Information processing device and program

Abstract

To enable the evaluation of a hitting feeling of a golf ball without requiring a hitting feeling evaluation by a tester.SOLUTION: An information processing device for estimating a hitting feeling of a golf ball includes feature amount acquisition means for acquiring a feature amount of a hitting sound with an estimation object ball as a sound source among hitting sounds of the estimation object ball, and estimation means for estimating the hitting feeling of the estimation object ball on the basis of correlation information showing a correlation between hitting feelings sensed by the tester in hitting each of a plurality of kinds of test balls and a feature amount of a hitting sound with the test ball as a sound source, and the feature amount acquired by the feature amount acquisition means.SELECTED DRAWING: Figure 10

Description

[Technical field]

[0001] The present invention relates to a technique for estimating the hitting feeling that a golfer feels when hitting a golf ball. [Background technology]

[0002] 2. Description of the Related Art There have been proposed techniques for analyzing the impact sound of a golf club hitting a golf ball and utilizing the analysis results in the development of golf clubs (Patent Documents 1 to 4). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2006-239132 A [Patent Document 2] JP 2003-325707 A [Patent Document 3] JP 2001-314534 A [Patent Document 4] Patent No. 3996517 Summary of the Invention [Problem to be solved by the invention]

[0004] One of the factors that golfers consider when selecting a golf ball to use from among various types of golf balls is the feel of the ball. Conventional evaluation of feel of a ball has relied on the impressions (sensory evaluation) of testers such as professional golfers who actually test-hit golf balls. In the golf ball development stage, in order to evaluate the feel of golf balls with various specifications, testers had to test the balls each time, leaving room for improvement in development efficiency.

[0005] An object of the present invention is to provide a technique that enables evaluation of the hit feel of a golf ball without the need for a tester. [Means for solving the problem]

[0006] According to the present invention, for example, An information processing device that estimates a hitting feeling of a golf ball, a feature amount acquiring means for acquiring a feature amount of an impact sound having the estimation target ball as a sound source, out of the impact sounds of the estimation target ball; and an estimation means for estimating the hitting feeling of the estimation target ball based on correlation information indicating a correlation between the hitting feeling felt by a tester when hitting each of a plurality of types of test balls and a feature amount of the hitting sound generated by the test ball as a sound source, and the feature amount acquired by the feature amount acquisition means. picture, the impact sound generated by the test ball as a sound source is identified based on vibration characteristic information indicating vibration characteristics of the test ball, the information being obtained through a prior vibration experiment; The present invention provides an information processing apparatus comprising: Effect of the Invention

[0007] According to the present invention, it is possible to provide a technique that enables evaluation of the feel of a golf ball without the need for evaluation of the feel by a tester. [Brief description of the drawings]

[0008] [Figure 1] 1 is a schematic diagram of an information processing system according to an embodiment of the present invention. [Diagram 2] 1A is a flowchart showing an example of a method for deriving correlation information, and FIG. 1B is a flowchart showing an example of a method for estimating a hit feel. [Diagram 3] FIG. 13 is a diagram showing an example of measurement of vibration characteristics of a ball. [Figure 4] 4A and 4B are diagrams showing an example of a measurement mode of an impact sound. [Diagram 5] FIG. 11 is a diagram showing an analysis example of an impact sound. [Figure 6] FIG. 13 is a diagram showing an example of quantifying a hit feeling. [Figure 7] FIG. 13 is a diagram showing an example of learning data. [Figure 8] A diagram showing an example of machine learning. [Figure 9] 11 is a flowchart showing an example of processing by an information processing device. [Figure 10] 11 is a flowchart showing an example of processing by an information processing device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, the embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the invention according to the claims, and not all combinations of features described in the embodiments are essential to the invention. Two or more features among the multiple features described in the embodiments may be arbitrarily combined. In addition, the same reference numbers are used for the same or similar configurations, and duplicated descriptions are omitted.

