JP2006220880A - Vocalization support device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a vocalization support device which is simpler compared with an existing vocalization supporting device, which can be easily used even by an aged patient who is not physically strong enough to breathe out strongly, which enables a user to feel the actual sensation of vocalization, and which can support sufficiently clear vocalization. <P>SOLUTION: The vocalization support device comprises: a vowel sensor 4 which detects positional change based on vocalizing actions and outputs an electric signal; a consonant signal generating section 5 which selectably outputs, as an electric signal, consonant elements based on vocalizing actions; a vowel identification section 6 which identifies vowels and generates vowel codes; a consonant identification section 7 which identifies electric signals with respect to consonant elements and generates consonant codes; a vocalization database 9 which stores voice data and/or character data; a control section 8 which reads voice data and/or character data out of the vocalization database 9; and an output section 10 which outputs voice data and/or character data. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an utterance assisting device used by an aphasic patient, a patient with a vocal cord disorder, and the like to practice utterance, and more particularly to an utterance assisting device capable of realizing that he / she has made a voice.

Conventionally, many techniques have been developed to promote vocalization in patients with vocal cord damage due to diseases such as laryngeal cancer or psychogenic aphasia.
For example, in Patent Document 1, as an “artificial larynx for assisting speech”, a pitch pattern signal generation operation for controlling the vibration frequency of the vibrator of the artificial larynx is described as operation control means and mountain pattern generation unit or descending pattern generation. An invention is disclosed in which accents and intonations can be naturally expressed by being controlled by means or the like.
By the operation of the operation control means of the present invention, the chevron pattern generating means generates a chevron pitch pattern signal, and this chevron pitch pattern rises from a low pitch level and turns to decrease through a maximum point, A pattern descending to the reference pitch level is formed. The descending pattern generating means forms a descending pitch pattern that has a downward convex curve shape and monotonously decreases to the reference pitch level.
In the invention disclosed in Patent Document 1, since it can be configured to automatically generate a pitch pattern having a smooth curved shape, the voice assisted by the artificial larynx exhibits a natural pitch change. Along with this, there is an effect that non-harmful voice is realized.

Next, in Patent Document 2, as an “electrical artificial larynx”, the biological information detection means for detecting biological information and the fundamental frequency and volume of the substitute sound source output from the acoustic transducer are controlled according to the detected biological information. An apparatus with control means is disclosed.
In this invention, there is no specific description of what kind of sensor is used and how the obtained biological information is used and controlled. However, according to this patent document, the basics of the substitute sound source using the biological information are described. Because the frequency and volume can be controlled, it is possible to live a comfortable life with a natural voice.

Patent Document 3 discloses an invention of “generation assisting device and program recording medium”. This invention is a low-frequency component that removes a low-frequency component from an input audio signal in order to solve a lack of sound volume and an insufficient symbiotic component of vocal cords with abnormalities in vocal cords such as esophageal voice and recurrent paralysis voice A high-frequency component removal filter that removes the high-frequency component, an analysis processing unit that analyzes and synthesizes the output of the high-frequency component removal filter, and adds the output of the analysis processing unit and the output of the low-frequency component removal filter An adder is provided.
According to such an invention, it is possible to improve the voice quality by increasing the sound component while improving the loudness of the patient's voice by removing the low frequency and high frequency noise unnecessary for communication.

Japanese Patent Laid-Open No. 11-69476 Japanese Patent Laid-Open No. 7-433 JP 2000-242287 A

However, in the invention described in Patent Document 1, by operating the switches SW1 to SW3, a pitch pattern such as a mountain shape or a descent is generated to give accents and intonation close to Japanese spoken words. The utterance itself is performed by the artificial larynx, and the generation of the previous pitch pattern is performed by the patient (speaker) by operation.
FIG. 3 discloses a flowchart of the operation of the switches SW1 to SW3, and pitch patterns uttered by turning these switches on and off are disclosed in FIGS. 4 to 11, but these switches are turned on and off by the patient. It has nothing to do with the utterance action. In such an invention, it is possible to change accents and intonations simply by operating the switch at hand without the patient actually performing the utterance operation, and the patient practices for more accurate utterance. In some cases, it becomes unsuitable.
In addition, since the utterance is performed with the artificial larynx, it may be possible to produce a sound by the action of some utterances, but the artificial larynx must be used against the throat, and it is determined to be suitable for utterance. It is necessary to install in the position, and it is necessary to exhale for vocalization. However, when the vocal cords are extracted at the time of surgery, the breath often leaks and physical strength is required. In addition, elderly patients were particularly unable to utter using an artificial larynx because of this physical stress.

  The invention disclosed in Patent Document 2 also uses an artificial larynx, detects biological information including respiration and myoelectric potential with a sensor, and uses these to generate a frequency of a substitute sound source output from the acoustic transducer. And so on. In this invention, it is unclear what biological signal is used and how it is acoustically converted. However, as in the case of Patent Document 1, an artificial larynx is used and breathing is required. In addition, since it is necessary to attach many sensors for detecting biological information to the body, it is a burden on the patient.

