Articulation disorder detection device and articulation disorder detection method

Abstract

To provide an articulation disorder detection device and an articulation disorder detection method which can quickly detect an articulation disorder.SOLUTION: An articulation disorder detection device comprises: a first line generation unit which generates a first line obtained by averaging voice data obtained by making a subject repeatedly utter a voice module including a voiced plosive by using a first window length that is set to be equal to or less than a standard utterance time of the voiced plosive; a second line generation unit which generates a second line obtained by averaging the voice data by using a second window length that is set to be equal to or greater than the standard time of the voice module and equal to or less than two times of the standard time; a section detection unit which detects a section in which a value of the first line is greater than a value obtained by multiplying a value of the second line by a prescribed positive real number; and a determination unit which determines an articulation disorder on the basis of the detection result of the section detection unit.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an apparatus and a method for detecting an articulation disorder. [Background technology]

[0002] When motor disorders occur due to strokes, traffic accidents, etc., the commands to move the muscles of the lips, tongue, etc. do not work properly, which can lead to problems with pronunciation (hereinafter referred to as articulation disorders). With such articulation disorders, the tongue does not move accurately, so sounds sound connected overall, and the rhythm and speed are disrupted. This can make conversations, etc., unclear.

[0003] As an example of an apparatus for detecting speech disorders and the like in a subject, Patent Document 1 discloses a configuration for estimating the subject's disease or symptoms based on features extracted from voice data uttered by the subject and a spectrogram image generated from the voice data. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 6854554 Summary of the Invention [Problem to be solved by the invention]

[0005] If such speech disorders could be detected quickly, treatment options would increase and symptoms would likely be alleviated, so there is a demand for a device that can quickly detect symptoms of speech disorders.

[0006] An object of the present disclosure is to provide an articulation disorder detection device and an articulation disorder detection method that are capable of quickly detecting articulation disorders. [Means for solving the problem]

[0007] The speech disorder detection device according to the present disclosure comprises: a first line generating unit that generates a first line by averaging voice data obtained by repeatedly uttering a voice module including a voiced plosive by a subject using a first window length that is set to be equal to or shorter than a standard utterance time of the voiced plosive; The audio data is converted into standard time data by the audio module. + a specified time length a second line generating unit that generates an averaged second line using a second window length set to an interval detection unit that detects an interval in which the value of the first line is greater than a value obtained by multiplying the value of the second line by a predetermined positive real number; A determination unit that determines an articulation disorder based on a detection result of the section detection unit; Equipped with.

[0008] The method for detecting an articulation disorder according to the present disclosure includes: A method for detecting an articulation disorder in an articulation disorder detection device, comprising: The articulation disorder detection device A step of averaging speech data obtained by repeatedly uttering a speech module including a voiced plosive by a subject using a first window length set to be equal to or shorter than a standard utterance time of the voiced plosive to generate a first line; The articulation disorder detection device The audio data is converted into standard time data by the audio module. + a specified time length generating a second averaged line using a second window length set to The articulation disorder detection device detecting an interval in which the value of the first line is greater than a value obtained by multiplying the value of the second line by a predetermined positive real number; The articulation disorder detection device determining whether or not an articulation disorder exists based on the detection result of the section; has.

[0009] According to the present disclosure, speech disorders can be quickly detected. [Brief description of the drawings]

[0010] [Figure 1] 1 is a block diagram showing an example of the configuration of an articulation disorder detection device according to an embodiment of the present disclosure. [Diagram 2] FIG. 13 is a diagram showing an example of a voice waveform when a subject speaks a voice module. [Diagram 3] FIG. 11 is a diagram showing an example of a first line generated based on an audio waveform. [Figure 4] FIG. 11 is a diagram showing an example of a second line generated based on a first line. [Diagram 5] FIG. 13 is a diagram showing an example of detection values ​​based on a first line and a second line. [Figure 6A] 11 is a diagram showing an example of a detection value in a case where a voiced explosive sound cannot be produced in the voice module. FIG. [Figure 6B] FIG. 13 is a diagram showing an example of a detection value in a case where a voiced plosive cannot be produced in the speech module. [Figure 7] FIG. 13 is a diagram showing an example of a first line, a second line, and detection values ​​when a subject repeats a voice module eight times. [Figure 8A] FIG. 1 is a diagram showing an example of a spectrogram image of a healthy subject. [Figure 8B] FIG. 1 is a diagram showing an example of a spectrogram image of a subject with an articulation disorder. [Figure 9] 4 is a flowchart showing an example of a detection control operation in the articulation disorder detection device. [Figure 10A] FIG. 11 is a diagram showing an example of the positions of normal data and abnormal data in a two-dimensional coordinate system in weak anomaly detection; [Figure 10B] 10B is a diagram for explaining the positional relationship between input data, normal data, and abnormal data in the two-dimensional coordinate system of FIG. 10A. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] (Embodiment) DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present disclosure will be described in detail with reference to the drawings. Fig. 1 is a block diagram showing an example of the configuration of an articulation disorder detection device 100 according to an embodiment of the present disclosure.