[0010] <System configuration> FIG. 1 is a schematic diagram of a system A using an information processing device 1 according to an embodiment of the present invention. The system A includes the information processing device 1, a microphone 2, and a signal processing device. Science Department 3, a display device 4, and an input device 5.

[0011] The microphone 2 is a device that collects the impact sound when a golfer or a swing robot hits a golf ball with a golf club. The signal processing unit 3 is an electric circuit that processes the audio signal collected by the microphone 2, and includes, for example, a circuit that converts an analog signal to a digital signal, a filter circuit that removes noise, etc. Time series data of the impact sound can be obtained from the microphone 2 and the signal processing unit 3.

[0012] The information processing device 1 is a computer that estimates the hitting feeling of the golf ball that the golfer feels when hitting the ball from the impact sound. The information processing device 1 includes a processing unit 11, a storage unit 12, and an I / F unit (interface unit) 13 that are electrically connected to each other. The processing unit 11 is a processor such as a CPU. The storage unit 12 includes one or more storage devices. The storage device is, for example, a RAM, a ROM, a hard disk, etc. The storage unit 12 stores programs executed by the processing unit 11 and various data. The storage unit 12 includes a database 12a that stores correlation information, learning data, audio signal data collected by the microphone 2, etc., which will be described later. The program executed by the processing unit 11 can be composed of multiple instructions that can be read by the processing unit 11. The program executed by the processing unit 11 can also be stored in a storage medium such as a CD-ROM or DVD, and installed in the information processing device 1.

[0013] The I / F unit 13 inputs and outputs data between an external device and the processing unit 11. The I / F unit 13 may include an input / output interface and a communication interface. The signal processing unit 3 is connected to the I / F unit 13, and the measurement result of the impact sound picked up by the microphone 2 is acquired by the information processing device 1.

[0014] A display device 4 and an input device 5 are connected to the information processing device 1. The display device 4 is, for example, an electronic image display device such as a liquid crystal display device, and displays the processing results (e.g., the estimated hit feel results) of the information processing device 1. The input device 5 is a mouse and a keyboard, and accepts data input and operation instructions for the information processing device 1.

[0015] <Flow of impact feeling estimation> In this embodiment, the feel of a golf ball is estimated from the impact sound. The impact feel is estimated based on correlation information that indicates the correlation between the impact sound (particularly the feature amount of the impact sound generated by the golf ball as the sound source) and the sensory evaluation of the impact feel. To distinguish the golf ball used in generating the correlation information from the golf ball that is the subject of the impact feel estimation, the former is conveniently called a test ball and the latter is called an estimation subject ball.

[0016] A rough processing flow will be described with reference to Fig. 2(A) and Fig. 2(B). An example of the processing executed by the information processing device 1 will be described later with reference to Fig. 9 and Fig. 10. The processing is roughly divided into a process of deriving correlation information and a process of estimating the hit feeling of the estimation target ball using the derived correlation information.

[0017] <Deriving correlation information> First, the process of deriving correlation information will be described with reference to Fig. 2(A), which is a flowchart showing an example of a method of deriving correlation information.

[0018] In S1, the vibration characteristics of the test ball are measured to obtain vibration characteristic information. In addition to the sound generated by the golf ball as a sound source, the impact sound includes the sound generated by the golf club head as a sound source and environmental sounds. In order to extract the sound generated by the golf ball as a sound source from the impact sound, the vibration characteristics of the test ball are measured in this embodiment. FIG. 3 is a diagram showing an example of the measurement of the vibration characteristics of the ball, showing the actual measured values ​​of a general ball. The golf ball 100 is mounted on a vibration measuring device 6, and the response characteristics to the excitation are measured. Although not shown, the vibration measuring device includes an acceleration sensor, a force sensor, and an excitation device. By eigenvalue analysis of the measurement results, the natural frequency of the golf ball 100 in the vibration mode of multiple orders is obtained as vibration characteristic information. This measurement of the vibration characteristics is performed for each of the multiple types of golf balls used as the test ball.

