Title of Invention: Otoacoustic Authentication Device, Otoacoustic Authentication Method, and Recording Medium
Technical field
[0001]
TECHNICAL FIELD The present invention relates to an otoacoustic authentication device, an otoacoustic authentication method, and a recording medium. The present invention relates to an otoacoustic authentication device, an otoacoustic authentication method, and a recording medium for authenticating a subject.
Background technology
[0002]
For example, fingerprint authentication, vein authentication, face authentication, iris authentication, and voice authentication are known as personal authentication technologies (called biometric authentication technologies) based on the personal characteristics of living organisms. Among personal authentication technologies, otoacoustic authentication, in particular, focuses on the individual characteristics of the internal structure of the human ear canal. In otoacoustic authentication, a test signal is input into the ear canal of an individual to be authenticated, and an echo signal based on the echo sound from the ear canal is used to authenticate the individual.
[0003]
　The individual who is the target of personal authentication (person to be authenticated) wears an earphone-shaped device (called an earphone-type device or a hearable device) with a built-in speaker and microphone on the auricle. A speaker of the earphone type device emits a test signal (sound wave) into the ear canal of the person to be authenticated. The earbud device's microphone detects echoes from the ear canal. Then, an echo signal based on the echo sound is transmitted from the earphone type device to the personal authentication device. The personal authentication device performs personal authentication by comparing one or more personal characteristics registered in advance with characteristics extracted from echo signals received from the earphone-type device.
[0004]
Otoacoustic authentication technology can complete personal authentication instantly and stably, and can perform personal authentication instantly while wearing an earphone-type device, even when the individual is moving or working. hands-free), and high secrecy regarding the internal structure of the human ear canal.
prior art documents
patent literature
[0005]
Patent Document 1: International Publication No. 2018/198310
SUMMARY OF THE INVENTION
Problems to be Solved by the Invention
[0006]
Strictly speaking, the echo signal, which is the object of measurement in ear acoustic authentication, depends not only on the acoustic characteristics of the subject's ear canal, but also on the acoustic characteristics of the earphone-type device (called an audio device) used for authentication. Therefore, depending on which audio device is used, there is a possibility that the accuracy of authenticating the subject may vary.
[0007]
The present invention has been made in view of the above problems, and an object of the present invention is to provide an otoacoustic device capable of accurately authenticating a subject regardless of which audio device is used for otoacoustic authentication. An authentication device, an otoacoustic authentication method, and a recording medium are provided.
Means to solve problems
[0008]
An otoacoustic authentication apparatus according to an aspect of the present invention uses an audio device to input a test signal to an ear of a subject, and when an echo signal is received from the subject, the voice device and the subject feature extracting means for extracting from the echo signal a first feature relating to a system consisting of the ears of the human ear; and correcting the first feature to a second feature if the audio device were a predetermined reference device. A correcting means, and an authenticating means for authenticating the subject by collating the second characteristics with characteristics indicated in authentication information registered in advance.
[0009]
An otoacoustic authentication method according to an aspect of the present invention uses an audio device to input a test signal to an ear of a subject, and when an echo signal is received from the subject, the audio device and the subject from the echo signal, correcting the first feature to a second feature if the audio device were a predetermined reference device; and authenticating the subject by matching the features of the above with features indicated in pre-registered authentication information.
[0010]
A recording medium according to an aspect of the present invention uses an audio device to input a test signal to an ear of a subject, and when an echo signal is received from the subject, the audio device and the ear of the subject are detected. extracting from the reverberant signal a first feature for a system consisting of: correcting the first feature to a second feature if the audio device were a predetermined reference device; and It stores a program for causing a computer to authenticate the subject by collating the second feature with the feature indicated in the pre-registered authentication information.
Effect of the invention
[0011]
According to one aspect of the present invention, it is possible to accurately authenticate a subject regardless of which audio device is used for ear acoustic authentication.
Brief description of the drawing
[0012]
1 is a block diagram showing the configuration of an ear acoustic authentication device according to Embodiment 1. FIG.
2 is a flow chart showing the flow of processing executed by the ear acoustic authentication device according to Embodiment 1. FIG.
3 is a diagram schematically showing the configuration of a system according to Embodiment 3; FIG.
4 is a block diagram showing the configuration of an ear acoustic authentication device according to Embodiment 3. FIG.
5 is a flow chart showing the flow of processing executed by a filter generation unit of the ear acoustic authentication device according to the third embodiment; FIG.
6 is a flow chart showing the flow of processing executed by an ear acoustic authentication device according to Embodiment 3. FIG.
7 is a block diagram showing the configuration of an ear acoustic authentication device according to Embodiment 4. FIG.
8 is a flow chart showing the flow of processing executed by the registration unit of the ear acoustic authentication device according to Embodiment 4. FIG.
9 is a diagram showing a hardware configuration of an ear acoustic authentication device according to Embodiments 1 to 4; FIG.
MODE FOR CARRYING OUT THE INVENTION
[0013]
[Embodiment 1]
Embodiment 1 will be described with reference to FIGS.
[0014]
(ear acoustic authentication device 10)
An ear acoustic authentication device 10 according to the first embodiment will be described with reference to FIG. FIG. 1 is a block diagram showing the configuration of the ear acoustic authentication device 10. As shown in FIG. As shown in FIG. 1 , the ear acoustic authentication device 10 includes a feature extractor 11 , a corrector 12 and an authenticator 13 .
[0015]
When the feature extraction unit 11 inputs a test signal to the subject's ear using an audio device and receives an echo signal from the subject, the first feature extraction unit 11 regarding a system consisting of the audio device and the subject's ear Features are extracted from the reverberant signal.
[0016]
Specifically, an audio device is attached in advance to the subject's ear. The feature extraction unit 11 controls the audio device to output a test signal. The test signal is, for example, an impulse wave. The test signal reverberates in a closed system that includes the audio device and the subject's ears. The audio device detects echo signals output from the subject's ears.
[0017]
The feature extraction unit 11 communicates with the audio device and receives echo signals from the audio device by wire or wirelessly. A feature extraction unit 11 extracts an impulse response from the echo signal. An impulse response is a response to a test signal that is an impulse wave. The feature extraction unit 11 Fourier-transforms or Laplace-transforms the impulse response. Thereby, the feature extracting unit 11 calculates a transfer function representing acoustic characteristics of a closed system including the audio device and the subject's ear. A transfer function is an example of an acoustic property, and an acoustic property is an example of a first feature. Alternatively, the feature extraction unit 11 may extract, as the first feature, another response function based on the echo signal instead of the transfer function. The feature extractor 11 transmits the first feature to the corrector 12 .