[0012] As shown in Fig. 1, the articulation disorder detection device 100 is a device that detects articulation disorders based on speech data uttered by a subject. The articulation disorder detection device 100 includes a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, and an input / output circuit (not shown), and detects articulation disorders in a subject based on a preset program.

[0013] The articulation disorder detection device 100 includes a speech waveform generation unit 110, a first line generation unit 120, a second line generation unit 130, a section detection unit 140, an image generation unit 150, a storage unit 160, and a determination unit 170.

[0014] The voice waveform generating unit 110 acquires voice data when the subject repeatedly utters a predetermined voice module a predetermined number of times, and generates a voice waveform based on the acquired voice data.

[0015] The predetermined sound module is a vocalization element for making the subject vocalize a plurality of sounds in succession. For example, the sound module may be configured to include a first syllable that is a voiced plosive sound and a voiced flap sound that follows the first syllable.

[0016] A voiced plosive is a sound that is relatively easy to pronounce even for a subject with an articulation disorder, i.e., a subject with tongue paralysis, and in this embodiment, it is "de". A voiced flap is a sound that is difficult for a subject with tongue paralysis to pronounce, and in this embodiment, it is "re". Therefore, in this embodiment, "dere", which is a combination of "de" and "re", becomes the speech module.

[0017] For example, when a subject utters "dere", the sound pressure level of the original speech changes over time as a vibration waveform, as shown in Figure 2. The left part of Figure 2 is the waveform corresponding to the voiced plosive "de", and the right part is the waveform corresponding to the voiced explosive "re".

[0018] The sound pressure level of each sound fluctuates so as to decrease after reaching a peak value. Also, because the subject utters a voiced plosive and a voiced elastic sound in succession, the sound pressure level changes so as to decrease slightly from the peak value of the voiced plosive and then increase again toward the peak value of the voiced elastic sound.

[0019] Also, FIG. 2 shows an example in which the peak value of a voiced plosive is larger than the peak value of a voiced elongated sound, since a voiced plosive is easier to pronounce than a voiced elongated sound.

[0020] The voice waveform generating unit 110 acquires an original voice in which the subject repeatedly utters "dere" for example eight times, and generates time-series data of the sound pressure level of the original voice as a voice waveform. Therefore, time-series data in which eight voice waveforms are arranged as shown in Fig. 2 is generated (not shown).

[0021] The first line generating unit 120 generates a first line for determining whether or not one sound is uttered by the subject in the standard utterance time.

[0022] The first line is an envelope formed of values ​​calculated as the root mean square (RMS) of the sound pressure level in the audio data for each first window length.

[0023] The standard utterance time is, for example, the time during which the subject can naturally utter one sound without any constraints, and can be set to any time. In this embodiment, the standard utterance time is set to, for example, 175 ms. In other words, in this embodiment, the time required for the subject to utter a voice module consisting of two sounds once is set to 350 ms, which is twice the standard utterance time.

[0024] The first window length is set to be equal to or shorter than the standard speaking time. If the subject speaks one sound accurately within the standard speaking time, the peak value of the sound pressure level will be present within the standard speaking time. For example, if the first window length is set to be longer than the standard speaking time, the peak values ​​of two sounds may fall within one first window length. Therefore, by setting the first window length to be equal to or shorter than the standard speaking time, it is possible to reliably extract the peak value of the sound pressure level when the subject speaks one sound.

[0025] In this embodiment, the first window length is set to, for example, ⅓ or more of the standard speaking time (approximately 64 ms) in consideration of fluctuations in the speaking of one sound within the standard speaking time.

[0026] Since sound pressure level is typically data that repeats small vibrations (see Figure 2), by taking the root mean square of the sound pressure level within the standard utterance time, the first line becomes a line that shows the general shape of the sound pressure level within the standard utterance time, as shown in Figure 3. Specifically, the first line shows a line that connects a part having a peak value corresponding to a voiced plosive sound and a part having a peak value corresponding to a voiced explosive sound.