[0019] In the illustrated example, the natural frequencies of the first vibration mode to the fourth vibration mode are obtained. The natural frequencies used are not limited to the first vibration mode to the fourth vibration mode as in the present embodiment, but may be the first vibration mode to the second vibration mode, or the first vibration mode to the fifth vibration mode or higher. However, in terms of the relationship between the impact sound and the impact feeling of the golf ball and the efficiency of calculation, the order of the vibration mode may be selected so that the natural frequency falls within the range of 1000 Hz to 12000 Hz. The vibration characteristics of the test ball may be analyzed by a simulation such as a finite element analysis using a physical model of the test ball.

[0020] Returning to FIG. 2(A), in S2, a tester hits a test ball and the impact sound is measured by microphone 2. The tester is, for example, a professional golfer. The test hit may be measured in an environment close to an actual game (a golf course or an environment close to it) in that it involves a sensory evaluation of the hitting feeling.

[0021] 4(A) shows an example of how to measure an impact sound. In the illustrated example, a tester 110 hits a golf ball 100, which is a test ball, with a golf club 101. The microphone 2 is, for example, a small microphone, and is provided near the ear of the tester 110 so that the impact sound heard by the tester 110 and the impact sound measured by the microphone 2 are as close as possible. The microphone 2 is attached to the ear of the tester 110 using, for example, a clip. This impact sound measurement is also performed for each of multiple types of golf balls used as test balls, and the measurement results are stored in association with individual information identifying each test ball.

[0022] Returning to Fig. 2(A), in S3, the impact sound measured in S2 is analyzed and the sensory evaluation (impact feeling information) of the impact feeling felt by the tester is collected. In the impact sound analysis, the feature quantity of the impact sound whose sound source is the test ball is extracted. Fig. 5 shows an example of the analysis.

[0023] The audio data D1 indicates audio signal data (temporal change in sound pressure) of the impact sound measured by the microphone 2 and processed by the signal processor 3. By performing a fast Fourier transform (FFT) on this audio signal data, frequency characteristic data D2 indicating the relationship between the frequency and sound pressure (sound pressure level) of the impact sound is obtained. As described above, the impact sound includes sounds generated by the golf club head as a sound source and environmental sounds. By using the vibration characteristics (natural frequency) of the test ball measured by S1, it is possible to extract the feature quantity of the impact sound generated by the test ball as a sound source from the frequency characteristic data D2.

[0024] A specific example will be described. The natural frequencies of the first vibration mode to the fourth vibration mode of the test ball are N1 to N4, respectively. Applying the example of Fig. 3, N1 = 3184 Hz, N2 = 4937 Hz, N3 = 6851 Hz, and N4 = 8698 Hz. Of the frequency characteristic data D2, the frequencies corresponding to N1 to N4 can be said to be characteristic frequencies of the impact sound generated by using the test ball as the sound source.

[0025] However, since the struck ball, which is the sound source, moves while vibrating, the Doppler effect does not necessarily mean that the frequency of the impact sound that reaches the microphone matches the measurement result of the vibration characteristics of the test ball S1 exactly. Therefore, for N1 to N4 of the test ball, the frequency that has a sound pressure peak within a predetermined frequency range is set as the characteristic frequency of the impact sound with the test ball as the sound source. In the example of FIG. 5, the frequency that has a sound pressure peak within a range of ±200 Hz centered on N1 to N4 is set as the characteristic frequency of the impact sound with the test ball as the sound source. For example, if N1=3184 Hz, the frequency that has a sound pressure peak within a range of 2984 Hz to 3384 Hz (F1 in the example of FIG. 5) is set as the characteristic frequency of the impact sound with the test ball as the sound source. In the example of FIG. 5, the characteristic frequencies of F1 to F4 are specified for N1 to N4 of the test balls. The characteristic frequencies F1 to F4 and their respective sound pressures (characteristic sound pressures) are set as the characteristic quantities of the impact sound with the test ball as the sound source. The characteristic sound pressure may be a peak value or an average value in the above-mentioned predetermined frequency range (range of ±200 Hz) (for example, the sound pressure corresponding to frequency F1 is the average value of sound pressure in the range of 2984 Hz to 3384 Hz).