  In the invention disclosed in Patent Document 3, the utterance signal is electrically processed, and only effective components are extracted and amplified, thereby enabling a more natural utterance. There is a problem that the voice is not effective when there is a certain amount of voicing, and it is difficult to utter in an elderly patient as described above. That is, in the invention disclosed in Patent Document 3, there is a problem that the effect varies depending on the utterance ability of the patient.

  The present invention has been made in response to the conventional situation, and is simpler than a conventional voice assist device, but is easy for elderly patients who are insufficient in physical strength and cannot exhale strongly. It is an object of the present invention to provide an utterance assisting device that can be used in a variety of situations and that can provide a sense of utterance and can assist the utterance clearly.

  In order to achieve the above object, an utterance assisting device according to the first aspect of the present invention includes a vowel sensor that is mounted on the face and detects a positional change based on a utterance operation and outputs an electrical signal, and a consonant component of utterance. A consonant signal generating unit that outputs an electrical signal in a selectable manner, a vowel identifying unit that discriminates a vowel based on an electrical signal accompanying a position change output by a vowel sensor and generates a vowel code corresponding to the vowel, and a consonant signal A consonant identification unit that discriminates an electrical signal related to a consonant component output by the generation unit and generates a consonant code corresponding to the consonant, and stores voice data and / or character data in a readable manner using the vowel code and the consonant code as keys. Control of reading voice data and / or character data from the utterance database using the utterance database and the generated vowel codes and consonant codes When, in which an output unit for outputting audio data and / or character data read.

  According to a second aspect of the present invention, in the utterance assisting device according to the first aspect, the control unit has a function of generating a utterance code by synthesizing a vowel code and a consonant code, and a utterance database. Instead of storing voice data and / or character data in a readable manner using vowel codes and consonant codes as keys, voice data and / or character data is stored in a readable manner using generated codes as keys.

  According to a third aspect of the present invention, in the utterance assist device according to the first or second aspect, the vowel sensor detects a change in position and outputs an electrical signal, It detects the expansion and contraction and outputs an electrical signal.

  According to a fourth aspect of the present invention, in the utterance assisting device according to the first or second aspect, the vowel sensor detects the change in position and outputs an electric signal, instead of detecting the position change. It detects the distance between the left and right and the displacement of the temporomandibular joint and outputs an electrical signal.

  According to a fifth aspect of the present invention, in the utterance assistance device according to any one of the first to fourth aspects, the consonant signal generation unit includes a plurality of instruction units. By associating consonant components one by one in advance, an electric signal related to the corresponding consonant component is output.

  The utterance assist device according to claim 6 is the utterance assist device according to any one of claims 1 to 4, wherein the consonant signal generator is attachable to each finger of the hand and is an ON-OFF signal. A plurality of switches capable of outputting a signal, and by outputting an electrical signal related to the corresponding consonant component in advance by associating the consonant components one by one with the combination of ON-OFF signals output from these switches. Is

  As described above, the utterance assisting device according to the present invention can detect a vowel that a patient is trying to utter by detecting a change in position, displacement, and expansion / contraction based on a generated operation by a vowel sensor mounted on the face. Since it is possible, it is not necessary to exhale, and even an elderly patient with weak physical strength can easily assist in speaking.

  In addition, only vowels are detected by a sensor and discriminated to generate vowel codes. Consonants can be selected by a separate means instead of a sensor, and consonant codes can be generated using these as keys or these codes. Since the synthesized utterance code is generated and the voice data or character data stored in the utterance database is read out using this as a key, it is possible to experience the utterance by performing the utterance operation. Since at least the consonant is selected by the operation, the utterance amount can be easily and highly accurately uttered.

DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments of a speech assisting apparatus according to the present invention will be described with reference to FIGS. (In particular, corresponding to claims 1, 2 and 6.)
FIG. 1 is a configuration diagram of the speech assisting device according to the present embodiment. In FIG. 1, the speech assisting device is roughly divided into a main body 3 and an output unit 10. The main body 3 includes a vowel sensor 4 mounted around the patient's mouth 1 and a consonant signal generator 5 provided in the hand 2.
Since the vowel sensor 4 attached around the mouth 1 moves simultaneously with the movement of the mouth 1 when the patient takes a utterance action, the change in the position detected by the movement is converted into an electrical signal and output. To do. A sensor that detects a change in position is, for example, converting the position change obtained by integrating the acceleration generated when the vowel sensor 4 moves to calculate the velocity and further calculating the distance. What to do is conceivable. Therefore, in this case, the position change can also be referred to as an acceleration change. In the present embodiment, a sensor that detects a change in acceleration is used as the vowel sensor 4 that detects a change in position.
The electrical signal output by the vowel sensor 4 is transmitted to the vowel identification unit 6 of the main body 3, and the vowel identification unit 6 determines the electrical signal of the output vowel and generates a vowel code corresponding to the vowel. The vowel code is a code for identifying “A”, “I”, “U”, “E”, “O”. A method for discriminating the electric signal of the vowel will be described in detail later.