[0027] More specifically, the first line generating unit 120 sets a first window length based on the standard speaking time, and generates a first line by averaging the audio waveform generated by the audio waveform generating unit 110 using the set first window length.

[0028] The second line generating unit 130 generates a second line that serves as a threshold for determining whether the subject can accurately pronounce a voice module within the standard time of the voice module (twice the standard pronunciation time).

[0029] Since the voice modules are repeatedly uttered by the subject, the voice waveform is considered to show a time change in which the waveform related to the utterance of each voice module is repeated at a constant cycle.

[0030] For example, as shown in FIG. 4, an example is shown in which after the utterance of a given voice module ends and the sound pressure level drops, the utterance of the next voice module starts.

[0031] When the subject accurately pronounces a voice module a predetermined number of times, it is considered that one voice module is pronounced at every fixed period, so that fixed period is set as the second window length, and the second line generating unit 130 generates a second line by averaging the voice waveform using the second window length.

[0032] The second line is an envelope formed of values ​​calculated as the root mean square of the sound pressure level in the audio data for each second window length.

[0033] The second window length is set, for example, to a time greater than or equal to the standard time and less than or equal to twice the standard time, so as to include the period when the healthy subject is speaking and the period between two voice modules (the period when he or she is not speaking).

[0034] For example, the standard time is the time length for the subject to utter the two sounds "de" and "re", and corresponds to twice the standard utterance time (350 ms). The second window length can be set, for example, to a value obtained by adding the standard time and the time length (any length of time) for the subject to naturally start uttering the next voice module after finishing uttering a given voice module. In this embodiment, the any length of time is set, for example, to 150 ms. That is, in this embodiment, the second window length is set to 500 ms, which is the sum of 350 ms and 150 ms.

[0035] Within the second window length, there are periods when sound is produced and periods when sound is not produced, so the second line, which is the average of these periods, is a line that indicates a value lower than the peak value of a single sound on the first line, i.e., a threshold line that serves as the criterion for determining the presence of a peak value.

[0036] The section detection unit 140 compares the first line with the second line, and detects a section in which the value of the first line is greater than the value obtained by multiplying the value of the second line by a predetermined positive real number. The predetermined positive real number may be, for example, 1, or may be a value that can be appropriately set according to the method of section detection, such as a, which will be described later. In the following description, the value obtained by multiplying the value of the second line by a predetermined positive real number will simply be referred to as the value of the second line.

[0037] For example, when the voice modules are correctly pronounced, the value of the first line will be greater than the value of the second line during the period when "dere" is pronounced due to the presence of peak values ​​of the sound pressure levels of each sound. On the other hand, during the period from the end of the pronunciation of one voice module to the start of the pronunciation of the next voice module, no voice is pronounced, so the value of the first line will be smaller than the value of the second line. Note that in the following, the detection value when the value of the first line is greater than the value of the second line is set to 1, and the detection value when the value of the first line is equal to or less than the value of the second line is set to 0.

[0038] Therefore, as shown in Figure 5, when a voice module is spoken correctly, sections with a detection value of 1 are detected the same number of times as the voice module is repeated, and the length of each section is approximately constant.

[0039] However, when one of the sound modules is not produced, the value of the first line is more likely to be equal to or less than the value of the second line, and therefore, at least, the length of the section where the value is 1 is shorter than when the sound module is produced correctly. For example, as described above, a subject with an articulation disorder has difficulty producing voiced sounds, so when he or she produces "dere", "re" is not produced correctly and the length of the section is likely to be shorter than the standard time.

[0040] For example, as shown in FIG. 6A, if a subject with an articulation disorder can pronounce the voiced plosive "de" relatively accurately but cannot pronounce the voiced elongated "re", the sound pressure level of the voiced elongated part on the first line will decrease. Therefore, the length of the section where the value is 1 will be shorter than the detection value of a healthy subject, for example. The detection value of a healthy subject is the detection value when a healthy subject without an articulation disorder pronounces the voice module accurately.

[0041] Also, if the subject prolongs the first "de" sound, the length of the section that becomes 1 will exceed the pronunciation time of one speech module. For example, a subject with an articulation disorder can relatively easily pronounce a voiced plosive, but when repeating speech modules, the tongue may not move well and the subject may prolong the "de" sound in one speech module. In this case, the length of the section is likely to be longer than the standard time.

[0042] For example, as shown in Fig. 6B, when a subject with an articulation disorder elongates the voiced plosive "de" and then utters the voiced volute "re", the voiced plosive portion on the first line becomes longer. Therefore, the length of the section that becomes 1 becomes longer than the detection value of a healthy subject, for example.