[0026] Furthermore, data D3 on the characteristic frequency of the impact sound and the vibration decay time (for example, the time it takes for the sound pressure to drop to 10% from the peak sound pressure) is obtained by performing a wavelet transform on the sound signal data D1. The decay times of the characteristic frequencies F1 to F4 are obtained and are used as the characteristic quantities (characteristic decay times) of the impact sounds generated by the test ball as the sound source.

[0027] As described above, in this embodiment, the characteristic frequencies F1 to F4 of the impact sounds corresponding to the natural frequencies N1 to N4 of the test ball, the characteristic sound pressures (sound pressure levels) at the characteristic frequencies F1 to F4, and the characteristic decay times of the characteristic frequencies F1 to F4 are used as the characteristic quantities of the impact sounds produced by the test ball as the sound source. These characteristic quantities are stored in association with individual information that identifies each individual test ball. Note that, although these three quantities are used as the characteristic quantities of the impact sounds produced by the test ball as the sound source in this embodiment, one or two of the three quantities may be used.

[0028] Fig. 6 shows an example of the types of hit feel (evaluation axes) and their numerical values. In the illustrated example, two evaluation axes of the hit feel, namely the hardness and weight of the golf ball, are shown. However, the evaluation axes of the hit feel may be one type or three or more types.

[0029] The hardness is an evaluation axis relating to whether the tester felt the golf ball as being hard or soft when hitting the ball, and the weight is an evaluation axis relating to whether the tester felt the weight of the golf ball as being heavy or light when hitting the ball.

[0030] Each type is quantified on an 11-point scale. 2 The tester reports the score of the type of hit feel, which is stored in association with the type of test ball as hit feel information. A "0" for the test ball means an average feel. The tester may first try a golf ball that will be the standard for the hit feel evaluation (value 0), and give it a score in comparison with the standard ball.

[0031] Returning to FIG. 2(A), in S4, correlation information is generated using the feature amount of the impact sound generated by using the test ball in S3 as the sound source and the hitting feeling information. The correlation information may be in the form of a calculation formula or a classification rule. Here, an example of deriving the correlation information as a calculation formula will be described, and in particular, an example of deriving the calculation formula by machine learning will be described.

[0032] FIG. 7 shows an example of learning data used in machine learning. The learning data includes feature values ​​and hitting feeling information for each test ball. The feature values ​​are feature values ​​of the impact sound with the test ball as the sound source, and are the characteristic vibration frequencies (F1 to F4 in FIG. 5) of the impact sound corresponding to the first vibration mode to the fourth vibration mode of the test ball, respectively, and the characteristic sound pressure and characteristic decay time at the characteristic vibration frequencies (D3 in FIG. 5). The hitting feeling information is each score of hardness and weight. In the illustrated example, one set of learning data is illustrated for one ball (for example, ball a), but if multiple testers participate in the test hit, it is possible to obtain the same number of sets of learning data as the number of testers.

[0033] FIG. 8 is a conceptual diagram showing an input / output structure in machine learning. Input data X (71) is data in the input layer of a learning model 72. Output data Y (73) is output as a result of recognizing the input data X using the learning model 72, which is a machine learning model. During learning, training data T (74) is given as correct answer data for the recognition result of the input data X, so that the output data Y and training data T are given to a loss function 75 to obtain the deviation amount L (76) of the recognition result from the correct answer. The learning model 72 is optimized by updating the coefficients, weighting, etc. of the learning model 72 so that the deviation amount L is small for a large amount of training data. The feature amount data of FIG. 7 is used as the input data X, and the hitting feeling information data of FIG. 7 is used as the training data T.

