JP5997369B2 - Rehabilitation support device and control method thereof - Google Patents

Description

  The present invention relates to a rehabilitation support apparatus that supports rehabilitation such as attention disorder.

  In language rehabilitation for attention disorder, the task of concentrating on one thing, such as solving a calculation problem or solving a national language problem, is given.

  In this field, there are some proposals related to a voice output device that replaces intention communication by sign language or writing (see, for example, Patent Document 1), and there is no suitable device for supporting language rehabilitation for attention disorder.

JP 2007-293395 A

  An example of selective attention in language rehabilitation for attention disorder is the cocktail party effect. The cocktail party effect can be achieved only by paying attention to a specific voice from multiple voices, so it is thought that it will be a good exercise for attention disorder, but as mentioned above, there is no appropriate training device at present. It is. Therefore, providing a device that supports language rehabilitation for attention disorder is an important requirement for effective training.

  Other objects and advantages will become apparent from the following description.

  According to one aspect of the present invention, a rehabilitation support apparatus for supporting rehabilitation of attention disorder, a voice data storage unit that stores a plurality of types of voice data, and a noise data storage unit that stores a plurality of types of noise data Noise generating means for creating mixed noise data obtained by selecting and mixing one or more pieces of noise data from the noise data storing means, and voice data selected from the voice data storing means by the noise generating means Degraded voice creating means for creating degraded voice data by superimposing the created mixed noise data, and acoustic output means for acoustically outputting the created degraded voice data, wherein the noise creating means includes the noise data It is possible to adjust at least one of the number of noise data to be selected and mixed from the storage means and the gain of the mixed noise data. Rehabilitation supporting device according to claim is provided.

  According to the present invention, language rehabilitation for attention disorder can be effectively performed.

  Other features and advantages of the present invention will become apparent from the following description with reference to the accompanying drawings. In the accompanying drawings, the same or similar components are denoted by the same reference numerals.

The accompanying drawings are included in the specification, constitute a part thereof, show an embodiment of the present invention, and are used to explain the principle of the present invention together with the description.
The figure which shows the external appearance structure of the robot in embodiment. The block diagram which shows the internal structure of the robot in embodiment. The figure which shows the example of the module structure of the program for rehabilitation in embodiment. The figure which shows the structural example of the data hold | maintained in the patient database in embodiment. The flowchart of the rehabilitation assistance process in embodiment. The figure which shows the example of the home screen in the rehabilitation assistance process in embodiment. The figure which shows the example of the sound source selection screen in embodiment. The figure which shows the example of the cocktail party mode setting screen in embodiment. The figure which shows the example of the rehabilitation execution screen in embodiment. The flowchart which concerns on the modification of the rehabilitation execution process in cocktail party mode. The figure which shows the example of the corresponding | compatible table of the difficulty, the number of people, and a noise level in embodiment. The figure explaining the example of the band restriction | limiting of the mixed noise data according to the difficulty in embodiment.

  DESCRIPTION OF EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. In addition, this invention is not limited to the following embodiment, It shows only the specific example advantageous for implementation of this invention. Moreover, not all combinations of features described in the following embodiments are indispensable for solving the problems of the present invention.

  FIG. 1 is a diagram illustrating an external configuration of a robot 1 as a language rehabilitation support apparatus according to an embodiment. The robot 1 in this embodiment interacts with a patient and performs language expression for rehabilitation (hereinafter abbreviated as “rehabilitation”) for the patient.

  The robot 1 may have a general computer appearance. However, since the robot 1 performs rehabilitation while interacting with the patient, the robot 1 may have an appearance configuration that allows the patient to relax and become familiar. It will be. The robot 1 includes an antenna 111 for performing wireless communication, for example. Further, the robot 1 is provided with a microphone 114 and a speaker 112 at positions corresponding to human ears and mouths, respectively. In addition, the robot 1 can connect a tablet terminal 150 that is a touch panel type display / input device for use by a speech hearing person or patient via a cable 151. It is assumed that the touch panel of the tablet terminal 150 can detect a tap or a tracing operation with a user's finger. However, such a function of the tablet terminal 150 may be configured in advance in the robot 1 itself.

  FIG. 2 is a block diagram showing the internal configuration of the robot 1. The robot 1 includes the following configuration, including a CPU 101 that controls the entire apparatus, a RAM 102 that functions as a main storage device, and a ROM 103 that stores control programs and fixed data.