[0043] In this way, by detecting sections where the value is 1, the section detection section 140 can determine the number of sections where the value is 1 and the length of the sections.

[0044] The section detection unit 140 may detect whether the value of the first line is larger than the value of the second line using the following equation (1), or may detect whether the value of the first line is larger than the value of the second line using a method other than equation (1) (for example, a method of simply comparing the value of the first line with the value of the second line).

[0045] Detection value = {1, (Rms1>a×Rms2) AND (Rms1>0.3×Rms3), 0, otherwise (1)

[0046] In formula (1), Rms1 is the value of the first line, and Rms2 is the value of the second line. Rms3 is the average value of the first line. a is an adjustment coefficient that takes into account the variation of each audio module. Specifically, a is changed within a range of, for example, 0.3 or more and 0.9125 or less, and a value that minimizes the evaluation function of the following formula (2) may be selected.

[0047] F(a)=std(dur)+std(interval)+100×Erest...(2)

[0048] std(dur) is the standard deviation of the detected interval length (dur), std(interval) is the standard deviation of the distance (interval) between the center positions of the detected intervals, and Erest is the energy of the undetected interval.

[0049] In formula (1), if (Rms1>a×Rms2) AND (Rms1>0.3×Rms3) are satisfied, the detection value is 1, and otherwise the detection value is 0.

[0050] In this way, by applying the adjustment coefficient that best suppresses the variation for each section to equation (1), even if there is a section with a larger degree of variation than others, the section can be detected while suppressing the degree of variation.

[0051] When the subject repeats the voice module eight times, the actual data of the first line (Rms1) and the second line (Rms2) are, for example, as shown in Figure 7. In Figure 7, the first line (Rms1) and the second line (Rms2) have relatively different shapes. The time of each voice module is in the range indicated by T1, T2, T3, T4, T5, T6, T7, and T8.

[0052] In the first line shown in Fig. 7, there are two rising edges, "de" and "re", at each of T1, T2, T3, T4, T5, T6, T7, and T8. In addition, in the first line shown in Fig. 7, there are three rising edges at T2, and only one rising edge at T7.

[0053] The section detection unit 140 compares the value of the first line with the value of the second line from the data shown in Fig. 7 using, for example, equation (1), and detects a section that is 1. In the example shown in Fig. 7, eight sections are detected, although there is variation in the length of each section.

[0054] The image generating unit 150 generates a spectrogram image based on the original sound. Specifically, the image generating unit 150 performs a Fourier transform on the original sound in the section detected by the section detecting unit 140 to generate a spectrogram image.

[0055] For example, since the length of one section is equivalent to two standard utterance times (the utterance time of one voice module), when a voice module is uttered accurately, the spectrogram image corresponding to the voice of one voice module has a substantially uniform width. When a voice module is uttered accurately, the spectrogram images corresponding to each section are likely to be generated with substantially uniform widths.

[0056] When a voice module is uttered correctly, for example, as shown in FIG. 8A, eight sections S1, S2, S3, S4, S5, S6, S7, and S8 detected by section detection unit 140 have approximately equal widths.

[0057] In addition, because the sound pressure level rises for each of the two sounds "de" and "re," the sound pressure level drops in the area between the two sounds. Therefore, it can be confirmed that the spectrogram image has vertical lines separating each sound in each section.

[0058] However, if one of the two sounds in the speech module is not produced, the corresponding spectrogram image will be shorter than if the speech module was produced correctly. Also, if the subject is unable to produce the sound of the speech module well and produces the first "de" sound by elongating it, the corresponding spectrogram image will be longer than if the speech module was produced correctly.

[0059] In addition, because subjects with speech disorders have difficulty moving their tongues, when they repeatedly pronounce a sound module, they may elongate the voiced plosive sound "de", which is easy to pronounce, or they may begin pronouncing the next sound module without pronouncing the voiced plosive sound "re", which is an easy sound, resulting in a mixture of long and short spectrogram images being generated.

[0060] When a subject with an articulation disorder speaks a voice module, for example, as shown in Figure 8B, a spectrogram image is generated in which the eight sections S11, S12, S13, S14, S15, S16, S17, and S18 detected by the section detection unit 140 are relatively long and have large variations in width.

[0061] In the example shown in Figure 8B, the subject with an articulation disorder is unable to pronounce the "re" sound that follows the relatively easy-to-pronounce "de," resulting in a spectrogram image in which the vertical lines that appear between the two sounds in each section, as shown in Figure 8B, are difficult to recognize.