[0034] The learning model 72 may be, for example, in the form of a linear equation: Hit feel: Hardness = a1 x primary characteristic frequency + b1 x primary characteristic sound pressure + c1 x primary characteristic decay time + a2 x secondary characteristic frequency + b2 x secondary characteristic sound pressure + c2 x secondary characteristic decay time + a3 x tertiary characteristic frequency + b3 x tertiary characteristic sound pressure + c3 x tertiary characteristic decay time + a4 x quaternary characteristic frequency + b4 x quaternary characteristic sound pressure + c4 x quaternary characteristic decay time The coefficients a1 to a4, b1 to b4, and c1 to c4 are optimized by machine learning. The same applies to the weight of the hitting feeling. Note that the learning model 72 may be a model using a neural network instead of the above.

[0035] As a result of the above machine learning, an equation is obtained that shows the correlation between the feature amount of the impact sound generated from the test ball as the sound source and the hit feel. As a result of the learning, terms whose coefficients are substantially zero may be deleted from the equation. According to the inventor's experiments, among the hit feel, hardness is relatively heavily influenced by the first-order characteristic frequency, characteristic sound pressure, and characteristic decay time, and weight is relatively heavily influenced by the second-order characteristic frequency, characteristic sound pressure, and characteristic decay time.

[0036] <Estimation of hitting sensation> Next, a process of estimating the hit feel of the estimation target ball from the impact sound using correlation information will be described with reference to Fig. 2(B), which is a flowchart showing an example of a hit feel estimation method.

[0037] In S5, the vibration characteristics of the estimation target ball are measured to obtain vibration characteristic information. This is the same process as S1, and the vibration characteristics of the estimation target ball (in this embodiment, each natural frequency of the first vibration mode to the fourth vibration mode) required for using the correlation information are measured. The vibration characteristics are measured for each golf ball to be estimated.

[0038] In S6, the estimation target ball is test-hit, and the impact sound is measured by the microphone 2. The test hit may be performed by a human (tester 110) as shown in FIG. 4(A) or by a swing robot 111 as shown in FIG. 4(B). At the stage of estimating the impact feeling, since a sensory evaluation of the impact feeling is not required in the test hit, it is efficient to perform the test hit with the swing robot 111. Even if a human performs the test hit, it is not necessary for a person with excellent sensitivity to the impact feeling, such as a professional golfer, to perform the test hit. Since it is sufficient to measure the impact sound, the test hit may be performed by a beginner to intermediate level golfer. The test hit may be measured in an environment (anechoic or semi-anechoic room) where only the impact sound can be measured more clearly, since no sensory evaluation of the impact feeling is required.

[0039] The swing robot 111 may be of a type in which a driving source such as a motor rotates an arm that holds the golf club 101, or of a type without a driving source in which the arm is manually lifted by the user to the top position and falls naturally (rotates naturally) by gravity. In the former type having a driving source, the driving sound may be easily included in the impact sound. Therefore, the latter type not having a driving source is advantageous in that less environmental sound is included in the impact sound. In addition, the surface of the swing robot 111 may be covered with a sound-absorbing material. This makes it possible to prevent the operating sound of the swing robot 111 from being included in the impact sound.

[0040] When using a swing robot 111 that does not have a driving source, the impact sound may be small. Therefore, as shown by the solid line in Fig. 4(B), the microphone 2 may be placed near the golf ball 100 (near the impact position, for example, within a few meters).

