KR100911725B1 - A language purifier robot - Google Patents

Abstract

According to an aspect of the present invention, there is provided a speech cleaning robot comprising: a slang table device for comparing a speech recognized by a speech recognizer to a "slang table signal" and a "not a table slang signal" in comparison with a slang table; A barrier sound slang detector which receives a "not a table slang signal" and outputs a "block noise slang signal" if the ratio of the barrier sounds in the voice is greater than or equal to a predetermined value, and outputs a signal that is not a barrier sound slang word; A strong interaction signal is output as the number of inputs of the "spoken slang signal" and the "table slang signal" is accumulated by receiving the "spoken slang signal", "not the slang signal", and "table slang signal", An interaction unit configured to output a weak interaction signal as the "no sound impairment signal" accumulates; And an output unit that informs the user visually, auditoryly, or tactilely of the sensitivity according to the intensity of the interaction signal. According to the present invention, children's use of slang hurts humans through interaction with the robot through the interaction with the robot to children who indiscriminately use the slang language at the time of language acquisition. You can reduce the use of slang by making you aware of what it is. It also helps other users with their language.

Language Cleaning Robot, Interaction, Disability Sounds, Slang Table, Bruise

Description

A language purifier robot}

1 to 3 are views for explaining the appearance of the language-cleaning robot according to the present invention;

4 is an overall block diagram of a language cleaning robot according to the present invention;

FIG. 5 is a block diagram illustrating the recognition unit 100 of FIG. 4.

FIG. 6 is a block diagram illustrating the interaction unit 200 of FIG. 4.

FIG. 7 is a block diagram illustrating the output unit 300 of FIG. 4.

<Description of reference numbers for the main parts of the drawings>

100: recognition unit 101: speech recognizer

102: slang table machine 103: slang slang detector

200: interaction unit 201: frequency, probability storage

202: 1 degree of freedom damped vibration model

203: means for storing the swear words

300: output unit 301: LED

302: vibration motor 303: speaker

FIELD OF THE INVENTION The present invention relates to language clean robots, in particular "bruise if correct." It is about a language-cleaning robot that is bruised emotionally by inappropriate language such as slang or swearing with an interaction concept.

Linguistic clean robots are used to purify the user's language, especially for children who are habitually acquiring and using inappropriate language. Can be.

The slang recognition technology must be included in the language clean robot, but the conventional slang recognition technology performs slang detection through comparison with a slang table stored in advance through speech recognition. However, this conventional method has a problem in that all the slang words must be stored in the slang table in advance, and the slang words which are not stored in the slang table have a limitation of not being recognized.

People are physically bruised and physically bruised, but are often mentally bruised by words. There have been various studies related to the emotional expression of robots, but most of them are mainly related to facial expressions of robots through LEDs, facial expressions of robots through movement of eyes, eyebrows, and mouth shapes, and voices of robots.

As we can see from the Uncanny Valley, introduced in his thesis published in 1970 by famous Japanese robot engineer Masahiro Mori, the robot's emotional expression that mimics the facial expressions and voices of existing humans The expectations of the users are increased, and when the interactions provided by the robots do not reach the expectations, the satisfaction of the robots may be lowered.

In T. Kanda's research, a passive robot, in which a person talks to a robot first and interacts with it, begins, rather than an interaction of an active robot, where a robot first talks to a person and begins to interact. It is found that the interaction of is preferred.

Therefore, the technical problem of the present invention is to detect slang words that are not stored in the slang table so that the slang table can be updated. It is to provide a language-cleaning robot that can appeal to users more freshly and emotionally through passive emotional expression.

According to an aspect of the present invention, there is provided a language cleaning robot, including: a voice recognizer for recognizing a user voice; A slang table that compares the speech recognized by the speech recognizer with a slang table and outputs a " table slang word " and a " not a table slang word " A barrier sound slang detector which receives the "not a table slang signal" and outputs a "block noise slang signal" if the ratio of the barrier sounds in the voice is equal to or greater than a predetermined value and outputs a signal that is not a barrier sound slang word; A strong interaction signal is output as the number of inputs of the "sound impaired slang signal" and the "table slang signal" is accumulated by receiving the "fault sound slang signal", "not a sound impaired signal", and a "table slang signal". And an interaction unit configured to output a weak interaction signal as the "no sound impairment signal" is accumulated; And an output unit configured to visually, auditoryly, or tactilely notify the user of the sensitivity according to the intensity of the interaction signal.

The output unit preferably outputs a negative emotional signal as the intensity of the interaction signal increases.

In the present invention, the interaction unit uses the frequency and probability values of the "disabled slang words", "not a slang words" signal, and a "table slang signal" as parameters, at a time modeled by a single degree of freedom attenuation vibration model. It is characterized in that for outputting the interaction signal according to.

The output unit may include an LED. In this case, the LED may be installed to express a variety of bruises, colors, or eyes through the color and brightness of the light according to the intensity of the interaction signal.

The output unit may include a vibration motor and a speaker. In this case, the vibration motor and the speaker output various vibrations and sounds according to the intensity of the interaction signal to express the emotional state of the robot.

Preferably, the interaction unit stores a voice corresponding to the impaired slang signal and uploads it to a slang table of the slang table.

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the present invention will be described in detail. Only features of the invention are shown in order to avoid repeated description with the prior art. The following examples are only presented to understand the content of the present invention, and those skilled in the art will be capable of many modifications within the technical spirit of the present invention. Therefore, the scope of the present invention should not be construed as being limited to these examples.

1 is a view for explaining the appearance of the language-cleaning robot according to the present invention. As shown in Figure 1, from the left can be a variety of forms, such as steamed bun robot, porcelain robot, egg robot. 'You bruise, you get hurt.' In order to highlight the characteristics of the passive robot, a simple appearance consisting of body and eyes is desirable, and the material is translucent and metal.