[0062] In this way, a clear difference occurs in the spectrogram images between when a speech is uttered correctly and when it is uttered incorrectly.

[0063] Therefore, in this embodiment, a machine learning model is used to determine whether a subject has an articulation disorder. Specifically, the learning method used in this embodiment may be, for example, an autoencoder.

[0064] The autoencoder is a machine learning model that is trained so that the output data is the same as the input data. In this embodiment, the spectrogram image of the speech module of a healthy person is trained, and when an abnormal image is input, the image cannot be appropriately restored, and the presence or absence of an articulation disorder is determined by using this fact.

[0065] The storage unit 160 stores a machine learning model that has been trained using spectrogram images (eg, images as shown in FIG. 8A) based on the original speech of a plurality of healthy subjects as training data.

[0066] The determination unit 170 calculates a score for each spectrogram image of each section detected by the section detection unit 140 based on the machine learning model stored in the storage unit 160. Specifically, a spectrogram image (input image) related to the original voice of the subject is input to an autoencoder, and an output image is output from the autoencoder. Then, the determination unit 170 calculates a difference image between the input image and the output image.

[0067] To calculate the difference image, the determination unit 170 may, for example, square all pixel values ​​of the difference image and calculate a mean square error (score) by averaging all the squared pixel values. The determination unit 170 may compare the calculated mean square error with a preset threshold value, and determine that the person has an articulation disorder if the mean square error is equal to or greater than the threshold value.

[0068] For example, the difference image between a spectrogram image with wide intervals as shown in FIG. 8B and a spectrogram image of a healthy subject as shown in FIG. 8A will be relatively large due to the difference in the interval widths.

[0069] Therefore, by comparing the score of the difference image with a predetermined threshold, an articulation disorder can be detected.

[0070] In this way, a machine learning model trained on spectrogram images of healthy individuals can be used to easily determine whether the subject's spectrogram image accurately represents the voice module.

[0071] Next, a description will be given of an example of the operation of the articulation disorder detection device 100. Fig. 9 is a flowchart showing an example of the detection control operation of the articulation disorder detection device 100. The process in Fig. 9 is started, for example, when the subject starts to utter the voice module.

[0072] 9, the articulation disorder detection device 100 acquires an original speech of a subject and generates a speech waveform (step S101). After step S101, the articulation disorder detection device 100 generates a first line (step S102) and a second line (step S103) based on the speech waveform.

[0073] After step S103, the speech disorder detection device 100 compares the first line with the second line, and detects a section in which the value of the first line is greater than the value of the second line (step S104). Next, the speech disorder detection device 100 generates a spectrogram image for each detected section (step S105).

[0074] After step S105, the articulation disorder detection device 100 calculates a difference image between the generated spectrogram image and a spectrogram image of a healthy subject in the learning model, and determines whether the score of the difference image is equal to or greater than a predetermined threshold (step S106).

[0075] As a result of the determination, if the score of the difference image is equal to or greater than the predetermined threshold (step S106, YES), the articulation disorder detection device 100 detects an articulation disorder (step S107). On the other hand, if the score of the difference image is less than the predetermined threshold (step S106, NO), the articulation disorder detection device 100 determines that the person does not have an articulation disorder (step S108). After step S107 or step S108, this control ends.

[0076] According to the present embodiment configured as described above, speech disorders are detected using the first and second lines generated from a speech waveform based on an original speech in which a subject repeatedly speaks a speech module whose first syllable is a voiced plosive a predetermined number of times.

[0077] In other words, by having the subject repeat a voice module starting with a voiced plosive, which is relatively easy to pronounce, a predetermined number of times, it is possible to make it easier to identify the peak value of the sound pressure level of a voice waveform based on a voiced plosive.

[0078] Specifically, by making it easier to set peak values ​​for the voice level, it becomes easier to detect sections on the first line that are larger than the second line, making it easier to detect speech disorders based on the detected sections.

[0079] Specifically, a spectrogram image based on that section is generated, and the image of the subject within that section is compared with a normal image based on a healthy subject, making it easier to detect speech disorders.

[0080] In the case of a subject with an articulation disorder, the length and number of sections to be detected, as well as the spectrogram image, are likely to differ from those of a healthy subject, making it easy to detect whether or not the subject has an articulation disorder.

