KR102100703B1 - Voice recognition kiosk with multiple input means - Google Patents

Abstract

The present invention relates to a voice recognition kiosk with a plurality of first input means comprising: a kiosk main body with a display of a certain size; a plurality of first input means provided on the left side or the right side of the display and receiving a voice signal of a speaker; a noise pickup unit connected to the first input means to remove the ambient noise from the speaker; and an output unit outputting information requested by the speaker to the display based on the voice signal of the speaker from which the ambient noise is removed.

Description

Voice recognition kiosk with multiple input means

The present invention relates to a speech recognition kiosk equipped with a plurality of input means, and more particularly, to a kiosk that is provided with a plurality of input means according to the height of the kiosk to increase the speech recognition rate of the talker.

Kiosk is an unmanned information terminal installed in public places such as government agencies, local governments, banks, department stores, and exhibition halls. It provides dynamic traffic information, public transportation information, route information, fare card distribution, reservations, and various phone numbers and address information. It can provide administrative procedures, product information, and how to use facilities.

In addition, it is an unmanned comprehensive information guidance system that provides efficient information to users, such as voice service and video realization, by utilizing advanced multimedia devices such as touch screens, sound, graphics, and communication cards.

However, conventionally, the function of inputting input data to the kiosk through only a touch has been mainly performed, and even for a kiosk equipped with voice recognition, a microphone is provided in a singular number, so that speech recognition satisfactory to people of all ages depending on the location of the microphone There was a problem that was not achieved. In addition, there is a problem in that the speech recognition rate is poor because the voice of the talker and the ambient noise can be input to the microphone.

Therefore, there is a need to provide a microphone for speech recognition of all ages in the kiosk and remove ambient noise input to the microphone.

Korean Patent Publication No. 10-2002-0087297

The present invention is to solve the problems of the prior art as described above, it is an object to increase the recognition rate of recognizing the speech of the speaker at various heights by providing a plurality of input means.

A voice recognition kiosk equipped with a plurality of input means according to the present invention for achieving the above object includes a kiosk body having a display in a predetermined size; A plurality of first input means provided on the left side or the right side of the display to receive a voice signal of a talker; A noise pickup unit connected to the plurality of first input means to remove ambient noise of a speaker; And an output unit that outputs information requested by the speaker to the display based on the voice signal of the speaker from which ambient noise is removed.

Here, the plurality of first input means is a microphone, characterized in that the microphone is provided at regular intervals from the top to the bottom of the kiosk body.

In addition, one of the microphones is provided at the top, middle, and bottom of the kiosk body, the microphone is connected to one noise pickup unit, and the noise pickup unit removes ambient noise of the microphone. .

In addition, the microphone is characterized by using an omni-directional microphone array.

In addition, when the plurality of first input means are arranged in a row on the left or right side of the kiosk main body, a gap of 5 cm or more is provided within the vertical length of the kiosk main body.

In addition, it includes a speech recognition unit for recognizing the speaker's voice and determining information requested by the speaker, and the speech recognition unit recognizes the speaker's requested information by recognizing the speaker's voice using a Recurrent Neural Networks (RNN) model It is characterized by delivering.

In addition, a sensor unit is provided in the vicinity of the plurality of first input means, and the sensor unit senses the key height of the talker and requests information from the plurality of first input means to the first input means close to the key height of the talker. It is characterized by guiding the following.

In addition, the guide is characterized in that the LED provided near the first input means is lit.

In addition, the sensor unit is characterized in that at least one or more of an ultrasonic sensor, an infrared sensor and an optical sensor.

The speech recognition kiosk according to the present invention has the following effects.

First, speech recognition can be customized for people of all ages. The speech recognition kiosk according to the present invention is provided with a plurality of input means on the front of the kiosk body. The plurality of input means is provided in a line by forming a predetermined distance from the top to the bottom of the front of the kiosk body. That is, the heights of the plurality of input means are different from each other. Accordingly, children of short height can input their own information to the kiosk through the microphone at the bottom, and teenagers who are taller than children can input their desired information to the kiosk through the microphone of the middle, youth Taller adults can enter their desired information into the kiosk through the microphone at the top. Therefore, speech recognition is possible according to the height of people of all ages.

Second, the user's needs can be quickly applied. The voice recognition kiosk according to the present invention allows a user to input desired information such as order purchase, ticketing, etc. by touching the display, but quickly displays information desired by the user through voice recognition or inputs information. It is also possible. Accordingly, the user can quickly apply the order, purchase, ticketing, etc. desired by the user.