On the other hand, the consonant signal generator 5 is configured so that consonants can be selected by the operation of the patient's hand. Although the configuration of the consonant signal generating unit 5 will be described later, an electric signal corresponding to the selected consonant is output, and the output electric signal is transmitted to the consonant identifying unit 7 of the main body 3. An electrical signal of sound is discriminated and a corresponding consonant code is generated. The consonant codes are "a", "ka", "sa", "ta", "na", "ha", "ma", "ya", "ra", "wa", "ga", "gya" ”,“ Za ”...,“ N ”. In the claims and specification of the present application, for the sake of convenience, “consonant”, “consonant component”, and “consonant chord” include a “row” sound that is a “row” consisting only of vowels. This is because, in order to specify the vowel sound by combining the vowels with the consonant, even if only the vowel such as “A” is specified, the consonant cannot be specified as “A” unless it is specified as “A”. However, it is needless to say that “no row” may be recognized when there is no electrical output from the consonant signal generator 5.
The vowel codes and consonant codes generated by the vowel identification unit 6 and the consonant identification unit 7 are transmitted to the control unit 8 included in the main body 3, and the control unit 8 synthesizes these codes and newly generates a utterance code. Generate. This utterance code is a code specified by any one of so-called 50 sounds including muddy sounds and repellent sounds.

  Further, the control unit 8 accesses the utterance database 9 of the main body 3, reads out the voice data and / or character data stored in the utterance database 9 using this utterance code as a key, and transmits it to the output unit 10. To do. Although the utterance database 9 is provided inside the main body 3 in the present embodiment, it may of course be externally attached.

  The output unit 10 is, for example, a hearing device such as a speaker for emitting voice data, or a visual device such as a liquid crystal display or a plasma display for displaying character data. Furthermore, any type of data may be an interface that can be output to an external device, and is not particularly limited.

FIG. 2 is a conceptual diagram schematically showing the data structure stored in the utterance database 9. In the utterance database 9a shown in FIG. 2A, as described above, a binary code 11 which is a utterance code generated by synthesizing a vowel code and a consonant code in the control unit 8, and the binary code 11 Corresponding to the above, voice data 12 and character data 13 are stored in a table.
The binary code 11 is a combination of a code representing a vowel and a code representing a consonant. The vowel is represented by a combination of three digits of 0 or 1, and the consonant is also represented by, for example, 10 digits of 0 or 1. Then, all utterances can be expressed by a combination of 0 or 1 of 13 digits in total. What is described in the column of the binary code 11 in the figure indicates “A” and “I” although numbers in the binary system are omitted on the way. Therefore, in the above example, the first 10 digits are a common binary code to indicate “A”, but the subsequent 3 digits indicate “A” and “I” of the vowel, respectively. The binary code 11 is different, and the overall binary code 11 is different.

  In the utterance database 9a of this figure, both the voice data 12 and the character data 13 are stored in association with the binary code 11, but depending on the application, only one of them may be stored. Further, the control unit 8 searches the table using the binary code 11 as a key to read out the voice data 12 or the character data 13, but either one or both may be used. In the figure, the same “A” and “I” are displayed. Of course, “A” in the voice data 12 column is “A” as voice data, and “A” in the character data 13 column. "" Is "a" as character data.

Next, the data structure shown in FIG. In the control unit 8 described above, a new utterance code is generated by synthesizing the vowel code and the consonant code. However, as shown in FIG. The database 9b separates the voice data 14 and the character data 15 to form a table. In each data, the vowel binary codes 17 and 19 and the consonant binary codes 16 and 18 are arranged in a matrix, and these are keyed. Depending on the combination, 50 sounds including muddy sound and repellent sound may be specified.
When data is configured in this way, it is not necessary to synthesize vowel codes and consonant codes by the control unit 8, and the control unit 8 itself can be simplified, but the utterance database 9b has a somewhat complicated matrix. Will be.
“000”, “011”, “111”, “010”, and “110” described in the column of the vowel binary codes 17 and 19 are vowel codes. Since there are five vowels in Japanese, a binary code requires at least three digits as a code structure. In this embodiment, the description is based on Japanese, but it is possible to deal with other languages that have vowels and consonants as well. When the number of vowels and consonants is large, it can be dealt with by appropriately increasing the number of digits of the vowel binary code or the consonant binary code.