[0081] In addition, since a voice module having a voiced plosive, which is relatively easy to pronounce, as the first syllable is used, it is possible to easily detect the peak value of the sound pressure level. For example, in a configuration in which detection is performed using a sound pressure module with no constraints, the detection results may differ depending on the contents of the voice module, and it may not be possible to perform detection quickly and accurately.

[0082] In contrast, in the present embodiment, a voice module that is easy to detect peak values ​​of sound pressure level is used, so that it is possible to quickly and accurately detect whether or not a person has an articulation disorder.

[0083] In addition, because speech disorders can be detected quickly and accurately, for example, cooperation with more specialized medical institutions can be quickly established. As a result, for example, by visiting a medical institution when the condition is still in its early stages, treatment can be started early, and the number of treatment options increases, which can lead to the alleviation of symptoms of speech disorders.

[0084] Furthermore, since the first window length is set to be equal to or shorter than the standard speaking time, the peak value of the sound pressure level of each sound can be reliably extracted.

[0085] Furthermore, by setting the first window length to approximately 1 / 3 of the standard speaking time, a first line can be generated that takes into account the fluctuations in the speaking of a single sound, making it easier to extract the peak value of the sound pressure level.

[0086] In addition, since the voice module includes voiced plosives that are easy to pronounce and voiced paponics that are difficult to pronounce, it is possible to easily cause differences in the peak value of the sound pressure level within a standard time corresponding to one voice module. As a result, it is possible to easily cause differences from normal results in the length of the sections, the number of sections, and the spectrogram image. As a result, it is possible to easily detect speech disorders.

[0087] In the above embodiment, an articulation disorder is detected based on a score based on a machine learning model, but the present disclosure is not limited to this, and an articulation disorder may be detected based only on the detection result of the section detection unit 140.

[0088] Specifically, the determination unit 170 may detect an articulation disorder based on the count number of the segments detected by the segment detection unit 140.

[0089] For example, if a subject accurately repeats a sound module eight times, eight sections are detected. However, if a subject with an articulation disorder repeats a sound module in the same way, eight sections are not necessarily detected.

[0090] For example, if a subject with an articulation disorder has difficulty moving his or her tongue when speaking a specific sound module and pronounces the "de" as if it were a drawn-out sound, the section of this sound module may be detected as being connected to the section of the next sound module.

[0091] In addition, subjects with speech disorders may not be able to produce two sounds from one speech module well, and this interval may not be detected.

[0092] In these cases, the section detection section 140 will detect a section that is less than eight.

[0093] In addition, when a subject with an articulation disorder utters a certain voice module, there may be a gap between the utterance of "de" and "re", resulting in two sections being detected for one voice module. In such a case, the section detection unit 140 will detect more than eight sections.

[0094] Therefore, the determination unit 170 may detect an articulation disorder when the count number of the section detected by the section detection unit 140 is not the number of repetitions of the voice module.

[0095] In this way, it is possible to improve the speed with which an articulation disorder can be detected.

[0096] Furthermore, the determining section 170 may determine, based on the length of a section detected by the section detecting section 140, whether or not to count the section.

[0097] As described above, subjects with articulation disorders may produce longer or shorter intervals, which may result in the detection of intervals that significantly exceed the standard time (e.g., see Figure 6B) or that are significantly short of the standard time (e.g., see Figure 6A).

[0098] Therefore, by determining not to count such sections, the determination unit 170 can exclude sections that are clearly affected by an articulation disorder. As a result, it is possible to easily differentiate the count number of sections related to a subject with an articulation disorder from the desired number of sections.

[0099] Furthermore, when detecting an articulation disorder based only on the detection result of the section detection unit 140, the image generation unit 150 and the storage unit 160 do not need to be provided.

[0100] Furthermore, the determination unit 170 may detect an articulation disorder based on both the number of segments counted by the segment detection unit 140 and a score based on a learning model.

[0101] For example, the determination unit 170 performs a primary determination of an articulation disorder based on the count number of the section. In the primary determination, for example, the determination unit 170 determines whether or not to perform a secondary determination based on the count number of the section. The secondary determination is a determination for detecting an articulation disorder based on a score based on a learning model.

[0102] For example, if the count number of the section is not within a predetermined range (for example, a range of 6 to 10 times), the judgment unit 170 judges not to perform the secondary judgment, and judges that there is an articulation disorder based on the result of the primary judgment.

[0103] On the other hand, if the count number of the section is within the predetermined range, the determination unit 170 determines to perform a secondary determination, and performs the secondary determination. In the secondary determination, a spectrogram image or the like is used to perform a detailed determination.