1 is a perspective view of a speech recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention.
2 is a front view of a speech recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention.
3 is a block diagram of a speech recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention.
4 is a view schematically showing a state in which a voice recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention performs micro voice recognition provided at the bottom.
5 is a view schematically showing a state in which a voice recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention performs micro voice recognition provided in a middle stage.
6 is a view schematically showing a state in which a voice recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention performs micro voice recognition provided at the top.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. However, some components irrelevant to the subject matter of the invention will be omitted or compressed, but the omitted components are not necessarily required in the present invention, and may be used in combination with those skilled in the art to which the present invention pertains. You can.

1 is a perspective view of a speech recognition kiosk with a plurality of input means according to an embodiment of the present invention, and FIG. 2 is a front view of a speech recognition kiosk with a plurality of input means according to an embodiment of the present invention, FIG. Is a block diagram of a speech recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention.

1 to 3, a voice recognition kiosk (hereinafter referred to as a "voice recognition kiosk") equipped with a plurality of input means according to an embodiment of the present invention is a kiosk body 100, a first input means ( 110, a second input means 120, a noise pickup unit 130, a voice recognition unit 140, a control unit 150, an output unit 160 and a database 190.

In the present invention, a voice recognition kiosk is mainly described, but as an application that replaces the kiosk, it can be implemented by using voice recognition in a ticket machine, a vending machine, or a digital information display (DID).

The kiosk main body 100 is configured to have a front surface and a rear surface in a predetermined size, and is formed in a substantially rectangular parallelepiped shape. The components constituting the speech recognition kiosk are located in the kiosk main body 100, and the kiosk main body 100 may serve as a case surrounding the inside. The kiosk main body 100 may be installed standing on the ground in a department store, airport, subway, restaurant, or the like, or fixed to a wall surface.

The first input means 110 is configured to receive the talker's voice signal. The first input means 110, for example, a microphone may be used. The first input means 110 may be provided in plural on the left or right side of the kiosk main body 100, and may be arranged in a vertical direction in front of the kiosk main body 100. At this time, the plurality of first input means 110 may be provided at the top, middle, and bottom of the kiosk main body 100, and the first input means 110 is a total of three, one at the top, middle, and bottom. Dogs may be provided. Here, as the microphone used as the first input means 110, an omni-directional microphone array may be used. In addition, the plurality of first input means 110 may be provided at intervals of 5 cm or more, respectively, within the vertical length of the kiosk body 100.

Omni-directional microphones have the same sensitivity in all directions, so they absorb sound in all directions without directing to a specific direction. It is mainly used to capture the reverberation of the recording space by receiving signals from all directions, front and rear, and left and right. In addition to this, it is used as a microphone used in close proximity to sound sources, such as a 'lavaliere mic (aka pin microphone)', because it reacts equally to sound signals coming from all directions without a proximity effect due to its physical characteristics. Here, the proximity effect refers to a phenomenon in which the volume of the low-frequency band generated when approaching the microphone is increased. It can be used to accommodate large-scale orchestras or hall and studio reverberations. It is used even when the direction of the sound source is not constant. Therefore, the voice recognition kiosk according to the present invention can output the information requested by the speaker by receiving the voice signal of the speaker even in a public place where the direction of the sound source is not constant.

The second input means 120 is another configuration that the talker can use when requesting information he or she wants from the speech recognition kiosk. The second input means 120 is a means for sensing the touch and transmitting the sensed input information. As the talker touches the kiosk screen, input information may be transmitted to the controller 150.

The noise pickup unit 130 is connected to the first input means 110 to remove ambient noise. The noise pickup unit 130 is provided on the rear of the kiosk main body 100 and is connected to the first input means 110, and when the talker transmits the voice signal to the first input means 110, the talker's voice signal is transmitted. Except for this, other sounds that occur around the talker can be considered as noise and blocked. Since kiosks are generally installed and used in public places, there is no choice but to generate ambient noise. Therefore, the noise pickup unit 130 blocks the ambient noise when the talker transmits the voice signal in front of the speech recognition kiosk, so that only the talker's voice signal is received.

For example, the noise pickup unit 130 may use a scheme such as active noise control (ANC), noise canceling, or active noise reduction (ANR). Specifically, when ambient noise is detected by the noise pickup unit 130, the interference may be generated in the circuit to cancel the noise to block the ambient noise.

The voice recognition unit 140 is configured to recognize the speaker's voice and determine information requested by the speaker. When the voice of the talker is input through the first input means 110, the voice recognition unit 140 analyzes the voice of the talker in the analog form, determines the input code indicated by the talker, and transmits it to the controller 150. Here, the input code means a code to be input to the central processing unit 120, and may be the same code that is transmitted to the control unit 150 when a corresponding item is selected from the second input means 120. That is, the voice recognition unit 140 may serve to determine and deliver what service item the voice spoken by is spoken.