[0018]
The correction unit 12 corrects the first feature to the second feature when the audio device is a predetermined reference device.
[0019]
Specifically, the correction unit 12 receives the first feature from the feature extraction unit 11 . As noted above, in one example, the first feature is the acoustic properties for a closed system that includes the audio device and the subject's ears. The corrector 12 corrects the first feature to a second feature that would be obtained from a closed system including a given reference device and the subject's ear. In other words, the correction unit 12 calculates, as the second feature, acoustic characteristics obtained when the condition regarding the subject remains unchanged and the audio device is replaced with the reference device.
[0020]
I will add a little more about the correction of the first feature. The acoustic characteristics, which are the first feature, can be mathematically expressed separately into (acoustic characteristics with the audio device as a factor) and (acoustic characteristics with the subject as a factor). For each audio device, it is possible to extract an acoustic characteristic with the audio device as a factor from the acoustic characteristic that is the first feature.
[0021]
However, when it is assumed that the subject uses a wide variety of audio devices for ear acoustic authentication, a closed system including the audio device and the subject's ear is used for all pairs of the audio device and the subject. It may not be realistic to obtain in advance the acoustic characteristics for .
[0022]
Therefore, in order to correct the acoustic characteristics of the closed system including the audio device and the subject's ears to the acoustic characteristics of the closed system including the reference device and the subject's ears, A parameter representing the relationship between (characteristics) and (acoustic characteristics with a predetermined reference device as a factor) is calculated in advance. Specifically, this parameter is expressed as (acoustic characteristics with a predetermined reference device as a factor)/(acoustic characteristics with an audio device as a factor). This parameter does not include subject-factored acoustic characteristics. Therefore, this parameter can be used regardless of the subject.
[0023]
Specifically, the correction unit 12 uses the above relational expression to convert the first feature consisting of (acoustic characteristics with the audio device as a factor) and (acoustic characteristics with the subject as a factor) into ( Acoustic characteristics with the reference device as a factor) and (acoustic characteristics with the subject as a factor) are corrected to second features. Details of the relational expressions described here will be described in more detail in the third embodiment.
[0024]
Alternatively, each of the first feature and the second feature may be useful information (data) for authenticating the subject, which is extracted from the acoustic characteristics rather than the acoustic characteristics themselves. In this case, the correction unit 12 extracts the first feature from acoustic characteristics obtained from a closed system including the audio device and the subject's ears. For example, the first feature is the mel-frequency cepstrum coefficient (MFCC). Further, the correction unit 12 corrects the first feature to the second feature. In this case, the second feature is the same type of feature as the first feature, extracted from acoustic signatures obtained from a closed system comprising a given reference device and the subject's ear.
[0025]
The authentication unit 13 authenticates the subject by matching the second feature with the feature indicated in the pre-registered authentication information.
[0026]
Specifically, the authentication unit 13 receives information indicating the second feature from the correction unit 12 . The authentication unit 13 collates the second feature with a pre-registered person's feature (authentication information). The authentication unit 13 calculates the degree of similarity (for example, a function of distance in the feature amount space) between the second feature and the feature of a person registered in advance. When the calculated similarity exceeds the threshold, the authentication unit 13 determines that the target person and the pre-registered person are the same person (authentication success). On the other hand, when the degree of similarity is equal to or less than the threshold, the authentication unit 13 determines that the target person and the pre-registered person are not the same person (authentication failure).
[0027]
The authentication unit 13 may present the authentication result on a display or the like by outputting the authentication result.
[0028]
(Operation of ear acoustic authentication device 10)
The operation of the ear acoustic authentication device 10 according to the first embodiment will be described with reference to FIG. FIG. 2 is a flow chart showing the flow of processing executed by the ear acoustic authentication device 10. As shown in FIG.
[0029]
As shown in FIG. 2, the feature extraction unit 11 uses an audio device to input a test signal to the subject's ear, and when receiving an echo signal from the subject, the sound A first feature for the system consisting of the voice device and the subject's ear is extracted from the echo signal (S1). The feature extraction unit 11 transmits information indicating the first feature to the correction unit 12 .
[0030]
The correction unit 12 receives information indicating the first feature from the feature extraction unit 11 . The correction unit 12 corrects the first feature to the second feature when the audio device is the predetermined reference device (S2). Correction unit 12 transmits information indicating the second feature to authentication unit 13 .
[0031]
The authentication unit 13 receives information indicating the second feature from the correction unit 12 . The authentication unit 13 authenticates the subject by matching the second feature with the feature indicated in the pre-registered authentication information. (S3). After that, the authentication unit 13 outputs the authentication result.
[0032]
Thus, the operation of the ear acoustic authentication device 10 according to the first embodiment is completed.
[0033]
(Effect of this embodiment)
According to the configuration of the present embodiment, the feature extraction unit 11 uses an audio device to input a test signal to the subject's ear, and when receiving an echo signal from the subject, the audio device and the subject's A first feature for the ear system is extracted from the echo signal. The correction unit 12 corrects the first feature to the second feature when the audio device is the predetermined reference device. The authentication unit 13 authenticates the subject by collating the second feature with pre-registered authentication information.
[0034]
That is, the ear acoustic authentication apparatus 10 performs ear acoustic authentication based on the second feature when the voice device is a predetermined reference device, not the first feature whose factor is the voice device. Therefore, regardless of which audio device is used for earacoustic authentication, the subject can be authenticated with high accuracy.
[0035]
[Embodiment 2]
In the second embodiment, the configuration of the ear acoustic authentication device 10 is as described in the first embodiment. Only the differences between the second embodiment and the first embodiment will be described below.
[0036]
Based on the second feature, for the purpose of authenticating the subject, the acoustic characteristics of the audio device as a factor are noise. That is, when an otoacoustic impression is performed based on the second feature described in the first embodiment, the authentication accuracy may vary among individual audio devices. Taking this into consideration, in the second embodiment, the correction unit 12 of the ear acoustic authentication device 10 corrects the first feature to another first feature. Here, the first feature is the acoustic properties themselves for the closed system that includes the audio device and the subject's ears. Alternatively, the first feature is a feature extracted from its acoustic properties. "Another first feature" is included in the concept of the first feature of the present invention, although it is different from the first feature described in the first embodiment. Note that the second feature can also be corrected using the method described below. In this case, in the following description, the "(original) first feature" should be read as "(original) second feature".