[0041] Returning to FIG. 2(B), in S7, the impact sound measured in S6 is analyzed. This is the same as the content of S3 and FIG. 5 above. As a result, the first to fourth characteristic vibration frequencies of the impact sound corresponding to the respective natural frequencies of the first to fourth vibration modes of the estimation target ball are obtained as the characteristic quantities of the impact sound whose sound source is the estimation target ball, each characteristic sound pressure (sound pressure level), and each characteristic decay time are obtained. In S8, the characteristic quantities of the impact sound whose sound source is the estimation target ball obtained in S7 are substituted into the calculation formula which is the correlation information derived in S4, and an estimation result of the hit feeling (hardness, weight) can be obtained.

[0042] As described above, in this embodiment, although the evaluation of the impact feeling by a tester such as a professional golfer is necessary for deriving the correlation information, if the impact sound can be measured in the subsequent evaluation of the impact feeling of the golf ball, the evaluation of the impact feeling by a tester is not necessary. Therefore, the development of golf balls can be carried out more efficiently. In addition, the relationship between the impact sound and the impact feeling, which was previously unknown, can be clarified.

[0043] According to the inventor's experiment, in a golf club having a head with a high hollowness such as a driver head, the impact sound tends to be dominated by the head rather than the golf ball. On the other hand, in a golf club having a solid or low hollowness head such as an iron head, particularly a wedge or putter head, the impact sound tends to show the characteristics of the golf ball relatively prominently. Therefore, the method of the present embodiment is particularly effective for evaluating the impact feeling of a golf ball when hitting a golf ball with an iron type club (particularly a wedge) or a putter club, and is even more effective because such golf clubs tend to prioritize the impact feeling of the golf ball. It is not necessarily necessary to use the same club or the same type of club (number, wood type, iron type, hollow, solid, loft angle, material, etc.) as the golf club in the test hit in S2 and the test hit in S6, but using the same club or the same type of club allows for more accurate impact feeling estimation.

[0044] <Example of processing by information processing device> The method of deriving correlation information in Fig. 2(A) and the method of estimating hit feel in Fig. 2(B) can be automated in whole or in part by the information processing device 1. An example of this will be described with reference to Fig. 9 and Fig. 10. Fig. 9 is a flowchart showing an example of processing by the processing unit 11 regarding derivation of correlation information, and Fig. 10 is a flowchart showing an example of processing by the processing unit 11 regarding hit feel estimation.

[0045] The description will begin with Fig. 9. In S11, input of vibration characteristic information of a test ball that has been measured in advance is accepted. The vibration characteristic information can be input by an operator using input device 5. The input content is stored in database 12a in association with individual information that identifies each individual test ball.

[0046] In S12, the impact sound is measured and stored. The tester hits the test ball 100 in the manner shown in FIG. 4(A), and the impact sound is measured by the microphone 2 and processed by the signal processing unit 3. The audio data (D1 in FIG. 5) is then stored in the database 12a in association with individual information identifying each test ball. In S13, input of impact feel information is accepted. The impact feel information can be a score of the impact feel (hardness, weight) heard from the tester, which the operator inputs using the input device 5. The input content is stored in the database 12a in association with individual information identifying each test ball.

[0047] In S14, a feature quantity of an impact sound having the test ball as a sound source is extracted from the sound data of the impact sound stored in S12. Here, the sound data is read from database 12a, and frequency characteristic data (D2 in FIG. 5) is generated by FFT. Furthermore, vibration characteristic information of the test ball is read from database 12a, and from the frequency characteristic data, characteristic frequencies (F1 to F4 in FIG. 5) of the impact sound corresponding to the first vibration mode to the fourth vibration mode of the test ball, respectively, and characteristic sound pressure at those frequencies are identified as feature quantities of the impact sound having the test ball as a sound source. In addition, a wavelet transform is performed on the sound data read from database 12a, and characteristic decay times (D3 in FIG. 5) of each characteristic frequency (F1 to F4 in FIG. 5) are identified as feature quantities of the impact sound having the test ball as a sound source. The extracted feature quantities are stored in database 12a in association with individual information identifying each test ball.