For example, the voice recognition unit 140 may detect a start point and an end point of the voice signal input through the first input means 110 and extract characteristics of the detected voice signal. Here, a filtering process of amplifying the input voice signal and removing noise outside the audible frequency band may be included. Thereafter, the extracted feature data can be compared with the voice data codes stored in the database 190 in advance to determine the most similar voice data codes. Here, the feature data may mean a general “feature parameter” used in speech recognition technology. Then, the input code corresponding to the determined voice data code can be grasped and transmitted to the control unit 150.

In addition, the voice recognition unit 140 may perform speech recognition by learning the speaker's voice. At this time, the voice recognition unit 140 may input a large amount of voice and text represented by the voice into the learner in pairs and generate a learning model. Subsequently, when voice information is input to the learning model, text corresponding to the voice information is returned. Such a structure may use Recurrent Neural Networks (RNN), which is a cyclic neural network capable of handling variable length data during deep learning.

The RNN is connected so that the result of the hidden layer goes back to the input of the same hidden layer. This feature allows the RNN to have characteristics that can be considered in terms of order or time. RNN is useful for dealing with sequence data because it has the characteristics that can be judged by considering the sequential aspect. A typical form of sequence data is data such as a 'sentence'. In the case of words in letters, the meaning of the current word can be interpreted through the relationship with the previous word, not the meaning of the current word alone. In addition, RNN can be used to process sequence or time series data such as genes, handwriting, voice signals, data sensed by sensors, and stock prices.

The control unit 150 is a component that retrieves response information for a corresponding selection item from the database 190 according to the input code transmitted from the voice recognition unit 140 and transmits it to the output unit 160. The response information may include response information for voice and response information for a screen. Accordingly, the response information for the voice may be output to the speaker 180, and the response information for the screen may be output to the display 170.

The output unit 160 is configured to receive and output response information for a voice and response information for a screen from the control unit 150. The output unit 160 may include a display 170 and a speaker 180.

The display 170 is provided on the front surface of the kiosk body 100. The display 170 may be formed in a predetermined size to output various user menus set in the speech recognition kiosk and information requested by the talker by voice or touch, so that the talker can visually confirm.

The speaker 180 is provided on one side of the kiosk main body 100 to guide the speaker with various sounds.

The database 190 is configured to store data on various user menus and various response information that are set in advance. Graphic data and voice data corresponding to the feature data may be stored in the database 190.

Hereinafter, it will be described with reference to the drawings that people of various ages use the speech recognition kiosk according to the present invention.

4 is a view schematically showing a state in which a voice recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention performs micro voice recognition provided at the bottom.

4, the speech recognition kiosk according to the present invention can be easily used by people of various ages. Also, people generally have different heights depending on their age. Moreover, kiosks are generally devices of high height and size. Accordingly, the size of the display 170 shown in the kiosk is also relatively large. Depending on the size of the display 170, children or adolescents having a small height may have difficulty using a kiosk.

That is, since the height of the key is small, it is difficult to select a menu that is relatively higher than the height of the user, and it can be cumbersome because it requires the assistance of another person every time to solve the problem.

However, in the speech recognition kiosk according to the present invention, the kiosk main body 100 is provided with a plurality of first input means 110 in a line at a predetermined interval in the longitudinal direction. These first input means 110 are located at the bottom, middle and top, respectively. Accordingly, children or adolescents having a shorter height can transmit a voice signal to the first input means 110 provided at the bottom by referring to information or order information they receive.

The voice recognition unit 140 may recognize the speaker's voice based on the voice signal received from the first input means 110 provided at the bottom to determine information requested by the speaker. Thereafter, the information desired by the talker is transmitted to the controller 150, and finally, by the control of the controller 150, the output unit 160 may output information requested by the talker to the speaker 180 and the display 170. In this way, as the first input means 110 is provided at the bottom of the kiosk main body 100, the speech recognition rate can be increased even for children or adolescents having a small height.

In addition, children who are tall or relatively tall can request information or order they want through the first input means 110 provided in the middle of the kiosk main body 100. This will be described with reference to FIG. 5.

5 is a view schematically showing a state in which a voice recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention performs micro voice recognition provided in a middle stage.

As shown in FIG. 5, in the voice recognition kiosk according to the present invention, the kiosk main body 100 is provided with a plurality of first input means 110 lined up in a longitudinal direction at predetermined intervals. These first input means 110 are located at the bottom, middle and top, respectively. Accordingly, adolescents who are taller than a child can transmit a voice signal to the first input means 110 provided for interruption by referring to information or an order to be provided.