[0037]
In the second embodiment, the correction unit 12 corrects the first feature to another first feature. Specifically, in the first embodiment, the correction unit 12 converts (scales) the original first feature into another first feature based on the following formula. In the following, it may be described as correction in the sense of conversion (scaling).
[0038]
[Number 1]

[0039]
Here, μ tar is the average of the features obtained from the closed system including the audio device and the ears of each person, for each combination of the audio device and a plurality of different persons. σ tar is the standard deviation of the features obtained from the closed system containing the audio device and each person's ear for each combination of audio device and different persons. The multiple different persons may or may not include the subject. Another first feature MY tar shown in equation (1) roughly indicates that we are interested in how much the first feature Y tar deviates from the average μ tar.
[0040]
μ tar means the average of features extracted from corrected acoustic characteristics when authenticating multiple persons using an audio device. In Equation (1), the acoustic characteristics of the audio device as a factor are commonly included in Y tar and μ tar. Therefore, by the operation of (Y tar−μ tar) in the above formula, the acoustic characteristics factored by the audio device in the other first feature MY tar either disappear or become the original first feature Y tar is at least less than
[0041]
Alternatively, the correction unit 12 may convert the first feature Y tar into another first feature MY tar' based on the following formula (2) instead of the formula (1) described above. .
[0042]
[Number 2]

[0043]
Here, μ ref is the average of the features obtained from the closed system including the audio device and each person's ear for each combination of the reference device and a plurality of different persons. F ref is a feature obtained from a closed system that includes a reference device and a specific person's ear. F tar is a feature obtained from a closed system that includes an audio device and a specific person's ear. A specific person is a person different from the subject. Note that the reference device may be one audio device, or virtually a combination of multiple audio devices. When the reference device is a synthesis of multiple (M>1) audio devices, the value (μ vref) obtained by dividing the average sum of features for each audio device by M is given by μ ref in equation (2). replace with Also, in this case, the value (σ vref) obtained by dividing the sum of the feature standard deviations for each audio device by M is replaced with σ ref in Equation (2).
[0044]
Referring to the right side of equation (2), the original first feature Y tar is transformed (scaled) based on the average μ ref of the reference device features. That is, another first feature MY tar' given by equation (2) can also be regarded as a scaled first feature Y tar. By substituting the scaled first feature Y tar (=MY tar') into equation (1), it is possible to obtain yet another first feature MY tar (variant).
[0045]
The correction unit 12 transmits to the authentication unit 13 information indicating another second feature obtained from the other first feature (MY tar or MY tar') obtained as described above.
[0046]
In the second embodiment, the authentication unit 13 authenticates the subject by comparing another second feature obtained from another first feature with pre-registered authentication information. That is, the authentication unit 13 calculates the degree of similarity between another second feature and the feature of the person registered in advance. When the degree of similarity exceeds the threshold, the authentication unit 13 determines that the target person and the pre-registered person are the same person (successful authentication). On the other hand, when the degree of similarity is equal to or less than the threshold, the authentication unit 13 determines that the target person and the pre-registered person are not the same person (authentication failure).
[0047]
(Effect of this embodiment)
According to the configuration of the present embodiment, the feature extraction unit 11 uses an audio device to input a test signal to the subject's ear, and when receiving an echo signal from the subject, the audio device and the subject's A first feature for the ear system is extracted from the echo signal. The correction unit 12 corrects the first feature to the second feature when the audio device is the predetermined reference device. The authentication unit 13 authenticates the subject by collating another second feature with pre-registered authentication information.
[0048]
That is, the ear acoustic authentication apparatus 10 performs ear acoustic authentication based on a second feature when the voice device is a predetermined reference device instead of the first feature whose factor is the voice device. Therefore, regardless of which audio device is used for earacoustic authentication, the subject can be authenticated with high accuracy.
[0049]
Furthermore, when considering the acoustic characteristics of the audio device as a factor, the second feature is that noise causes variations for each individual audio device. Taking this into consideration, in the present embodiment, the ear acoustic authentication device 10 corrects the first feature to another first feature described above. Thereby, in the alternative first feature, the acoustic characteristics attributed to the audio device are eliminated or at least reduced compared to the original first feature.
[0050]
(Modified example)
In one variant, the further first feature MY tar' given by equation (2) above is considered to be the scaled first feature Y tar. In this modified example, MY tar′ shown in Equation (2) is introduced as Y tar on the right side of Equation (1). However, in this case μ tar and σ tar in equation (1) also need to be scaled similarly to Y tar based on the average μ ref of the reference device features.
[0051]
In this modified example, the above formula (1) is modified as the following formula (1)'.
[0052]
[Number 3]

[0053]
Here, the second feature Y tar' is equal to MY tar' shown in Equation (2) above. Also, μ tar′ is obtained by recalculating μ tar described above in correspondence with the scaling of the second feature Y tar. σ tar′ is obtained by recalculating σ tar as described above, corresponding to the scaling of the second feature Y tar.
[0054]
According to the configuration of this modified example, the second feature Y tar is scaled based on the average μ ref of the features of the reference device according to the formula shown on the right side of Equation (2). Another first feature MY tar is then calculated from the scaled second feature Y tar' according to equation (1)'. In this way, from the scaled first feature Y tar' another first feature MY tar can be obtained.
[0055]
[Embodiment 3]
Embodiment 3 will be described with reference to FIGS. 3 to 6. FIG.
[0056]
(System 1)
FIG. 3 is a diagram schematically showing the configuration of the system 1 according to the third embodiment. In FIG. 3 , the system 1 includes an ear acoustic authentication device 100 , a storage device 200 and an audio device 300 . The ear acoustic authentication device 100, the storage device 200, and the audio device 300 may be part of an authentication device for ear acoustic authentication.
[0057]
Alternatively, the ear acoustic authentication device 100 and the storage device 200 may be on a network server. In this case, the audio device 300, the ear acoustic authentication apparatus 100, and the storage device 200 are communicably connected wirelessly, for example, by a mobile network. Alternatively, only storage device 200 may reside on a network server. In this case, the ear acoustic authentication device 100 and the audio device 300 are part of the authentication device, and the ear acoustic authentication device 100 and the storage device 200 are communicably connected by a wireless network.