[0104] In this way, a simplified determination is made by the primary determination, and if a judgment cannot be made by the primary determination, a detailed determination is made by the secondary determination, thereby making it possible to improve the efficiency of detection.

[0105] Moreover, the primary determination and the secondary determination may be performed at all times. By performing both the primary determination and the secondary determination, the accuracy of detection can be improved.

[0106] Furthermore, the result of the primary determination may be used to perform the determination again. For example, when there is a problem with the speech acquisition in the articulation disorder detection device 100, and the count number of the section in the primary determination is clearly incorrect, the accuracy of detection can be improved by performing the determination again.

[0107] Furthermore, in the above embodiment, an autoencoder technique is used to detect an articulation disorder, but the present disclosure is not limited to this, and an articulation disorder may be detected using a technique other than an autoencoder.

[0108] As an example of a method other than autoencoder, there is a method for weak anomaly detection.

[0109] Weak anomaly detection is a method that uses both normal and anomalous data for learning. In weak anomaly detection, a metric learning method is used, and learning is carried out so that the positions of the features of the normal data and the features of the anomalous data become farther apart from each other.

[0110] Specifically, a DNN (Deep Neural Network) model is trained so that the "distance" between two feature vectors reflects the "similarity" of the data. For example, the DNN model is trained so that the distance between feature vectors is small for samples that belong to similar classes, and large for samples that belong to dissimilar classes.

[0111] In the present embodiment, as the learning of the DNN model progresses, the DNN model is trained so that, for example, data on the speech of healthy subjects (normal data) are concentrated in one place. The DNN model is trained so that data on the speech of subjects with dysarthria (abnormal data) are distant from the data group on the speech of healthy subjects.

[0112] For example, as shown in FIG. 10A, in a two-dimensional coordinate system with the horizontal axis being X and the vertical axis being Y, the DNN model is trained so that normal data is located at a certain location, and the first abnormal data and the second abnormal data are located at a distance from the normal data.

[0113] By treating this distance as the degree of abnormality, the determining unit 170 detects whether or not there is an articulation disorder.

[0114] For example, as shown in FIG. 10B, when input data is within the range of normal data, the data is positioned within the range of normal data in the two-dimensional coordinate system (see the white circle).

[0115] On the other hand, if the input data is close to the first abnormal data, the data will be located close to the first abnormal data (see black circle). Also, if the input data is close to the second abnormal data, the data will be located close to the second abnormal data (see black square).

[0116] In this way, by treating the distance between the input data and normal data as the degree of abnormality, it is possible to detect whether or not a person has an articulation disorder.

[0117] In addition, as a method other than the autoencoder, for example, a time domain anomaly detection method can be mentioned.

[0118] The time domain anomaly detection method uses a speech rate and a standard deviation of the center interval of a section to obtain a predetermined index A as a score of the degree of anomaly. The predetermined index A is calculated, for example, using the following formula (3).

[0119] A = 1 / Mpsec + wσ (3)

[0120] Mpsec is the average number of times the subject utters the voice module and the average speech speed when the subject speaks continuously. σ is the standard deviation of the center interval of the sections detected by the section detection unit 140. w is a weighting coefficient, and is a value determined experimentally (e.g., 2, etc.). The center interval of the sections is the center interval between adjacent sections.

[0121] According to a technology that uses Mpsec and σ to evaluate the impression of speech ("Evaluation index of acoustic features that determine the elemental sensations received from speech", Nishida Yuu and 3 others, FIT2017, 16th Forum on Information Science and Technology, J-018, pp.377-378), it is possible to evaluate the degree to which a voice is perceived as clear using Mpsec, and the degree to which a voice is perceived as having intonation using σ.

[0122] Using such index A, for example, when σ becomes large, the variance of the center interval of the section becomes large, and it becomes possible to detect that the rhythm of the subject's vocalization by the voice module is disturbed. In other words, the allowable range of index A is set based on healthy subject data, and the judgment unit 170 can detect an articulation disorder when the index calculated from the subject's data falls outside this allowable range.

[0123] In the above embodiment, the voice module is composed of two sounds, a voiced plosive and a voiced elastic sound, but the present disclosure is not limited to this, and may be composed in any way as long as it includes a voiced plosive, for example, it may not include a voiced elastic sound, or may be composed of three or more sounds. However, since it is easier to make a difference in the detected section by having the subject speak with varying strengths and weaknesses in the voice module, it is preferable that the voice module includes two sounds, a voiced plosive and a voiced elastic sound, in succession.

