluation results in a short time.
Claims (4)
1. A subjective evaluation method for automobile sound quality is characterized by comprising the following steps:
dividing M sound samples into N groups, wherein each group comprises 4-5 sound samples, two associated sound samples are arranged in each group, space transformation is carried out on the groups and the groups through linear models of the two associated sound samples, the M sound samples are converted into the same space, and K evaluators respectively compare the sound samples in the groups with each other by adopting a grade grading method to obtain grade scores of the sound samples in the groups;
step two, adopting a system clustering method to cluster evaluators with similar subjective feelings of the sound samples in each group into one group;
step three, determining a cluster set with the highest contact ratio of the evaluators in the N groups, and setting the number of cluster subsets of the evaluators in the nth group as QnThen there are Q sound samples in the Nth group of space1×Q2...×Qn...×QNA seed level score vector;
the determination of the cluster set with the highest contact ratio specifically comprises the following steps:
in the formula, V*Is a cluster set with the highest contact ratio of evaluators in the N groups, (V)*)nIs a V*The cluster subset used for establishing the linear transformation model in the nth group is V1×Q2...×Qn...×QNOne ofThe collection of the species clusters is performed,
is the q-th group in the n-th groupnA cluster subset of class evaluators, K is an evaluator serial number in the cluster set V, and K belongs to {1, 2., K }, count (K) is a statistic of the occurrence frequency of K in the cluster set V, and N is a statistic of the occurrence frequency of K in the cluster set V*In order to evaluate the lower threshold value of the human overlap times,
for counting the number of coincidence times in the cluster set V is not less than N*Total number of evaluators of (1);
step four, acquiring a clustering subset (V) used for linear model conversion establishment in the nth group*)nCalculating the grade scores of all the sound samples in the nth group by using an averaging method to obtain grade score vectors s of all the sound samples in the nth groupn；
Step five, grade fraction vectors s of all sound samples in the nth groupnConverting the linear models of the grade scores of the two associated sound samples into the Nth group of space to obtain the subjective evaluation result of the automobile sound quality; the method specifically comprises the following steps: the nth group and the nth group are respectively provided with two related sound samples l and m according to a calculation formula
in the formula, SlnAnd SmnRespectively the grade scores obtained by the sound samples l and m after all the sound samples in the nth group are mutually compared, SlNAnd SmNRespectively the grade scores, s, obtained by the sound samples l and m after all the sound samples in the Nth group are compared with each othernThe rank score vectors for all sound samples in the nth set,
grade score direction of all sound samples in the nth group spaceAmount of the compound (A).
2. The method for subjective evaluation of automotive acoustic quality according to claim 1, wherein N in the third step*The calculation formula of (A) is as follows:
in the formula, N*E {1, 2.., N }, σ is a lower threshold of the qualification rate of the evaluator, and the value σ is 0.7.
3. The subjective evaluation method of automotive sound quality according to claim 1 or 2, characterized in that: the associated sound samples in the first step are at a medium sound quality level in each group, and the sound quality difference between different associated sound samples is obvious.