[0058]
The audio device 300 generates a test signal and transmits the test signal from the speaker (FIG. 3) of the audio device 300 toward the subject's ear. The inspection signal is, for example, an impulse wave, as in the first embodiment. The audio device 300 transmits the transmitted test signal to the feature extraction unit 103 of the ear acoustic authentication apparatus 100 .
[0059]
The audio device 300 observes echo signals from the subject. Specifically, the audio device 300 uses the microphone (FIG. 3) of the audio device 300 to detect echo signals from the subject, and transmits the detected echo signals to the feature extraction unit 103 of the ear acoustic authentication apparatus 100. do.
[0060]
(ear acoustic authentication device 100)
The configuration of the ear acoustic authentication device 100 according to the third embodiment will be described with reference to FIG. FIG. 4 is a block diagram showing the configuration of the ear acoustic authentication device 100. As shown in FIG. As shown in FIG. 4, the ear acoustic authentication device 100 includes a feature extraction unit 103, a correction unit 104 and an authentication unit 105, a filter generation unit 106 is provided. The ear acoustic authentication apparatus 100 may further include an audio device control section (not shown). In this case, all or part of the operation of the audio device 300 described above is controlled by the audio device control section of the ear acoustic authentication apparatus 100 .
[0061]
The feature extraction unit 103 uses the audio device 300 to input a test signal to the subject's ear, and when receiving an echo signal from the subject, the feature extraction unit 103 extracts a first 1 features are extracted from the echo signal.
[0062]
Specifically, the audio device 300 is attached in advance to the ear of the subject. The feature extraction unit 103 controls the audio device 300 to output a test signal. The test signal is, for example, an impulse wave. The test signal reverberates in a closed system that includes the audio device 300 and the subject's ear. The audio device 300 detects echo signals output from the subject's ears.
[0063]
The feature extraction unit 103 communicates with the audio device 300 and receives echo signals from the audio device 300 by wire or wirelessly. A feature extraction unit 103 extracts an impulse response from the echo signal. The feature extraction unit 11 performs Fourier transform or Laplace transform on the impulse response to calculate a transfer function indicating acoustic characteristics of a closed system including the audio device 300 and the subject's ear. Acoustic properties and transfer functions are examples of first features. Alternatively, the feature extraction unit 103 may extract, as the first feature, another response function based on the echo signal instead of the transfer function. Feature extraction section 103 transmits the first feature to correction section 104 .
[0064]
The correction unit 104 corrects the first feature to the second feature when the audio device 300 is the predetermined reference device.
[0065]
Specifically, the correction unit 104 receives the first feature from the feature extraction unit 103 . A corrector 104 corrects the first feature to a second feature that would be obtained from a closed system including a given reference device and the subject's ear. In other words, the correction unit 104 calculates, as the second feature, acoustic characteristics obtained when the condition regarding the subject remains unchanged and the audio device 300 is replaced with the reference device. Alternatively, the correction section 104 may correct the second feature to another first feature, like the correction section 12 according to the second embodiment. In this case, in the following description, "second feature" is read as "another first feature".
[0066]
The first and second features mentioned above will be additionally explained. Here, it is assumed that the acoustic characteristic, which is the first feature, is the transfer function for the system consisting of the audio device and the subject's ear. When the subject wears the audio device, the audio device and the subject's ear canal can be considered as one sound system. In this case, as is well known, the acoustic characteristics of the entire acoustic system, that is, the acoustic characteristics as the first feature, are the product of the acoustic characteristics with the audio device as a factor and the acoustic characteristics with the subject as a factor. is represented by A transfer function is an example of an acoustic characteristic. Alternatively, the feature extraction unit 103 may extract, as the first feature, another response function based on the echo signal instead of the transfer function. A case where the acoustic characteristic, which is an example of the first feature, is a transfer function will be described below. Specifically, the acoustic characteristic (an example of the first feature), which is the first feature, is represented by the following formula. In equation (3), ω is the frequency of the sound.
[0067]
[Number 4]

[0068]
In Equation (3), G tar is defined as the first feature, the acoustic characteristic, of the system consisting of the audio device 300 and the ears of a specific person. D tar(ω) is defined as the acoustic characteristic factored by the audio device 300 . Also, p(ω) is defined as an acoustic characteristic with the subject as a factor.
[0069]
The feature extraction unit 103 calculates the acoustic characteristics of the system consisting of the audio device 300 and the subject's ears. That is, the first feature is G tar in equation (3).
[0070]
Here, it is assumed that the acoustic characteristic, which is the second feature, is the transfer function for the system consisting of the reference device and the subject's ear. In this case, the acoustic characteristic, which is the second feature, is represented by the product of the acoustic characteristic having the reference device as a factor and the acoustic characteristic having the subject as a factor. Specifically, the acoustic characteristic (which is an example of the second feature) calculated by the correction unit 104 is represented by the following formula. In equation (4), ω is the frequency of the sound.
[0071]
[Number 5]

[0072]
In Equation (4), G ref is defined as the second feature, the acoustic characteristic, of the system consisting of the reference device and the ear of a specific person. D ref is defined as an acoustic characteristic factored by the reference device. Also, p is the acoustic characteristic with the subject as a factor, as described above. Note that the first feature and the second feature are not limited to transfer functions. The first feature may be an acoustic characteristic different from the transfer function, as long as it is expressed in the form of Equation (3) above. Also, the second feature may be an acoustic characteristic different from the transfer function as long as it is expressed in the form of the above-described equation (4).
[0073]
The filter generation unit 106 generates a correction filter that is the ratio of the acoustic characteristic with the audio device 300 as a factor and the acoustic characteristic with the reference device as a factor. Then, the filter generator 106 stores the calculated correction filter in the storage device 200 of the system 1 as device information.
[0074]
(Generation of correction filter)
A procedure for the filter generation unit 106 according to the third embodiment to generate the correction filter described above will be described. For example, the filter generator 106 generates a correction filter for the audio device 300 for each type of audio device 300, each production lot, or each manufacturer.
[0075]
FIG. 5 is a flowchart showing the flow of processing executed by the filter generation unit 106. FIG. Here, it is assumed that device information indicating acoustic characteristics (G'ref) relating to a system consisting of a reference device and a specific person is registered in the storage device 200 of the system 1 in advance. The specific person may or may not be the subject himself/herself.