Next, the mounting state of the vowel sensor will be described with reference to FIGS.
FIGS. 3A and 3B are conceptual diagrams schematically showing a vowel sensor attached to a patient's face. In FIG. 3A, the horizontal acceleration sensor 20 is attached to the side of the patient's mouth, and the vertical acceleration sensor 21 is attached to the lower side of the mouth. The calibration acceleration sensor 22 is attached to the patient's occipital region as shown in FIG. Although it is a sensor for measuring horizontal acceleration and vertical acceleration based on utterance movements, it may detect acceleration even when the entire face moves in addition to changes in the mouth rotation based on utterance movements, A calibration acceleration sensor 22 is provided on the back of the head that does not affect the movement of the mouth when speaking, and arithmetic processing is performed using the acceleration detected by this sensor and the acceleration detected by the horizontal acceleration sensor 20 and the vertical acceleration sensor 21. , It is possible to measure the acceleration generated around the mouth accompanying the utterance action. As a specific example of the arithmetic processing, based on the acceleration detected by the calibration acceleration sensor 22, it is subtracted from the acceleration detected by the horizontal acceleration sensor 20 and the vertical acceleration sensor 21, so that the acceleration generated around the mouth accompanying the utterance operation May be measured. However, in such a case, when the neck is moved, the movement around the mouth may be larger than the movement of the back of the head, and calibration is performed from the acceleration detected by the horizontal acceleration sensor 20 or the vertical acceleration sensor 21. It may be considered that accurate acceleration may not be obtained simply by subtracting the acceleration detected by the acceleration sensor 22. In such a case, it is conceivable to improve the accuracy by performing correction or increasing the number of sensors.

  FIG. 5 is a conceptual diagram schematically showing the shape of the mouth when vowels are uttered. 5A to 5E correspond to “A” to “O”, respectively. The vowels are identified by measuring the horizontal and vertical accelerations generated when the lips move to these mouth shapes.

6A and 6B are directions determined for convenience with respect to the direction in which acceleration occurs. In this specification, the x-axis direction means a horizontal direction parallel to the face, the y-axis direction means a vertical direction parallel to the face, and the z-axis direction is perpendicular to the face. Means the right direction. In the present embodiment, the horizontal acceleration sensor 20 shown in FIG. 4 measures the acceleration in the x-axis direction, and the vertical acceleration sensor 21 measures the acceleration in the y-axis direction and the z-axis direction. Each direction shown in the figure is the positive direction.
Table 1 is a vowel discrimination table showing the vowel utterance operation and the positive / negative state of each axial component of the acceleration detected by the horizontal acceleration sensor 20 and the vertical acceleration sensor 21.
In Table 1, for example, if “A”, the horizontal acceleration sensor 20 indicates a positive (+) value in the x-axis direction, and the vertical acceleration sensor 21 indicates a positive (+) value in the y-axis direction and the z-axis direction. Indicates.

Compared with the shape of the mouth shown in FIG. 5A, since the mouth is slightly opened left and right and up and down to utter the vowel “A”, positive acceleration occurs in the x-axis and y-axis. The positive acceleration is generated on the z-axis because the teeth push the lower lip forward slightly when speaking.
If “I” is uttered, the horizontal acceleration sensor 20 can only detect positive acceleration in the x-axis direction, and “U” can utter without moving the mouth. Not done. Similarly, for “e” and “o”, + can be written in the column of each axis. In addition, since it is thought that the detection of these accelerations also has an individual difference, it is desirable to be able to set a vowel, and to make this Table 1 amendable suitably according to an individual difference.
A vowel binary code can be generated by setting the + part shown in Table 1 to 1 and setting the part where acceleration was not detected, that is, the part left blank in Table 1 to 0.

The steps up to the generation of the vowel code will be described while summarizing the above with reference to FIG. FIG. 7 is a process diagram showing a signal processing process in the vowel identification unit when the acceleration sensor is a vowel sensor.
In FIG. 7, the electrical signal of the acceleration generated by the vowel sensor 4 comprising the horizontal acceleration sensor 20, the vertical acceleration sensor 21 and the calibration acceleration sensor 22 mounted around the patient's mouth shown in FIG. (Step S1).
As described above, the relative acceleration change of the horizontal acceleration sensor 20 and the vertical acceleration sensor 21 is calculated using the calibration acceleration sensor 22 in order to eliminate the influence of the movement of the entire face (step S2).
Based on the relative acceleration change of the horizontal acceleration sensor 20 and the vertical acceleration sensor 21, vowels are discriminated by the vowel discrimination table shown in Table 1 stored in the vowel discriminator 6 (step S3), and as described above. A binary code is generated (step S4).

Next, in the case where a vowel sensor 4 that detects the expansion and contraction of a muscle and outputs an electrical signal is employed, a process until the vowel identification unit 6 determines a vowel and generates a vowel code will be described. (In particular, corresponding to claim 3)
The place where the vowel sensor 4 is mounted is as shown in FIGS. 4A and 4B, similarly to the acceleration sensor. The vowel sensor 4 that detects the expansion and contraction of the muscle discriminates the vowel depending on how the muscle around the mouth expands and contracts during the vowel utterance operation. The horizontal extension / contraction sensor 24 is attached to the lower side of the patient's mouth, and the vertical extension / contraction sensor 23 is attached to the side of the mouth. The occlusal expansion / contraction sensor 25 is attached to a so-called “gill” portion of the jaw of the patient's face.
The axial directions of x, y, and z shown in FIG. 6 are common to the sensors mounted in this way. In the vowel sensor 4 that senses muscle expansion and contraction, vowels are discriminated by a vowel discrimination table as shown in Table 2.