  The wireless communication controller 105 controls wireless communication performed via the antenna 111. The HDD 106 is a hard disk device that stores an operating system (OS) 107 and a rehabilitation program 108, and also stores an original sound database (DB) 116, a noise DB 117, and a patient DB 118. The original sound DB 116 stores data recording words, phrases, and sentences that are pronounced in a quiet environment such as an anechoic room. These should be categorized by fields such as greetings, political economy and science. Further, it is desirable that the sound pressure level of each sound data in the original sound DB 116 is normalized. The noise DB 117 stores data of various noises such as white noise and pink noise, which are often used for sound quality evaluation, as well as automobile running sound, factory noise, and birds and animals. In addition, a plurality of types (for example, 20 types) of human roaring data for forming the noise of a person simulating a cocktail party are stored. It is desirable that the sound pressure level (noise level) of each noise data in the noise DB 117 is also normalized. The patient DB 118 holds various information including personal information of a patient who is a subject, a failure type, and a severity thereof.

  An interface (I / F) 109 connects the tablet terminal 150 via a cable 151. The audio controller 110 includes an A / D converter, a D / A converter, an antialiasing filter, and the like (not shown), and performs audio output using the speaker 112 and audio input from the microphone 114.

  FIG. 3 shows an example of the module configuration of the rehabilitation program 108. The patient registration / search module 121 is a function module related to new registration processing in the patient DB 118 and patient search processing from the patient DB 118. The rehabilitation support main module 123 is responsible for executing rehabilitation by the robot 1. The speech synthesis module 124 performs speech synthesis when performing acoustic output during rehabilitation by the robot 1. The voice recognition module 125 recognizes a patient's utterance. This speech recognition has not only a word recognition function but also a continuous speech recognition function capable of recognizing sentences.

  FIG. 4 is a diagram illustrating a data structure example of the patient DB 118. As shown in the figure, each data entry is given a unique patient ID and describes the patient's personal information such as name, date of birth, etc., as well as the disorder type and its severity, and rehabilitation history. The The disorder type includes, for example, attention disorder and aphasia. Further, here, for example, 1 is set for mild, 2 is set for moderate, and 3 is set for severe. The rehabilitation history includes the date and time when rehabilitation was performed in the past, the number of voices presented, the number of times, and the like. By referring to the rehabilitation history, it is easier for a language hearing person to plan future rehabilitation.

  FIG. 5 is a flowchart of the rehabilitation support process in the present embodiment. A program corresponding to this flowchart is included in the rehabilitation program 108, loaded into the RAM 102, and executed by the CPU 101. When this program is executed, a home screen as shown in FIG. As illustrated, the home screen includes a patient registration button 601, a patient selection button 602, and a rehabilitation start button 603. When a user (for example, a language auditor) taps any button, the corresponding screen can be displayed.

  The patient registration or selection is performed by tapping the patient registration button 601 or the patient selection button 602 (S1). Since details of patient registration and selection are not directly related to the present invention, illustration of these screen examples is omitted. At the time of registration, items as shown in FIG. 4 are described. When registration or selection is completed, the screen returns to the home screen.

  In S2, it waits for the rehabilitation start button 603 to be tapped. When the rehabilitation start button 603 is tapped, the process proceeds to a process (S3) for selecting an original sound and noise. Here, the screen changes to a sound source selection screen as shown in FIG. When the user taps the original audio selection button 701 on the sound source selection screen of FIG. 7, a candidate window 702 is displayed, and a desired genre can be selected from the candidate window 702. When “0: automatic selection” is selected, the selection of the original sound is left to the apparatus side. At this time, the robot 1 may select the original sounds in a predetermined order or may select them at random. Next, when the user taps the noise type selection button 703, a candidate window 704 is displayed, from which a desired noise type can be selected. When “0: automatic selection” is selected, the robot 1 selects noise in a predetermined order or at random, as described above.

  When the user taps the cocktail party mode selection button 705, the robot 1 invalidates the noise selection by the noise type selection button 703, and enters a mode for discriminating the cocktail party effect (that is, selective listening of voice). Processing in this cocktail party mode will be described later.