[0048] In S15, the hit feel information saved in S13 and the feature amount extracted in S14 are read out from the database 12a to obtain them, and learning data (FIG. 6) is generated from the obtained hit feel information and feature amount. In S16, an arithmetic expression is derived as correlation information by machine learning. In S17, the correlation information derived in S16 is saved in the database 12a. This completes the process of deriving correlation information.

[0049] The process of estimating the impact feeling in FIG. 10 will be described. In S21, an estimated target object measured in advance is Elephant BoThe vibration characteristic information of the ball is inputted by an operator using the input device 5. The inputted contents are stored in the database 12a in association with individual information that identifies each ball to be estimated.

[0050] In S22, the impact sound is measured and stored. Here, the same process as in S12 is performed. That is, the test ball 100 is hit in the manner shown in FIG. 4(A) or FIG. 4(B). After the impact sound is measured by the microphone 2 and processed by the signal processing unit 3, the resulting sound data (similar to D1 in FIG. 5) is stored in the database 12a in association with individual information that identifies each estimation target ball.

[0051] In S23, feature quantities of the impact sound whose sound source is the estimation target ball are extracted from the audio data of the impact sound saved in S22. Here, the same process as in S14 is performed. That is, the audio data is read from database 12a, and frequency characteristic data (similar to D2 in FIG. 5) is generated by FFT. Furthermore, vibration characteristic information of the estimation target ball is read from database 12a, and from the frequency characteristic data, the characteristic frequencies (similar to F1 to F4 in FIG. 5) of the impact sound corresponding respectively to the first vibration mode to the fourth vibration mode of the test ball, and the sound pressure at the characteristic frequencies are identified as feature quantities of the impact sound whose sound source is the estimation target ball. In addition, a wavelet transform is performed on the audio data read from database 12a, Estimated target The characteristic decay time (similar to D3 in FIG. 5) of each characteristic frequency is identified as a characteristic amount of the impact sound generated by the ball as the sound source. The extracted characteristic amount is stored in database 12a in association with individual information that identifies each estimation target ball.

[0052] In S24, the feature amount saved in S23 is read out from the database 12a to be acquired. In S25, the hit feel of the estimation target ball is estimated from the feature amount acquired in S24 and the correlation information read out from the database 12a. In S26, the estimation result of S25 is displayed on the display device 4. This completes the process related to hit feel estimation. In the example of FIG. 10, when the measurement of the impact sound is started in S22, the process up to the output of the estimation result in S26 is performed automatically. However, for example, in the process of acquiring the feature amount in S24, the feature amount calculated separately may be acquired by accepting an input by the operator, in which case the operator can input the feature amount using the input device 5.

[0053] Although the embodiment of the invention has been described above, the invention is not limited to the above embodiment, and various modifications and variations are possible within the scope of the gist of the invention. [Explanation of symbols]

[0054] 1. Information processing device

Claims

1. An information processing device that estimates a hitting feeling of a golf ball, a feature amount acquiring means for acquiring a feature amount of an impact sound having the estimation target ball as a sound source, out of the impact sounds of the estimation target ball; an estimation means for estimating a hit feeling of the estimation target ball based on correlation information indicating a correlation between a hit feeling felt by a tester when hitting each of a plurality of types of test balls and a feature amount of an impact sound generated from the test ball as a sound source, and the feature amount acquired by the feature amount acquisition means, the impact sound generated by the test ball as a sound source is identified based on vibration characteristic information indicating vibration characteristics of the test ball, the information being obtained through a prior vibration experiment; 23. An information processing apparatus comprising:

2. 2. The information processing device according to claim 1, the feature amount of the impact sound generated by the estimation target ball as a sound source is identified based on vibration characteristic information indicating vibration characteristics of the estimation target ball, the vibration characteristic information being obtained by a prior vibration experiment; 23. An information processing apparatus comprising:

3. 3. The information processing device according to claim 2, the vibration characteristic information indicating the vibration characteristics of the estimation target ball includes information on each natural frequency of a plurality of vibration modes of the estimation target ball, the feature amount of the impact sound generated by the estimation target ball as a sound source includes a frequency having a sound pressure peak within a predetermined frequency range for each of the natural frequencies; 23. An information processing apparatus comprising:

4. 4. The information processing device according to claim 3, the feature amount of the impact sound generated by the estimation target ball as a sound source includes a sound pressure of the vibration frequency having a peak sound pressure within the predetermined vibration frequency range and a decay time of the sound pressure, 23. An information processing apparatus comprising:

5. 2. The information processing device according to claim 1, an impact sound acquisition means for acquiring a measurement result of the impact sound of the estimation target ball; an extraction means for extracting the feature amount of the impact sound generated by the estimation target ball as a sound source based on the measurement result acquired by the impact sound acquisition means and vibration characteristic information of the estimation target ball obtained by a prior vibration experiment, The feature acquisition means acquires the feature extracted by the extraction means.

23. An information processing apparatus comprising:

6. 2. The information processing device according to claim 1, The correlation information is an arithmetic expression having the feature amount of an impact sound generated by the estimation target ball as a sound source as a variable and the level of the hit feeling as a solution, 23. An information processing apparatus comprising:

7. 2. The information processing device according to claim 1, The hit feeling is specified by a plurality of types of evaluation axes, One of the multiple evaluation axes is the hardness of the golf ball felt by the hitter when hitting the ball.

23. An information processing apparatus comprising:

8. 2. The information processing device according to claim 1, The impact sound of the estimation target ball is an impact sound generated when the estimation target ball is hit by a swing robot.

23. An information processing apparatus comprising:

9. 2. The information processing device according to claim 1, The impact sound of the estimation target ball is an impact sound when the estimation target ball is hit with an iron club or a putter club.

23. An information processing apparatus comprising:

10. a feature acquisition means for acquiring a feature of an impact sound generated by the test ball as a sound source from among impact sounds of each of a plurality of types of test balls; a hit feeling information acquiring means for acquiring information on a hit feeling felt by a tester when hitting each of the plurality of types of test balls; a derivation means for deriving an arithmetic expression indicating a correlation between the feature amount and the hit feeling by machine learning using the feature amount acquired by the feature amount acquisition means and the information acquired by the hit feeling information acquisition means as teacher data, the impact sound generated by the test ball as a sound source is identified based on vibration characteristic information indicating vibration characteristics of the test ball, the information being obtained through a prior vibration experiment; 23. An information processing apparatus comprising:

11. A computer is provided to estimate the feel of a golf ball. a feature amount acquisition means for acquiring a feature amount of an impact sound having the estimation target ball as a sound source, out of the impact sounds of the estimation target ball; an estimation means for estimating the hit feeling of the estimation target ball based on correlation information indicating a correlation between the hit feeling felt by a tester when hitting each of a plurality of types of test balls and a feature amount of an impact sound generated from the test ball as a sound source, and the feature amount acquired by the feature amount acquisition means; It is a program that functions as the impact sound generated by the test ball as a sound source is identified based on vibration characteristic information indicating vibration characteristics of the test ball, the information being obtained through a prior vibration experiment; A program characterized by:

12. Computer, a feature acquisition means for acquiring a feature of an impact sound generated by the test ball as a sound source from among impact sounds of each of a plurality of types of test balls; a hit feeling information acquiring means for acquiring information on the hit feeling felt by a tester when hitting each of the plurality of types of test balls; a derivation means for deriving an arithmetic expression indicating a correlation between the feature amount and the hit feeling by machine learning using the feature amount acquired by the feature amount acquisition means and the information acquired by the hit feeling information acquisition means as teacher data; It is a program that functions as the impact sound generated by the test ball as a sound source is identified based on vibration characteristic information indicating vibration characteristics of the test ball, the information being obtained through a prior vibration experiment; A program characterized by:

Images