Table 2 is a vowel discrimination table showing the vowel voicing operation and the expansion / contraction state of each axial component of the muscle detected by the horizontal expansion / contraction sensor 24 and the vertical expansion / contraction sensor 23.
In Table 2, for example, “A” indicates that the muscle around the mouth extends in the x-axis direction and the y-axis direction, so that the horizontal expansion / contraction sensor 24 and the vertical expansion / contraction sensor 23 indicate expansion, and the occlusal expansion / contraction sensor 25 indicates contraction. If such expansion and contraction are 1 and 0, respectively, a vowel binary code can be generated as in the case of the acceleration sensor.

With reference to FIG. 8, the steps until generation of a vowel code will be described as in the case of the acceleration sensor. FIG. 8 is a process diagram showing a signal processing process in the vowel identification unit when the expansion / contraction sensor is a vowel sensor.
In FIG. 8, the muscular expansion / contraction electrical signal generated by the vowel sensor 4 composed of the horizontal expansion / contraction sensor 24, the vertical expansion / contraction sensor 23, and the occlusal expansion / contraction sensor 25 shown in FIG. ).
The vowel identification unit 6 calculates the expansion / contraction change in each sensor (step S2), and determines the vowel by the vowel determination table shown in Table 2 stored in the vowel identification unit 6 by the expansion / contraction change by the three sensors (step S2). S3), thereby generating a vowel binary code as described above (step S4).

Furthermore, when a vowel sensor 4 that detects the distance between the upper and lower lips and the left and right of the lips and the displacement of the temporomandibular joint and outputs an electrical signal is used, the vowel identification unit 6 discriminates the vowel and generates a vowel code. The process up to this will be described. (In particular, corresponding to claim 4)
The vowel sensor 4 is mounted differently from the acceleration sensor or the muscle expansion / contraction sensor, as shown in FIG. 9, horizontal distance sensors 27a and 27b are mounted on the left and right sides of the lips around the mouth, and the vertical distance sensors 26a and 26b are Mounted in two places on the top and bottom. Further, as shown in FIG. 10, the occlusal position displacement sensor 31 is mounted on the “gill” portion of the jaw of the patient's face.
The vertical distance sensors 26a and 26b and the horizontal distance sensors 27a and 27b detect the vowels by detecting the distance between the sensors, respectively.

FIG. 11 is a conceptual diagram simply showing the structure of the vertical distance sensor 26 and the horizontal distance sensor 27. The sensor shown in FIG. 11 (a) is, for example, a vertical distance sensor 26a, which is formed in a circular patch shape, is provided with a magnet 28 at the center on the back side, and is coated with an adhesive 29 so that it can be worn around the mouth. Can be affixed to.
On the other hand, the sensor shown in FIG. 11 (b) is, for example, a vertical distance sensor 26b, and a reed switch 30 is provided on the back side thereof. The reed switch 30 reacts when the magnet approaches within a certain distance. Therefore, the switch is turned on and off when the upper and lower and left and right ends of the lips approach or separate from each other due to the movement of the lips. Since the adhesive material 29 is similarly applied to the back surface, it can be easily attached. Instead of the reed switch 30, a magnetic sensor capable of sensing the magnetism of the magnet 28 may be used to detect the distance between the upper, lower, left and right ends of the lips, and the approach / separation.
The vertical distance sensor 26 sticks the sensors shown in FIGS. 11A and 11B on the upper and lower sides of the lips, and the horizontal distance sensor 27 sticks on the left and right.
This type of vowel sensor 4 discriminates vowels according to how the shape of the mouth, particularly the distance between the upper, lower, left and right of the lips, changes during the vowel utterance operation.

  Table 3 shows the vowel utterance operation, the open / closed state of the lips detected by the horizontal distance sensors 27a and 27b and the vertical distance sensors 26a and 26b, and whether or not the teeth detected by the occlusal position displacement sensor 31 are bitten. This is a vowel discrimination table for discriminating vowels.

  In Table 3, for example, if it is “A”, the degree of opening of the lips is large, the degree of opening of the lips is also large, and whether the teeth are bitten (the jaw is closed) It shows the situation of not chewing. For example, a vowel binary code can be generated by setting 0 for a large opening, 1 for a small opening, 1 for a case where the teeth are biting, and 0 for a case where the teeth are not biting. .

  Table 4 exemplifies the vowel binary code generated in this way. In “A”, the horizontal and vertical lip opening degree is large, so that it is 0, 0, and further, it is 0 because the teeth are not chewed. Therefore, all three-digit binary numbers are 0.

  According to the vowel discrimination table of Table 3, the vowels of Table 4 are binary codes as shown.