  When the cancel button 706 is tapped, the input content on this screen is discarded and the screen returns to the home screen in FIG. On the other hand, when the next button 707 is tapped, the screen transitions to the next screen according to the input content on this screen. Specifically, the ON / OFF state of the cocktail party mode selection button 705 is confirmed in S4. If the cocktail party mode selection button 705 is OFF, the process proceeds to S5. In S5, voice data is generated by superimposing the selected original sound data and noise data. In this specification, sound data obtained by superimposing noise data on original sound data is referred to as “degraded sound data”. At this time, the screen transitions to a rehabilitation execution screen as shown in FIG. When the sound generation execution button 901 is tapped on the rehabilitation execution screen of FIG. 9, the CPU 101 passes the created deteriorated sound data to the sound controller 110 and outputs the sound via the speaker 112 (S9). The patient listens to the output deteriorated sound, for example, and repeats the sound. Alternatively, the speech auditor may ask questions about the topic of the output degraded speech and have the patient answer it. Listening to degraded speech generally requires more concentration than listening to clear original sound. For this reason, listening to such degraded speech is a good exercise for improving attention for patients with attention disorders. In addition, listening to degraded speech with superimposed noise leads to activation of brain activity, which may have a positive effect on attention-impaired patients.

  The speech hearing person can input correct / incorrect information about the patient's answer (S10). For example, if the patient's answer is correct, the correct answer button 903 in FIG. 9 is tapped, and if the answer is incorrect, the incorrect answer button 904 is tapped. Alternatively, the CPU 101 may call the voice recognition module 125 to recognize the utterance related to the patient's answer input via the microphone 114 and automatically determine the correct / incorrect answer. When the correct answer button 903 is tapped by the language auditor, or when the correct answer is determined by speech recognition, the CPU 101 calls the speech synthesis module 124 to generate a synthesized sound “Sure answer!” By the speech synthesis. Depart from. At the same time, the robot 1 may be happy. When the “return” button 905 is tapped, the screen returns to the previous screen. When “next sound” button 906 is tapped, it is determined to repeat the process (NO in S11), and the process returns to S3 and the process is repeated. When the “quit” button 907 is tapped, information on the correct / incorrect answers so far is stored in the patient DB 118 as a rehabilitation history.

  On the other hand, when the cocktail party mode selection button 705 is in the ON state, the process proceeds from S4 to S6, and the processes of S6 to S8 are executed instead of the process of S5. When the process proceeds from S4 to S6, the screen changes to the cocktail party mode setting screen of FIG. In the cocktail party mode setting screen of FIG. 8, when the user taps the “number of people shaking” setting button 801, a candidate window 802 is displayed, from which a desired number of people P can be selected. Also, the user can change the noise level L of the noise data created by sliding the slide bar 803. In this example, the noise level L can be reduced by moving the slide bar 803 to the left, and the noise level L can be increased by moving the slide bar 803 to the right. When the “return” button 804 is tapped, the setting value here is discarded and the screen returns to the previous screen. When the next button 805 is tapped, the setting value here is validated, and the process proceeds to S7. In S7, the CPU 101 obtains and mixes the set P noises from the noise DB 117, and amplifies the mixing result with a gain corresponding to the set noise level L, thereby mixing. Create noise data. Here, the gain corresponding to the noise level L is a gain for achieving the noise level L. If the noise level of each noise data in the noise DB 117 is normalized, the gain can be easily obtained by the ratio of the set noise level L to the original noise level. The amplification by the gain may be performed by multiplying each sample of the digital data by the gain value as described above, and the amplification factor of a variable amplifier (not shown) that can be included in the audio controller 110 or the speaker 112 corresponds to the gain. Those skilled in the art will readily understand that this may be done by setting the value. In the present embodiment, it is assumed that the number of rumbling persons P can be 1. Therefore, the mixed noise in the case of P = 1 is substantially a single type of noise itself. Thereafter, the process proceeds to S9, and the CPU 101 creates degraded voice data by superimposing the mixed noise data created in S8 on the original sound data selected in S3. At this time, the screen changes to the rehabilitation execution screen of FIG. Then, as described above, when the sound generation execution button 901 is tapped on the rehabilitation execution screen of FIG. 9, the CPU 101 passes the created deteriorated sound data to the sound controller 110 and outputs the sound via the speaker 112 ( S9).