[0124] In addition, in the above embodiment, the subject is made to repeatedly speak the voice module eight times, but the present disclosure is not limited to this, and the subject may be made to repeatedly speak a number of times other than eight.

[0125] In the above embodiment, the first window length is set to be equal to or shorter than the standard speaking time and equal to or longer than 1 / 3 of the standard speaking time, but the present disclosure is not limited to this, and the lower limit of the first window length may be set arbitrarily as long as it is equal to or shorter than the standard speaking time. In the above embodiment, the second window length may be set to be longer than the first window length, and may be set appropriately so that the section can be detected well. In addition, the window function may be a window function such that both ends are smoothly attenuated, such as a Hanning window.

[0126] Furthermore, in the above embodiment, the voiced plosive sound is "de", but the present disclosure is not limited to this, and any sound may be used as long as the vowel in the syllable is e and it is relatively easy to pronounce, such as "be" or "ge".

[0127] In addition, in the above embodiment, the voiced plosive is the first syllable of the speech module, but the present disclosure is not limited to this, and it does not have to be the first syllable of the speech module.

[0128] Furthermore, in the above embodiment, the voiced flute is "re", but the present disclosure is not limited to this, and any sound may be used as long as it is a sound that is relatively difficult to pronounce.

[0129] In addition, in the above embodiment, the first line and the second line are calculated by calculating the root mean square of the audio waveform, but the first line and the second line may also be calculated by averaging the audio waveform using a different method.

[0130] In addition, the above-mentioned embodiments are merely examples of the embodiment of the present disclosure, and the technical scope of the present disclosure should not be interpreted as being limited by them. In other words, the present disclosure can be embodied in various forms without departing from the gist or main features thereof. [Industrial Applicability]

[0131] The speech disorder detection device and speech disorder detection method of the present disclosure are useful as a speech disorder detection device and a speech disorder detection method capable of quickly detecting speech disorders. [Explanation of symbols]

[0132] 100 Articulation disorder detection device 110 Audio waveform generator 120 1st line generation section 130 Second line generation section 140 Section detection unit 150 Image Generation Unit 160 Storage area 170 Judgment section

Claims

1. a first line generating unit that generates a first line by averaging voice data obtained by repeatedly uttering a voice module including a voiced plosive by a subject using a first window length that is set to be equal to or shorter than a standard utterance time of the voiced plosive; a second line generating unit that generates a second line by averaging the voice data using a second window length that is set to a value obtained by adding a predetermined time length to a standard time of the voice module; an interval detection unit that detects an interval in which the value of the first line is greater than a value obtained by multiplying the value of the second line by a predetermined positive real number; A determination unit that determines an articulation disorder based on a detection result of the section detection unit; An articulation disorder detection device comprising:

2. an image generating unit that generates an image based on the audio data; A storage unit for storing a learning model trained using images based on original speech of a plurality of healthy subjects as training data; Equipped with the determination unit calculates a score for each section detected by the section detection unit based on the learning model, and determines the articulation disorder based on the calculated score. The speech disorder detection device according to claim 1 .

3. The determination unit determines the articulation disorder based on a count number of the sections detected by the section detection unit. The speech disorder detection device according to claim 1 or 2.

4. The determination unit determines whether or not to count the section based on the length of the section detected by the section detection unit. The speech disorder detection device according to claim 3 .

5. the first line generating unit sets the first window length to ⅓ or more of the standard speaking time; The speech disorder detection device according to any one of claims 1 to 4.

6. the voiced plosive is the initial syllable of the speech module; The speech disorder detection device according to any one of claims 1 to 5.

7. The speech module includes the voiced plosive and a continuous voiced explosive. The speech disorder detection device according to any one of claims 1 to 6.

8. The vowel of the voiced plosive syllable is / e / . The speech disorder detection device according to any one of claims 1 to 7.

9. A method for detecting an articulation disorder of an articulation disorder detection apparatus, comprising: generating a first line by averaging speech data obtained by repeatedly uttering a speech module including a voiced plosive by the subject using a first window length set to be equal to or shorter than a standard utterance time of the voiced plosive, using the speech disorder detection device; generating a second line by averaging the speech data using a second window length set to a value obtained by adding a predetermined time length to a standard time of the speech module, by the speech disorder detection device; detecting, by the articulation disorder detection device, a section in which the value of the first line is greater than a value obtained by multiplying the value of the second line by a predetermined positive real number; determining an articulation disorder based on a detection result of the section by the articulation disorder detection device; The method for detecting an articulation disorder comprises:

Images