The language-cleaning robot according to the present invention is composed of a body part expressing a bruised or bloody interaction and an eye part expressing language perception. The body part includes a slang recognition device that distinguishes between slang and general language, and expresses a bruise (FIG. 2) and a color (FIG. 3) through color LEDs and a vibration module. The eye part tells the user that he is aware of the slang language by blinking an eye through the color LEDs whenever the language is recognized.

4 is an overall block diagram of a language cleaning robot according to the present invention. As shown in FIG. 4, the recognition unit 100 includes a recognition unit 100, an interaction unit 200, and an output unit 300.

FIG. 5 is a diagram for describing the recognizer 100 of FIG. 4. The recognition unit 100 includes a voice recognizer 101, a slang table 102, and a disability sound slang detector 103. The voice of the user is recognized through the voice recognizer 101. The slang table 102 compares the voice recognized by the speech recognizer 101 with the slang table and outputs a "table slang signal" to the interaction unit 200 if it corresponds to the slang word in the slang table. If it does not correspond to a slang word, a "not a table slang word" signal is output to the scam slang detector 103.

The disability sound slang detector 103 that receives the "not a table slang signal" detects the obstructions included in the user's voice by using a spectral center to detect the ratio of the disability sound included in the voice. Detect slang through That is, if the ratio of the barrier sounds is greater than or equal to the predetermined value, the "speech noise slang signal" is output to the interaction unit 200, and if the ratio of the barrier sounds is less than that, the signal "not a barrier sound slang word" is output to the interaction unit 200.

FIG. 6 is a block diagram illustrating the interaction unit 200 of FIG. 4. As illustrated in FIG. 6, in the interaction unit 200, the "table slang signal" output from the slang table 102 and the "speech slang signal" and the "not slang" slang output from the sneak slang detector 103 are not. Signal, and store their frequency and probability value 201, and output the interaction signal modeled by the 1 degree of freedom attenuation vibration model 202 to the output unit 300.

The 1 degree of freedom damping vibration model is an over-damping case model of mass, spring, and damper system, and can appropriately model attenuation over time of an output signal. The parameters of the 1 DOF vibration model are determined according to the frequency and the probability value.

Therefore, the interaction unit 200 outputs a strong interaction signal as the number of recognition of the slang words is accumulated, and outputs a weak interaction signal as the number of recognition of the normal language is accumulated. Thus, the time-dependent strength and weakness of the interaction signal output by the inputs of the "spoken slang and the table slang" and the "normal language" are the "speech slang", "not the slang" and "table slang" signals It will change according to the frequency and probability value.

The interaction unit 200 stores the voice, which is determined by the obstacle sound slang detector 103, in the sound impaired slang storage means 203, and periodically uploads the sound to the slang table of the slang table machine 102. To update the slang table machine 102.

FIG. 7 is a block diagram illustrating the output unit 300 of FIG. 4. The output unit 300 displays the emotional state according to the intensity of the interaction signal output from the interaction unit 200 through the LED 301, the vibration motor 302, and the speaker 303 through visual, tactile, or hearing to the user. Emotionally informs. As the intensity of the interaction signal increases, the output unit 300 outputs a negative emotional signal such as a large portion of the bruise or a darker bruise.

The LED 301 expresses various kinds of bruises, colors, eyes, etc. by using three colors, and is installed to express various feelings of bruises, colors, or eyes through kinds and brightness of light according to the intensity of the interaction signal. . The vibration motor 302 and the speaker 303 output various vibrations and sounds according to the intensity of the interaction signal to express the emotional state of the robot to provide positive or negative feedback to the user.

Children at the time of language acquisition do not know whether slang is bad or hurt people, and use habitually slang words according to the present invention. Bruises or positive reactions can be used to reduce the use of slang words by making children aware that slang words are hurting people. It also helps other users with their language.

Claims (6)

A voice recognizer for recognizing a user's voice; A slang table that compares the speech recognized by the speech recognizer with a slang table and outputs a " table slang word " and a " not a table slang word " A barrier sound slang detector that receives the "not a table slang signal" and outputs a "block noise slang signal" if the ratio of the barrier sounds in the voice is greater than or equal to a predetermined value and outputs a signal that is not a barrier sound slang word; The frequency of the "disabled sound slang signal", "not a disabled sound slang signal", and a "table slang signal" by receiving the "disabled sound slang signal", "not a disabled sound slang signal", and a "table slang signal" Output the interaction signal according to the time modeled by the 1 degree of freedom attenuation vibration model using the probability value as a parameter, and the stronger the interaction signal is output as the number of times of recognition of the "failure slang word" and the "table slang word signal" is accumulated. And an interaction unit for outputting a weak interaction signal as the “no sound impairment signal” is accumulated. And And an output unit for visually, auditoryly, or tactilely notifying the user of the sensitivity according to the intensity of the interaction signal. The language clean robot of claim 1, wherein the output unit outputs a negative emotional signal as the intensity of the interaction signal increases. delete The method of claim 1, wherein the output unit comprises an LED, wherein the LED is installed to express a variety of bruises, colors, or eyes through the color and brightness of the light according to the intensity of the interaction signal. Language clean robot. The method of claim 1, wherein the output unit comprises a vibration motor and a speaker, the vibration motor and the speaker outputs various vibrations and sounds according to the intensity of the interaction signal to represent the emotional state of the robot Language clean robot. The language clean robot of claim 1, wherein the interaction unit stores a voice corresponding to the impaired slang signal and periodically uploads it to a slang table of the slang table.

Images