With reference to FIG. 12, steps up to the generation of a vowel code will be described as in the case of an acceleration sensor or a muscle expansion / contraction sensor. FIG. 12 is a process diagram showing a signal processing process in the vowel identification unit when the distance sensor is a vowel sensor.
In FIG. 12, the lip distance electrical signal and the positional displacement electrical signal generated by the vowel sensor 4 composed of the vertical distance sensor 26, the horizontal distance sensor 27, and the occlusal position displacement sensor 31 shown in FIGS. It is read by the identification unit 6 (step S1).
The vowel identification unit 6 calculates a change in distance for the vertical distance sensor 26 and the horizontal distance sensor 27 (step S2), and calculates a displacement of the position of the occlusal position displacement sensor 31 based on a change in acceleration or a change in muscle stretch ( In step S2), the vowels are discriminated by the vowel discrimination table shown in Table 3 stored in the vowel identification unit 6 (step S3), thereby generating a vowel binary code (step S4).

Heretofore, the description has been made up to the generation of the vowel code in the utterance assistance device according to the present embodiment. Next, a consonant discrimination method using the consonant signal generation unit 5 and the consonant identification unit 7 shown in FIG. 1 will be described.
Unlike a vowel that can be identified mainly by the movement of the lips that accompanies the vocalization, the consonant is determined by how to exhale and how the tongue moves, so it is difficult to recognize it from the mouth and lips. In addition, elderly patients and the like often have weak physical strength and often have a weak way of exhaling, so it is particularly difficult to determine consonants.
Therefore, the consonant signal generation unit 5 generates an electric signal related to the consonant component regardless of the generation operation.

  For example, in this embodiment, an electric signal is generated by using an apparatus as shown in FIG. FIG. 13 is an outline view showing the structure of the consonant signal generator according to the present embodiment. In FIG. 13, the consonant signal generator includes a rubber sack 32 and a switch 35 above the finger base of the rubber sack 32, and a piano wire 33 extending from the switch 35 is passed through a guide 34 while the tip of the finger. It fixes to the fastener 41 provided in the side.

FIG. 14 is a conceptual diagram schematically showing the structure of the switch 35. In FIG. 14, an indenter 36 is built in the switch 35, and the piano wire 33 extends from the indenter 36. The indenter 36 receives a repulsive force by a spring 38 provided in the switch 35 in the same manner according to the tension state of the piano wire 33, but if it is strongly pulled by the piano wire 33, it is separated from the switch element 37 and the tension of the piano wire 33 is weak. If it is, it will be urged | biased by the repulsive force of the spring 38, will touch the switch element 37, and will be in an ON state.
By generating an electrical signal when the indenter 36 contacts the switch element 37 by such a switch 35, the switch 35 is turned on by fitting the rubber sack 32 to the finger and bending or stretching the finger. -OFF can generate an electrical signal.

There are nine consonants, K, S, T, N, H, M, Y, R, and W. As explained earlier, there are 10 consonants including the “A” line. Each consonant component can be generated by attaching ten consonant signal generators 5 corresponding to each finger of both hands in order to correspond to ten fingers of both hands. Turbidity and sound repellent can be dealt with by turning on a plurality of switches among these ten.
Tables 5 and 6 show specific examples of this.

  Table 5 shows the correspondence between the fingers of both hands and the consonant signal generator shown in FIG. 14, and Table 6 shows that the finger is bent, that is, the switch 35 is OFF, including muddy sound and sound repellent. FIG. 5 is a table summarizing a binary code, where 1 is a state in which a finger is extended, that is, a state in which the switch 35 is ON.

With reference to FIG. 15, steps until generation of a consonant code will be described in the same manner as in the case of the vowel sensor 4 and the vowel identification unit 6. FIG. 15 is a process diagram showing signal processing steps in the consonant identification unit 7 when the consonant signal generation unit 5 shown in FIGS. 13 and 14 is attached to each finger of both hands.
In FIG. 15, the electrical signal generated by the switch element 37 of the consonant signal generator 5 shown in FIGS. 13 and 14 is read by the consonant identifier 7 (step S1). In this step S1, when there is a signal for all fingers in order to confirm the presence / absence of an electrical signal for each finger, it is read. The “right parent” in step S1 in FIG. 1 means the switch 35 that is the consonant signal generation unit 5 attached to the thumb of the right hand, and the “left child” is attached to the little finger of the left hand. The switch 35 is meant. In the meantime, the switches 35 attached to the index finger, the middle finger, the ring finger, the little finger, the left thumb, the index finger, the middle finger, and the ring finger of the right hand are omitted.
The consonant discriminating unit 7 discriminates the consonant from the electrical signal output from each switch 35 by using the consonant discrimination table shown in Table 6 stored in the consonant discriminating unit 7 (step S2), thereby the binary code of the consonant. Is generated (step S3).