4. The subjective evaluation method of automotive sound quality according to claim 1 or 2, characterized in that: the grade scoring method in the first step is to grade a certain subjective evaluation attribute of the sound sample into a plurality of grades, and after an evaluator listens to a single sound sample, the corresponding grade score is selected according to subjective feeling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910805315.6A CN110530653B (en) | 2019-08-29 | 2019-08-29 | Subjective evaluation method for automobile sound quality |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910805315.6A CN110530653B (en) | 2019-08-29 | 2019-08-29 | Subjective evaluation method for automobile sound quality |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110530653A CN110530653A (en) | 2019-12-03 |
| CN110530653B true CN110530653B (en) | 2021-04-06 |
Family
ID=68665044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910805315.6A Active CN110530653B (en) | 2019-08-29 | 2019-08-29 | Subjective evaluation method for automobile sound quality |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110530653B (en) |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7499856B2 (en) * | 2002-12-25 | 2009-03-03 | Nippon Telegraph And Telephone Corporation | Estimation method and apparatus of overall conversational quality taking into account the interaction between quality factors |
| KR100929589B1 (en) * | 2007-10-31 | 2009-12-03 | 한양대학교 산학협력단 | Sound quality evaluation method using MTS |
| CN102475554B (en) * | 2010-11-24 | 2014-05-28 | 比亚迪股份有限公司 | Method for guiding interior sound package by utilizing sound quality |
| CN102354391B (en) * | 2011-09-26 | 2013-07-24 | 武汉理工大学 | Omnidirectional search mechanism variogram based optimal remote sensing scale selection method |
| CN103471709A (en) * | 2013-09-17 | 2013-12-25 | 吉林大学 | Method for predicting noise quality of noise inside passenger vehicle |
| JP6571623B2 (en) * | 2016-10-12 | 2019-09-04 | 日本電信電話株式会社 | Sound quality evaluation apparatus, sound quality evaluation method, and program |
| CN107832483A (en) * | 2017-08-30 | 2018-03-23 | 山东赛马力发电设备有限公司 | Diesel engine noise method for evaluating quality based on UDS and AHP |
| CN109141623B (en) * | 2018-08-14 | 2020-11-20 | 中车青岛四方机车车辆股份有限公司 | Method and device for evaluating sound quality in train |
| CN109443792A (en) * | 2018-10-12 | 2019-03-08 | 安徽江淮汽车集团股份有限公司 | A kind of automobile drives at a constant speed the evaluation method of sound quality |
-
2019
- 2019-08-29 CN CN201910805315.6A patent/CN110530653B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110530653A (en) | 2019-12-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103440873B (en) | A kind of music recommend method based on similarity | |
| CN102682760B (en) | Overlapped voice detection method and system | |
| CN107292330A (en) | A kind of iterative label Noise Identification algorithm based on supervised learning and semi-supervised learning double-point information | |
| CN102262642B (en) | Web image search engine and realizing method thereof | |
| CN104240719B (en) | The feature extracting method of audio, the sorting technique of audio and relevant apparatus | |
| Nieto et al. | Perceptual evaluation of automatically extracted musical motives | |
| CN109255029A (en) | A method of automatic Bug report distribution is enhanced using weighted optimization training set | |
| CN109461458A (en) | A kind of audio method for detecting abnormality based on generation confrontation network | |
| CN107944487B (en) | Crop breeding variety recommendation method based on mixed collaborative filtering algorithm | |
| CN106951728B (en) | Tumor key gene identification method based on particle swarm optimization and scoring criterion | |
| CN103810210B (en) | Search result display methods and device | |
| CN110530653B (en) | Subjective evaluation method for automobile sound quality | |
| CN106251861A (en) | A kind of abnormal sound in public places detection method based on scene modeling | |
| CN111863135B (en) | False positive structure variation filtering method, storage medium and computing device | |
| Cervi et al. | Comparing the reputation of researchers using a profile model and scientific metrics | |
| CN110010204A (en) | Prognosis biomarker recognition methods based on converged network and more marking strategies | |
| CN111611781B (en) | Data labeling method, question answering device and electronic equipment | |
| CN104318255A (en) | Gender identifying method and system for mobile shooting type APP user | |
| CN111368131A (en) | User relationship identification method and device, electronic equipment and storage medium | |
| CN113344031B (en) | Text classification method | |
| CN112738724B (en) | Method, device, equipment and medium for accurately identifying regional target crowd | |
| CN114822557A (en) | Method, device, equipment and storage medium for distinguishing different sounds in classroom | |
| CN111309608B (en) | Test case selection method and device, electronic equipment and readable storage medium | |
| CN111538669B (en) | Test case extraction method and device based on historical problem backtracking analysis | |
| CN113889274A (en) | Method and device for constructing risk prediction model of autism spectrum disorder |
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 |