[0076]
As shown in FIG. 5, the filter generation unit 106 first acquires device information related to the audio device 300 from the storage device 200 of the system 1 (S201). The filter generation unit 106 includes the acoustic characteristics (G'tar) related to the system consisting of the audio device 300 and the specific person from the device information. In addition, the filter generation unit 106 acquires acoustic characteristics (G' ref) related to the system composed of the reference device and the specific person (S202).
[0077]
Here, we define the following parameter Ftar that indicates the relationship between the acoustic characteristic G'tar and the acoustic characteristic G'ref. G'tar is obtained by replacing the acoustic characteristic p(ω) whose factor is the target person with the acoustic characteristic p'(ω) whose factor is a specific person in Equation (3). G'ref is obtained by replacing the acoustic characteristic p(ω) with the subject as a factor with the acoustic characteristic p'(ω) with a specific person as a factor in Equation (4).
[0078]
[Number 6]

[0079]
G'tar is obtained by replacing p(ω) with p'(ω) in equation (3), and G'ref is obtained by replacing p(ω) with p'(ω) in equation (4). are substituted into the equation (5) respectively.
[0080]
[Number 7]

[0081]
In Equation (6), Dtar(ω) is an acoustic characteristic with the audio device 300 as a factor. p'(ω) is an acoustic characteristic with a specific person as a factor. As noted above, the specific person may be determined independently of the subject.
[0082]
That is, the parameter F tar represents the relationship between the acoustic characteristic D ref whose factor is the reference device and the acoustic characteristic D tar whose factor is the audio device 300 . If the parameter F tar is applied to G tar, the acoustic characteristic D tar factored by the audio device 300 is removed (filtered out). In this sense, the parameter F tar is hereinafter referred to as a correction filter. The correction filter F tar is the ratio of the acoustic characteristic D tar factored by the audio device 300 and the acoustic characteristic D ref factored by the reference device.
[0083]
Acoustic characteristics G′tar for the system consisting of the audio device 300 and the ears of the specific person are calculated under the same conditions as when calculating the acoustic characteristics G ref for the system consisting of the reference device and the ears of the specific person. . Specifically, the audio device 300 is worn by the same person as the specific person for whom the acoustic characteristic G ref is to be calculated. The audio device 300 transmits the same test signal as when the acoustic characteristic Gref was calculated from the speaker. The audio device 300 uses the microphone of the audio device 300 to observe echo signals from a specific person's ear. Audio device 300 transmits the observed echo signal to filter generator 106 .
[0084]
The filter generation unit 106 acquires the acoustic characteristics G'tar of the system composed of the audio device 300 and the specific person's ear based on the known test signal and the echo signal received from the audio device 300 (S203). .
[0085]
The filter generation unit 106 generates a correction filter F tar for the combination of the reference device and the audio device 300 based on Equation (6) described above (S204). Then, the filter generator 106 stores the calculated correction filter F tar in the storage device 200 of the system 1 as device information (S205).
[0086]
This completes the generation of the correction filter.
[0087]
The correction unit 104 shown in FIG. 4 acquires the device information stored in the storage device 200 of the system 1, and uses the correction filter F tar shown in Equation (6) to convert the first feature (G tar) to the second (G ref ). That is, the correction unit 104 performs the following calculations.
[0088]
[Number 8]

[0089]
The correction unit 104 transmits information indicating the second feature (G ref) thus calculated to the authentication unit 105 .
[0090]
The authentication unit 105 authenticates the subject by matching the second feature with the feature indicated in the pre-registered authentication information.
[0091]
Specifically, the authentication unit 105 receives information indicating the second feature from the correction unit 104 . The authentication unit 105 collates the second feature with pre-registered features of the person (authentication information). The authentication unit 105 calculates a degree of similarity (for example, Mel frequency cepstrum coefficient (MFCC)) between the second feature and a pre-registered person feature. When the degree of similarity exceeds the threshold, the authentication unit 105 determines that the target person and the pre-registered person are the same person (successful authentication). On the other hand, when the degree of similarity is equal to or less than the threshold, the authentication unit 105 determines that the target person and the pre-registered person are not the same person (authentication failure).
[0092]
The authentication unit 105 outputs the authentication result. For example, the authentication unit 105 may cause a display device (not shown) to display information indicating whether authentication has succeeded or failed.
[0093]
(Modified example)
A modified example related to generation of correction filters by the filter generation unit 106 will be described.
[0094]
In this modified example, p'(ω) in Equation (6) corresponds to multiple persons. In other words, the particular person mentioned above is virtually a composite of multiple persons. Specifically, let p i (i=1, 2, . In this case, the correction filter F tar is determined as follows.
[0095]
[Number 9]

[0096]
　According to formula (6)', the reference device and the specific person ( Acoustic characteristics G ref for a system consisting of the ears of a plurality of persons (corresponding to a plurality of persons in this modified example) are acoustic characteristics D ref whose factor is the reference device and acoustic characteristics p i (i= 1, 2, . . . ). Further, the acoustic characteristics G′ tar for a system composed of the audio device 300 and the ears of a specific person (in this modification, corresponding to a plurality of persons) are the acoustic characteristics D tar with the audio device 300 as a factor, and a plurality of It can be separated into the sum of acoustic characteristics p i (i=1, 2, . . . ) with the person as a factor. Note that the reference device may virtually be a combination of multiple (N>1) audio devices 300 . In this case, the value (D vref) obtained by dividing the sum of the acoustic characteristics D ref with each audio device 300 as a factor by N is replaced with D ref in Equation (6)′.
[0097]
Looking at the right side of formula (6)' related to this modification, it is the same as the right side of formula (6) described above. This is because the sum of acoustic characteristics p i (i=1, 2, . . That is, although the process of calculating the correction filter F tar is different in this modified example, the correction filter F tar obtained by Equation (6)′ is the same as the correction filter F tar obtained by Equation (6).
[0098]
In this modified example, it is assumed that the sum (or the average) of the acoustic characteristics p i (i=1, 2, . can be regarded For this reason, it can be expected that fluctuations (noise) in measured values ​​of acoustic characteristics caused by each individual person can be canceled out.
[0099]
(Operation of ear acoustic authentication device 100)
The operation of the ear acoustic authentication device 100 according to the second embodiment will be described with reference to FIG. FIG. 6 is a flow chart showing the flow of processing executed by the ear acoustic authentication device 100 .