Next, a case where a device having a plurality of keyboard keys is used as the consonant signal generator 5 will be described. (Especially corresponding to claim 5)
The consonant signal generating device 39 shown in FIG. 16 includes keyboard keys 40a to 40c in which “rows” of consonants including “a row” are arranged including muddy sounds and repellent sounds. When these keyboard keys are pressed, electrical signals relating to the 50 consonant components including muddy and repellent sounds displayed on these keyboard keys can be output.
According to the third embodiment, a consonant component code is generated by combining signals from all the switches 35 as in the consonant signal generating unit 5 including the switches 35 attached to the fingers of both hands. Instead, a consonant component code can be generated by one pressed key, so that consonant discrimination is easy.
In addition, instead of the plurality of keyboard keys of the consonant signal generation unit 5, as an instruction unit that can generate an electrical signal related to the consonant component, a plurality of touch sensors that react without being pressed, such as a keyboard key, are used. If “row” is displayed on the surface including muddy sound and sound repellent sound, it is possible to generate an electrical signal related to the consonant component only by touching. Furthermore, for example, by using a gaze sensor that can identify the position of the gaze, the gaze sensor is attached to the face, and the patient follows the display section in which the “row” of each consonant is displayed. It is also possible to determine which consonant component the sensor has and transmit it as an electrical signal.

Similarly, steps up to the generation of the consonant code will be described with reference to FIG. FIG. 17 is a process diagram showing signal processing steps in the consonant identification unit when a consonant signal generation unit having keyboard keys is employed.
In FIG. 17, the electrical signal output by the keyboard key shown in FIG. 16 is read by the consonant identification unit 7 (step S1).
The consonant identification unit 7 discriminates the consonant by detecting which keyboard key is used to output the electric signal (step S2). Therefore, although not shown in the table or figure, the consonant identification unit 7 stores a consonant discrimination table in which each keyboard key and consonant components, that is, consonant components in each row including muddy and repellent sounds correspond to each other. . Of course, “A row” is also included as described above.
A consonant binary code is generated by such a consonant discrimination table (step S3).

The vowel code and the consonant code generated independently by the vowel identification unit 6 and the consonant identification unit 7 are read by the control unit 8. The control unit 8 accesses the utterance database 9 and uses the vowel binary codes 17 and 19 or the consonant binary code 16 as a search key to generate voice data 14 or character data as shown in FIG. The corresponding speech and characters are extracted from 15. The extracted voice data 14 and character data 15 are output as voice and characters via the output unit 10, or are output as an interface to other external devices.
The vowel code and the consonant code are read by the control unit 8, but the control unit 8 may read the consonant code first and then read the vowel code so as to search voice data and character data. When vowels and consonant chords are consecutive, for example, when the order is such that the vowel chords are sandwiched between consonant codes, it may be difficult to determine whether the vowels are related to the preceding or following consonant codes. It is. However, the consonant code may be read after the vowel code to search for character data or voice data.

An example of the external device is a telephone. It is also possible to enjoy a telephone conversation by connecting the output unit 10 to a microphone terminal of the telephone.
Further, the control unit 8 has a function of synthesizing the previous vowel code and the consonant code, thereby synthesizing it as an utterance code, and accessing the utterance database 9a as shown in FIG. The voice data 12 and the character data 13 may be searched using the binary code 11 of the utterance code as a key.
Note that the voice data 12 and 14 are obtained by collecting the patient's own voice in advance. For example, when practicing utterance after surgery to delete the vocal cords, There is also a high motivation for practice and assistance.

As described above, according to the utterance assisting device according to the present embodiment, vowels that can be discriminated by the movement of the mouth or lips or their changes by the utterance operation are detected on the face including the periphery of the mouth and the chin. It is possible to enhance the sensation of utterance by discriminating by wearing vowels and uttering vowels along the utterance action. For consonants that are clearly discriminated by exhaling, a consonant signal generator that uses a finger to generate consonant components is used so that even elderly patients with weak physical strength can use them. Thus, it is possible to suppress the exhaustion of physical strength with respect to the practice and assistance of utterance, and it is not necessary to be ready, and it is possible to perform utterance training with ease and length. In addition, the accuracy is improved because the consonant is selected by the operation of the finger.
In this embodiment, Japanese language is assumed, but it is possible to cope with all languages such as English, Korean, Chinese, French, German, and Spanish. However, the contents of the vowel discrimination table and the consonant discrimination table need to be modified as appropriate in correspondence with those languages.

  As described above, the utterance assisting device of the present invention can be used for rehabilitation for the utterance of a patient whose vocal cords have been extracted, for example, in a medical institution such as a hospital. It can also be used for home treatment such as surgery and psychogenic aphasia even in ordinary households.