  As described above, according to the present embodiment, the robot 1 as the rehabilitation support apparatus has the cocktail party mode, and when creating the environmental sound of the cocktail party, at least one of the number of people noisy and the noise level is selected. It can be adjusted. Increasing the number of people rumbling and increasing the noise level can increase the level of difficulty regarding the intelligibility (whether correctly understood) by the listener of the degraded voice data that has been acoustically output. On the other hand, if the number of people is noisy and the noise level is reduced, the difficulty level can be reduced. Thereby, the cocktail party effect can be efficiently trained for a patient with attention disorder.

<Modification 1>
In the following, a process for automatically creating appropriate deteriorated voice data according to the patient's listening ability in the cocktail party mode will be described. FIG. 10 is a flowchart according to a modification of the rehabilitation execution process in the cocktail party mode. When entering the cocktail party mode, the CPU 101 first sets the difficulty level D (S21). Here, the difficulty level refers to the level of difficulty related to the intelligibility of deteriorated voice data by the listener of the sound output. For example, the initial value of the difficulty level D may be a value according to the severity of the disorder described in the patient DB 118 shown in FIG. That is, as described above, when the severity value is set to 1 for mild, 2 for moderate, and 3 for severe, the initial value of difficulty D corresponds to 1, 2, and 3, respectively. To do.

  Next, the number of people and the noise level are set to values depending on the difficulty level D (S22). That is, the number of people rumbling is represented as P (D), and the noise level is represented as L (D). Here, the robot 1 may hold a correspondence table of the difficulty level, the number of rumors, and the noise level as shown in FIG. According to this correspondence table, for example, when the difficulty level D is 3, the number P of people is calculated as 3, and the noise level L is calculated as 1.

  Next, the CPU 101 acquires and mixes the set P noises from the noise DB 117 and amplifies the mixing result with a gain corresponding to the set noise level L, thereby mixing. Noise data is created (S23). Next, the CPU 101 selects original sound data from the genre selected on the screen of FIG. 7 or at random (S24). Then, the CPU 101 creates degraded sound data by superimposing the mixed noise data created in S23 on the original sound data selected in S24 (S25). At this time, the screen changes to the rehabilitation execution screen of FIG. Then, as described above, when the sound generation execution button 901 is tapped on the rehabilitation execution screen of FIG. 9, the CPU 101 passes the created deteriorated sound data to the sound controller 110 and outputs the sound via the speaker 112 ( S26).

  The patient listens to the output deteriorated sound, for example, and repeats the sound. The speech hearing person inputs correct / incorrect information about the patient's answer (S27). For example, if the patient's answer is correct, the correct answer button 903 in FIG. 9 is tapped, and if the answer is incorrect, the incorrect answer button 904 is tapped. Next, in S28, if the original sound data to be presented still remains, the process returns to S24, and other original sound data is selected and the process is repeated. In this way, the parameter of the determination result for calculating the correct answer rate is accumulated.

  In S29, the correct answer rate CR is calculated. It is determined whether or not the calculated accuracy rate CR exceeds the first threshold value TH1 (S30). When the correct answer rate CR exceeds the first threshold value TH1 (YES in S30), it can be evaluated that the questions were easy for the patient. Therefore, in this case, the difficulty level D is increased by one level (S31). Here, unless the “quit” button 907 in FIG. 9 is tapped and there is no end interrupt, the process returns to S22 and the process is repeated.

  Next, when the correct answer rate CR falls below the second threshold value TH2 that is lower than the first threshold value TH1 (YES in S32), it can be evaluated that the questions are difficult for the patient. Therefore, in this case, the difficulty level D is lowered by one step (S33). Here, unless the “quit” button 907 in FIG. 9 is tapped and there is no end interrupt, the process returns to S22 and the process is repeated.

  According to the above processing, it is possible to automatically and efficiently set the difficulty level for the patient.

<Modification 2>
In the first modification described above, at least one of the number of mixed noise data and the noise level is changed so that the degree of difficulty is changed according to the calculated accuracy rate. On the other hand, the frequency band of noise may be adjusted so that the difficulty level is changed according to the calculated accuracy rate. Hereinafter, this will be described as a second modification.

  For example, the rehabilitation program 108 further includes a band limiting filter module that can change the frequency band of the mixed noise data. The band limiting filter module forms a variable bandpass filter, for example.