[0100]
As shown in FIG. 6, the audio device 300 generates a test signal and transmits the test signal from the speaker (FIG. 3) of the audio device 300 to the subject's ear (S101). The audio device 300 transmits the emitted inspection signal to the feature extraction unit 103 .
[0101]
The audio device 300 observes echo signals from the subject (S102). The audio device 300 transmits the observed echo signal to the feature extraction unit 103 .
[0102]
The feature extraction unit 103 uses the audio device 300 to input a test signal to the subject's ear, and when receiving an echo signal from the subject, the feature extraction unit 103 extracts a first 1 features are extracted from the echo signal. Specifically, the feature extraction unit 103 calculates the acoustic characteristic G tar, which is the first feature, based on the inspection signal and the echo signal (S103). Feature extraction section 103 transmits information indicating the first feature to correction section 104 .
[0103]
The correction unit 104 receives information indicating the first feature from the feature extraction unit 11 . The correction unit 104 corrects the first feature to the second feature when the audio device 300 is the predetermined reference device.
[0104]
Specifically, the correction unit 104 acquires the correction filter Ftar stored as device information in the storage device 200 (S104).
[0105]
Then, the correction unit 104 corrects the acoustic characteristic G tar (first feature) to the acoustic characteristic G ref (second feature) using the correction filter F tar (S105). Correction unit 104 transmits information indicating the second feature to authentication unit 105 .
[0106]
The authentication unit 105 receives information indicating the second feature from the correction unit 12 . The authentication unit 105 authenticates the subject by matching the second feature with the feature indicated in the pre-registered authentication information. (S106). After that, the authentication unit 105 outputs the authentication result (S107).
[0107]
Thus, the operation of the ear acoustic authentication device 100 according to the second embodiment is completed.
[0108]
(Effect of this embodiment)
According to the configuration of this embodiment, the feature extraction unit 103 uses the audio device 300 to input the test signal to the ear of the subject, and receives the echo signal from the subject. A first feature relating to the human ear system is extracted from the echo signal. The correction unit 104 corrects the first feature to the second feature when the audio device 300 is the predetermined reference device. The authentication unit 105 authenticates the subject by matching the second feature with the feature indicated in the pre-registered authentication information.
[0109]
That is, the ear acoustic authentication apparatus 100 performs ear acoustic authentication based on the second feature when the voice device 300 is a predetermined reference device, not the first feature with the voice device 300 as a factor. Therefore, regardless of which audio device 300 is used for earacoustic authentication, the subject can be authenticated with high accuracy.
[0110]
Furthermore, the filter generation unit 106 generates an acoustic characteristic G′ tar for the system consisting of the audio device 300 and the specific person's ear, and an acoustic characteristic G ref for the system consisting of the reference device and the specific person's ear. Generate a correction filter F tar that is the ratio. As a result, the correction unit 104 uses the correction filter F tar to correct the acoustic characteristic G′ tar as the first feature to the second feature G ref when the audio device 300 is the predetermined reference device. can do.
[0111]
[Embodiment 4]
Embodiment 4 will be described with reference to FIGS. 7 and 8. FIG.
[0112]
(ear acoustic authentication device 100a)
The configuration of an ear acoustic authentication device 100a according to Embodiment 4 will be described with reference to FIG. FIG. 7 is a block diagram showing the configuration of the ear acoustic authentication device 100a. As shown in FIG. 7 , the ear acoustics authentication device 100 a includes a registration unit 107 in addition to the feature extraction unit 103 , the correction unit 104 and the authentication unit 105 . The ear acoustic authentication apparatus 100a may further include an audio device control section (not shown). In this case, all or part of the operation of the audio device 300 (FIG. 3), which will be described later, is controlled by the audio device control section of the ear acoustic authentication apparatus 100a.
[0113]
If the target person is the target person for registration, the registration unit 107 associates the second feature corrected by the correction unit 104 with the target person and registers it as authentication information.
[0114]
The processes executed by the feature extraction unit 103, the correction unit 104, and the authentication unit 105 of the ear acoustic authentication device 100a according to the fourth embodiment are the same as those of the ear acoustic authentication device 100 according to the third embodiment, unless otherwise specified below. is similar to
[0115]
(Operation of ear acoustic authentication device 100a)
The operation of the ear acoustic authentication device 100a according to the fourth embodiment will be described with reference to FIG. FIG. 8 is a flow chart showing the flow of processing executed by the ear acoustic authentication device 100a.
[0116]
As shown in FIG. 8, the audio device 300 generates a test signal and transmits the test signal from the speaker of the reference device (FIG. 3) toward the ears of the subject (S301). The reference device transmits the emitted inspection signal to the feature extraction unit 103 . The reference device is a specific pre-selected audio device 300 .
[0117]
The reference device observes echo signals from the subject (S302). The audio device 300 transmits the observed echo signal to the feature extraction unit 103 .
[0118]
The feature extraction unit 103 uses the audio device 300 to input the test signal to the subject's ear, and when receiving the echo signal from the subject, the feature extraction unit 103 extracts the sound related to the system composed of the audio device 300 and the subject's ear. A characteristic G tar is calculated (S303). The feature extraction unit 103 transmits information indicating the acoustic characteristic Gtar to the registration unit 107 .
[0119]
The registration unit 107 receives information indicating the acoustic characteristic Gtar from the feature extraction unit 103 . The registration unit 107 uses the acoustic characteristic G tar and the correction filter F tar to calculate the acoustic characteristic G ref to be stored as authentication information in the storage device 200 (S304). Here, the acoustic characteristic G ref is obtained as follows.
[0120]
[Number 10]

[0121]
That is, the registration unit 107 obtains the acoustic characteristic G ref by multiplying the acoustic characteristic G tar by the correction filter F tar. The registration unit 107 stores information indicating the acoustic characteristic G ref in the storage device 200 as authentication information.
[0122]
With this, the operation of the ear acoustic authentication device 100a according to the present embodiment is completed.
[0123]
(Effect of this embodiment)
According to the configuration of this embodiment, the feature extraction unit 103 uses the audio device 300 to input the test signal to the ear of the subject, and receives the echo signal from the subject. A first feature relating to the human ear system is extracted from the echo signal. The correction unit 104 corrects the first feature to the second feature when the audio device 300 is the predetermined reference device. The authentication unit 105 authenticates the subject by comparing another first feature with pre-registered authentication information.