It is a block diagram of the speech assistance apparatus which concerns on this Embodiment. Both (a) and (b) are conceptual diagrams schematically showing the data structure stored in the utterance database. It is the conceptual diagram which represents typically the vowel sensor with respect to a patient's side face, (a) is a conceptual diagram at the time of seeing from a front, (b) is a side view. It is the conceptual diagram which represents typically the vowel sensor with respect to a patient's side face, (a) is a conceptual diagram at the time of seeing from a front, (b) is a side view. (A) vowel (A), (b) vowel (I), (c) vowel (U), (d) vowel (E), (e) vowel (O) It is a conceptual diagram which shows typically the shape of the mouth at the time of doing. It is a conceptual diagram which shows the direction defined for convenience about the direction where an acceleration generate | occur | produces, (a) is a conceptual diagram which shows the direction defined as x and y-axis direction, (b) is the direction defined as z-axis direction. It is process drawing which shows the process of the signal processing in a vowel identification part at the time of using an acceleration sensor as a vowel sensor. It is process drawing which shows the process of the signal processing in a vowel identification part at the time of using an expansion-contraction sensor as a vowel sensor. It is a conceptual diagram which represents typically the mounting state of a vertical distance sensor and a horizontal distance sensor. It is a conceptual diagram which represents typically the mounting state of the occlusal expansion-contraction sensor. Both (a) and (b) are conceptual diagrams simply showing the structures of a vertical distance sensor and a horizontal distance sensor. It is process drawing which shows the process of the signal processing in a vowel identification part at the time of using a distance sensor as a vowel sensor. It is an external view which shows the structure of the consonant signal generation part which concerns on this Embodiment. It is a conceptual diagram which shows the structure of a switch typically. It is process drawing which shows the process of the signal processing in a consonant identification part at the time of mounting | wearing each finger | toe with the consonant signal generation part shown by FIG. It is a conceptual diagram of the consonant signal generator provided with the some keyboard key. It is process drawing which shows the process of the signal processing in the consonant identification part at the time of employ | adopting the consonant signal generator provided with a keyboard key.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Mouth 2 ... Hand 3 ... Main body 4 ... Vowel sensor 5 ... Consonant signal generation part 6 ... Vowel identification part 7 ... Consonant identification part 8 ... Control part 9, 9a, 9b ... Speech database 10 ... Output part 11 ... Binary code DESCRIPTION OF SYMBOLS 12 ... Voice data 13 ... Character data 14 ... Voice data 15 ... Character data 16 ... Consonant binary code 17 ... Vowel binary code 18 ... Consonant binary code 19 ... Vowel binary code 20 ... Horizontal acceleration sensor 21 ... Vertical acceleration sensor DESCRIPTION OF SYMBOLS 22 ... Calibration acceleration sensor 23 ... Vertical expansion / contraction sensor 24 ... Horizontal expansion / contraction sensor 25 ... Occlusion expansion / contraction sensor 26a, 26b ... Vertical distance sensor 27 ... Horizontal distance sensor 28 ... Magnet 29 ... Adhesive material 30 ... Reed switch 31 ... Occlusal position displacement sensor 32 ... Rubber sack 33 ... Piano wire 34 ... Guide 35 ... Switch 36 ... Indenter 37 ... Switch element 38 ... Spring 39 ... Consonant No. generator 40a~40c ... keyboard key 41 ... fastener

Claims (6)

  A vowel sensor that is mounted on the face and detects a position change based on the utterance action and outputs an electrical signal, a consonant signal generator that outputs a consonant component of the utterance as an electrical signal, and is output by the vowel sensor A vowel discriminating unit that discriminates a vowel based on an electric signal accompanying a change in position and generates a vowel code corresponding to the vowel, and an electric signal related to a consonant component output by the consonant signal generation unit is discriminated and corresponds to this consonant. A consonant identification unit that generates a consonant code; a utterance database that readablely stores voice data and / or character data using the vowel code and the consonant code as a key; and the utterance using the generated vowel code and the consonant code A control unit that reads the voice data and / or character data from a database; and the read voice data and / or Vocalization assisting device and having an output section for outputting the character data.   The control unit has a function of generating a utterance code by synthesizing the vowel code and a consonant code, and the utterance database stores voice data and / or character data in a readable manner using the vowel code and the consonant code as a key. Instead, the voice assisting device according to claim 1, wherein voice data and / or character data is stored in a readable manner using the generated code as a key.   The vowel sensor according to claim 1 or 2, wherein the vowel sensor detects an expansion and contraction of a muscle and outputs an electric signal instead of detecting an electrical change and outputting an electric signal. Auxiliary device.   The vowel sensor detects the distance between the upper and lower lips and between the left and right sides and the displacement of the temporomandibular joint and outputs an electrical signal instead of detecting an electrical change and outputting an electrical signal. The speech assisting device according to claim 1 or 2.   The consonant signal generation unit includes a plurality of instruction units, and outputs an electrical signal related to a corresponding consonant component by previously associating the consonant components one by one with the instruction units. The speech assisting device according to any one of claims 1 to 4. The consonant signal generation unit includes a plurality of switches that can be attached to each finger of the hand and can output an ON-OFF signal. One consonant component is previously added to a combination of ON-OFF signals output from these switches. The utterance assisting device according to any one of claims 1 to 4, wherein an electrical signal related to a corresponding consonant component is output by making each correspond.