  Here, it is assumed that the difficulty level D is four levels of 1, 2, 3, and 4 in order of ease. FIG. 12 is a schematic diagram of an audio frequency band. The assumed audio frequency band is divided into four as shown in the figure, and F1, F2, F3, and F4 are set in ascending order. It may be an equal division or an unequal division. When the difficulty level D = 1, the frequency band of the mixed noise data is set to F4. In this case, since the band on which the noise is superimposed is limited to only F4, the difficulty of hearing will be low. When the difficulty level D = 2, the frequency band of the mixed noise data is set to F3 + F4. When the difficulty level D = 3, the frequency band of the mixed noise data is set to F2 + F3 + F4. When the difficulty level D = 4, the frequency band of the mixed noise data is set to F1 + F2 + F3 + F4. That is, in this case, there is no band limitation, that is, noise is superimposed on the entire band, so that the difficulty of hearing is maximized.

  With such allocation, as in the process of FIG. 10, when the accuracy rate exceeds the first threshold value TH1, the band limiting filter is used to widen the frequency band of the mixed noise data so that the difficulty level D increases. Can be controlled. Further, when the accuracy rate is lower than the second threshold value TH2, which is lower than the first threshold value TH1, the band limiting filter is controlled so as to narrow the frequency band of the mixed noise data so that the difficulty level D decreases. That's fine.

  The present invention is not limited to the above embodiments, and various changes and modifications can be made without departing from the spirit and scope of the present invention. Therefore, in order to make the scope of the present invention public, the following claims are attached.

Claims (6)

A rehabilitation support device for supporting rehabilitation of attention disorder,
Audio data storage means for storing a plurality of types of audio data;
Noise data storage means for storing a plurality of types of noise data;
Noise generating means for generating mixed noise data obtained by selecting and mixing one or more noise data from the noise data storage means;
Degraded speech creation means for creating degraded speech data by superimposing the mixed noise data created by the noise creation means on speech data selected from the speech data storage means;
Acoustic output means for acoustically outputting the created degraded voice data;
Have
The rehabilitation support apparatus, wherein the noise creating means is capable of adjusting at least one of the number of noise data to be selected and mixed from the noise data storage means and the gain of the mixed noise data. Control means for repeatedly executing the operations of the degraded sound creation means and the sound output means for a plurality of sound data selected from the sound data storage means;
Calculating means for calculating a correct answer rate regarding intelligibility of the deteriorated voice data by a listener of sound output repeatedly executed by the control means;
The noise generating means further comprises
If the calculated accuracy rate exceeds a first threshold, increase at least one of the number of the noise data and the gain so that the difficulty level regarding the intelligibility increases,
When the calculated accuracy rate is lower than a second threshold value lower than the first threshold value, at least one of the number of the noise data and the gain is set so that the difficulty level is lowered. The rehabilitation support device according to claim 1, wherein the rehabilitation support device is lowered. An acquisition means for acquiring the severity of the patient;
A setting means for setting an initial value of the difficulty level based on the acquired severity;
The rehabilitation support apparatus according to claim 2, further comprising: The noise generating means is
Band limiting filter means capable of changing a frequency band of the mixed noise data,
When the calculated accuracy rate exceeds a first threshold, the band limiting filter is controlled to widen the frequency band of the mixed noise data so that the difficulty level regarding the intelligibility increases.
When the calculated accuracy rate is lower than a second threshold value that is lower than the first threshold value, the band limiting filter so as to narrow the frequency band of the mixed noise data so as to reduce the difficulty level. The rehabilitation support device according to claim 1, wherein the rehabilitation support device is controlled. A control method of a rehabilitation support device for supporting rehabilitation of attention disorder,
A noise creation step of creating mixed noise data obtained by selecting and mixing one or more noise data from a noise data storage means for storing a plurality of types of noise data;
A deteriorated voice creation step of creating degraded voice data by superimposing the mixed noise data created in the noise creation step on voice data selected from voice data storage means for storing a plurality of types of voice data;
An acoustic output step for acoustically outputting the created degraded voice data;
Have
The method of controlling a rehabilitation support apparatus, wherein the noise generation step is capable of adjusting at least one of the number of noise data to be selected and mixed from the noise data storage means and the gain of the mixed noise data.   The program for making a computer perform each step of the control method of the rehabilitation assistance apparatus of Claim 5.

Images