[0124]
That is, the otoacoustic authentication apparatuses 100 and 100a do not use the first feature with the voice device 300 as a factor, but based on another first feature when the voice device 300 is a predetermined reference device. Authenticate. Therefore, regardless of which audio device 300 is used for earacoustic authentication, the subject can be authenticated with high accuracy.
[0125]
(Modified example)
In one modification, a synthesis of multiple audio devices 300 is used as a virtual reference device. Specifically, in step S303 of the flow shown in FIG. 8 , the feature extraction unit 103 uses N (>1) audio devices 300 to extract acoustic information related to a system composed of each audio device 300 and the subject's ear. Calculate each characteristic. Feature extraction section 103 transmits information indicating the calculated acoustic characteristics to registration section 107 . In the following, G i (i=1 to N) is the acoustic characteristic of a system composed of each of the first to N-th audio devices 300 and the subject's ear.
[0126]
The registration unit 107 calculates an acoustic characteristic G ref regarding a system composed of a virtual reference device and the subject's ear based on the acoustic characteristic G i (i=1 to N). In this modified example, the second feature is acoustic characteristics G i (second acoustic characteristics) (i=1 to N) for a plurality of systems composed of a plurality of different audio devices 300 and the subject's ears. is the average of Specifically, the acoustic characteristic G ref according to this modified example is expressed as follows.
[0127]
[Number 11]

[0128]
Here, D i(ω) is an acoustic characteristic whose factor is the i-th audio device 300 synthesized to the virtual reference device. p(ω) is an acoustic characteristic with the subject as a factor. Also, the overlined D i(ω) on the right side of equation (8) represents the average of (part of) the acoustic characteristics of the first to N-th audio devices 300 as factors.
[0129]
According to Equation (8), in this modification, the average of (part of) the acoustic characteristics with the N audio devices 300 as factors corresponds to the acoustic characteristics with the virtual reference device as a factor. For this reason, it can be expected that fluctuations (noise) in the measured values ​​of the acoustic characteristics, which are caused by the individual audio devices 300, can be canceled out.
[0130]
(About hardware configuration)
Each component of the ear acoustic authentication devices 100 and 100a described in Embodiments 1 to 4 represents a functional unit block. Some or all of these components are realized by an information processing device 900 as shown in FIG. 9, for example. FIG. 9 is a block diagram showing an example of the hardware configuration of the information processing device 900. As shown in FIG.
[0131]
As shown in FIG. 9, the information processing device 900 includes the following configuration as an example.
[0132]
· CPU (Central Processing Un it) 901
· ROM (Read Only Memory) 902
· RAM (Random Access Memory) 903
· Program 904 loaded into RAM 903
· Storage device 905 for storing program 904
· A drive device 907 that reads and writes the recording medium 906
· A communication interface 908 that connects to the communication network 909
· Input/output interface 910 for data input/output
· Bus 911 connecting each component
Each component of the ear acoustic authentication devices 100 and 100a described in Embodiments 1 to 4 is implemented by the CPU 901 reading and executing the program 904 that implements these functions. A program 904 that implements the function of each component is stored in advance in, for example, the storage device 905 or the ROM 902, and is loaded into the RAM 903 and executed by the CPU 901 as necessary. The program 904 may be supplied to the CPU 901 via the communication network 909 or may be stored in the recording medium 906 in advance, and the drive device 907 may read the program and supply it to the CPU 901 .
[0133]
According to the above configuration, the ear acoustic authentication devices 100 and 100a described in the first to fourth embodiments are implemented as hardware. Therefore, the same effects as those described in the first to fourth embodiments can be obtained.
[0134]
[Appendix]
Some or all of the above embodiments can also be described as the following additional remarks, but are not limited to the following.
[0135]
(Appendix 1)
When an audio device is used to input a test signal to the subject's ear and an echo signal is received from the subject, the first feature relating to the system consisting of the audio device and the subject's ear is described above. a feature extraction means for extracting from the echo signal;
a correction means for correcting the first feature to a second feature when the audio device is a predetermined reference device;
An authentication means for authenticating the subject by comparing the second feature with pre-registered authentication information
Ear acoustic authentication device.
[0136]
(Appendix 2)
The feature extraction means calculates, as the first feature, a first acoustic characteristic of a system consisting of the audio device and the subject's ear based on the test signal and the echo signal,
The correction means corrects the first acoustic characteristic to the second acoustic characteristic when the audio device is the reference device
The ear acoustic authentication device according to Supplementary Note 1, characterized by:
[0137]
(Appendix 3)
The first acoustic characteristic is represented by the product of the acoustic characteristic with the voice device as a factor and the acoustic characteristic with the subject as a factor,
The second acoustic characteristic is represented by the product of the acoustic characteristic with the reference device as a factor and the acoustic characteristic with the subject as a factor
The ear acoustic authentication device according to Supplementary Note 2, characterized by:
[0138]
(Appendix 4)
The second feature is the average of the second acoustic characteristics for a plurality of systems consisting of a plurality of different audio devices and a specific person's ear
An otoacoustic authentication device according to Supplementary Note 2 or 3 characterized by:
[0139]
(Appendix 5)
The correction means calculates an average of features relating to a plurality of systems consisting of the audio device and the ears of a plurality of different persons, and subtracts the calculated average of the features from the first feature. Compute the first feature
An ear acoustic authentication device according to any one of appendices 1 to 4, characterized by:
[0140]
(Appendix 6)
Filter generating means for generating a correction filter for correcting the acoustic characteristics of the system consisting of the audio device and the ears of a specific person to the acoustic characteristics of the system consisting of the reference device and the ears of the specific person. further comprising
The correction means corrects the first feature to the second feature using the correction filter
An ear acoustic authentication device according to any one of appendices 1 to 5, characterized by:
[0141]
(Appendix 7)
The correction filter is the ratio of the acoustic characteristics with the audio device as a factor and the acoustic characteristics with the reference device as a factor
The ear acoustic authentication device according to Supplementary Note 6, characterized by:
[0142]
(Appendix 8)
The first feature is a first acoustic characteristic of a system consisting of the audio device and the subject's ear,
The first acoustic characteristic is represented by the product of the acoustic characteristic with the voice device as a factor and the acoustic characteristic with the subject as a factor
The ear acoustic authentication device according to Supplementary Note 7, characterized by:
[0143]
(Appendix 9)
The filter generating means generates the correction filter using the acoustic characteristics of a system consisting of the audio device and the ears of one person and the acoustic characteristics of a system consisting of the reference device and the ears of the same person. generate
The ear acoustic authentication device according to any one of appendices 6 to 8, characterized in that:
[0144]
(Appendix 10)
The filter generating means generates acoustic characteristics of a plurality of systems composed of the audio device and the ears of a plurality of different people, and acoustic characteristics of a plurality of systems composed of the reference device and the ears of the different people. using the characteristics to generate the correction filter
The ear acoustic authentication device according to any one of appendices 6 to 8, characterized in that:
[0145]
(Appendix 11)
　If the subject is a subject of registration,
further comprising registration means for registering the second feature corrected from the first feature by the correction means as authentication information in association with the subject person
An ear acoustic authentication device according to any one of appendices 1 to 10, characterized in that:
[0146]
(Appendix 12)
When an audio device is used to input a test signal to the subject's ear and an echo signal is received from the subject, the first feature relating to the system consisting of the audio device and the subject's ear is described above. extracted from the reverberant signal,
correcting the first feature to the second feature when the audio device is a predetermined reference device,
　Authenticate the subject by comparing the second feature with pre-registered authentication information
Ear acoustic authentication method including.
[0147]
(Appendix 13)
When an audio device is used to input a test signal to the subject's ear and an echo signal is received from the subject, the first feature relating to the system consisting of the audio device and the subject's ear is described above. extracting from the echo signal;
correcting the first feature to a second feature when the audio device is a predetermined reference device;
　Authentication of the subject by comparing the second feature with pre-registered authentication information
A non-temporary recording medium that stores a program that causes a computer to execute
[0148]
Although the present invention has been described with reference to the embodiments (and examples), the present invention is not limited to the above-described embodiments (and examples). Various changes can be made to the configurations and details of the embodiments (and examples) within the scope of the present invention that can be understood by those skilled in the art.
Industrial applicability
[0149]
The present invention can be used, for example, in biometric authentication, electronic devices such as hearing aids and earphones that are worn on people's ears, and security and other services.
Code explanation
[0150]
　　10 Ear acoustic authentication device
11 Feature extraction part
12 correction unit
13 Authentication section
100 ear acoustic authentication device
100a ear acoustic authentication device
103 Feature extraction unit
104 correction unit
105 authentication unit
106 filter generation unit
107 Registry Department
　300 Audio device
The scope of the claims
[Claim 1]
When an audio device is used to input a test signal to the subject's ear and an echo signal is received from the subject, the first feature relating to the system consisting of the audio device and the subject's ear is described above. a feature extraction means for extracting from the echo signal;
a correction means for correcting the first feature to a second feature when the audio device is a predetermined reference device;
an authentication means for authenticating the subject by comparing the second feature with the feature indicated in the pre-registered authentication information
Ear acoustic authentication device.
[Claim 2]
The feature extraction means calculates, as the first feature, a first acoustic characteristic of a system consisting of the audio device and the subject's ear based on the test signal and the echo signal,
The correction means corrects the first acoustic characteristic to the second acoustic characteristic when the audio device is the reference device
The ear acoustic authentication device according to claim 1, characterized by:
[Claim 3]
The first acoustic characteristic is represented by the product of the acoustic characteristic with the voice device as a factor and the acoustic characteristic with the subject as a factor,
The second acoustic characteristic is represented by the product of the acoustic characteristic with the reference device as a factor and the acoustic characteristic with the subject as a factor
The ear acoustic authentication device according to claim 2, characterized by:
[Claim 4]
The second feature is the average of the second acoustic characteristics for a plurality of systems consisting of a plurality of different audio devices and a specific person's ear
The ear acoustic authentication device according to claim 2 or 3, characterized in that:
[Claim 5]
The correction means calculates an average of features relating to a plurality of systems consisting of the audio device and the ears of a plurality of different persons, and subtracts the calculated average of the features from the first feature. Compute the first feature
The ear acoustic authentication device according to any one of claims 1 to 4, characterized by:
[Claim 6]
Filter generating means for generating a correction filter for correcting the acoustic characteristics of the system consisting of the audio device and the ears of a specific person to the acoustic characteristics of the system consisting of the reference device and the ears of the specific person. further comprising
The correction means corrects the first feature to the second feature using the correction filter
The ear acoustic authentication device according to any one of claims 1 to 5, characterized by:
[Claim 7]
The correction filter is the ratio of the acoustic characteristics with the audio device as a factor and the acoustic characteristics with the reference device as a factor
The ear acoustic authentication device according to claim 6, characterized by:
[Claim 8]
The first feature is a first acoustic characteristic of a system consisting of the audio device and the subject's ear,
The first acoustic characteristic is represented by the product of the acoustic characteristic with the voice device as a factor and the acoustic characteristic with the subject as a factor
The ear acoustic authentication device according to claim 7, characterized by:
[Claim 9]
The filter generating means generates the correction filter using the acoustic characteristics of a system consisting of the audio device and the ears of one person and the acoustic characteristics of a system consisting of the reference device and the ears of the same person. generate
The ear acoustic authentication device according to any one of claims 6 to 8, characterized by:
[Claim 10]
The filter generating means generates acoustic characteristics of a plurality of systems composed of the audio device and the ears of a plurality of different people, and acoustic characteristics of a plurality of systems composed of the reference device and the ears of the different people. using the characteristics to generate the correction filter
The ear acoustic authentication device according to any one of claims 6 to 8, characterized by:
[Claim 11]
　If the subject is a subject of registration,
further comprising registration means for registering the second feature corrected from the first feature by the correction means as authentication information in association with the subject person
The ear acoustic authentication device according to any one of claims 1 to 10, characterized by:
[Claim 12]
When an audio device is used to input a test signal to the subject's ear and an echo signal is received from the subject, the first feature relating to the system consisting of the audio device and the subject's ear is described above. extracted from the reverberant signal,
　The first feature is the Corrected to the second feature when it was a device,
　Authenticate the subject by comparing the second feature with the feature indicated in the pre-registered authentication information
Ear acoustic authentication method including.
[Claim 13]
When an audio device is used to input a test signal to the subject's ear and an echo signal is received from the subject, the first feature relating to the system consisting of the audio device and the subject's ear is described above. extracting from the echo signal;
correcting the first feature to a second feature when the audio device is a predetermined reference device;
　Authentication of the subject by comparing the second feature with the feature indicated in the pre-registered authentication information, and
A non-temporary recording medium that stores a program that causes a computer to execute