The voice recognition unit 140 may recognize the speaker's voice based on the voice signal received from the first input means 110 provided in the interruption to determine information requested by the speaker. Thereafter, the information desired by the talker is transmitted to the controller 150, and finally, by the control of the controller 150, the output unit 160 may output information requested by the talker to the speaker 180 and the display 170. In this way, as the first input means 110 is provided at the middle of the kiosk main body 100, the speech recognition rate can be increased even for adolescents having a height higher than that of a child.

In addition, adults having a high height can request information or an order they want through the first input means 110 provided at the top of the kiosk body 100 at the top. This will be described with reference to FIG. 6.

6 is a view schematically showing a state in which a voice recognition kiosk equipped with a plurality of input means according to an embodiment of the present invention performs micro voice recognition provided at the top.

As shown in Fig. 6, in the speech recognition kiosk according to the present invention, the kiosk main body 100 is provided with a plurality of first input means 110 lined up in a longitudinal direction at predetermined intervals. These first input means 110 are located at the bottom, middle and top, respectively. Accordingly, it is possible for a tall adult to transmit a voice signal to the first input means 110 provided at the top by referring to information or an order to be provided.

The voice recognition unit 140 may recognize the speaker's voice based on the voice signal received from the first input means 110 provided at the upper end to determine information requested by the speaker. Thereafter, the information desired by the talker is transmitted to the controller 150, and finally, by the control of the controller 150, the output unit 160 may output information requested by the talker to the speaker 180 and the display 170. Thus, as the first input means 110 is provided at the top of the kiosk main body 100 at the top, it is possible to increase the speech recognition rate even for adults having a high height.

On the other hand, a sensor unit is provided around each of the first input means 110 to detect the key height of the talker standing in front of the kiosk main body 100, and to receive a voice signal to the talker according to the detected key height of the talker. One can guide the input means (110).

Specifically, a sensor unit (not shown) may be provided around the first input means 110 provided at the bottom, middle, and top of the kiosk body 100 at the front. An infrared sensor, an ultrasonic sensor, and an optical sensor may be used as the sensor unit.

The infrared sensor, the ultrasonic sensor, and the optical sensor may be used as only one type of starch at the bottom, middle, or top, and one type may be used for each. That is, an ultrasonic sensor may be used at the bottom, an infrared sensor may be used at the middle, and an optical sensor may be used at the top. In addition, at least one infrared sensor, an ultrasonic sensor, and an optical sensor may be used in combination with at least one of the lower, middle, and upper ends.

The provided sensor unit senses the key height of the talker standing in front of the kiosk main body 100, and the first input means 110 that approaches the key height of the talker among the first input means 110 provided at the bottom, middle, and top. ) To guide the ignition.

First, the sensor unit sends a sensing signal for sensing toward the front of the kiosk main body 100. That is, the infrared sensor emits infrared rays, the ultrasonic sensor emits ultrasonic waves, and the optical sensor emits light. At this time, if the key height of the talker is close to the height of the first input means 110 provided at the bottom, the sensor unit provided at the bottom may be received again by infrared or ultrasonic or light reflection on the talker's body.

However, since the height of the talker does not reach the sensor unit provided at the top and the middle, infrared rays, ultrasonic waves or light cannot be reflected. Therefore, since infrared, ultrasonic, or light is reflected only the sensor unit provided at the bottom, the controller 150 can sense that the key height of the talker is near the first input means 110 provided at the bottom, and accordingly The LED provided near the first input means 110 provided at the bottom lights or the speaker 180 outputs and guides a voice requesting information desired by the speaker to the first input means 110 provided at the bottom. have.

In addition, if the key height of the talker is close to the height of the first input means 110 provided at the middle, the sensor unit provided at the bottom and the middle may be received again by infrared or ultrasonic or light reflection on the talker's body. However, since the height of the talker does not reach the sensor unit provided at the top, infrared rays, ultrasonic waves, or light cannot be reflected. Therefore, since infrared, ultrasonic, or light reflections are only the sensor unit provided at the bottom and the sensor unit provided at the middle, the controller 150 says that the height of the talker is near the first input means 110 provided at the middle. LED that is provided in the vicinity of the first input means 110 provided in the middle can be lit or the speaker 180 requests the desired information to the first input means 110 provided in the middle with the speaker 180. You can output and guide the voice.

Finally, if the key height of the talker is close to the height of the first input means 110 provided at the top, the sensor units provided at the bottom, middle and top are all reflected by infrared or ultrasonic waves or light reflected on the talker's body and received again. Can be. Therefore, since the infrared, ultrasonic, or light reflections are both the sensor unit provided at the bottom, the sensor unit provided at the middle, and the sensor unit provided at the division, the controller 150 has a first input having a key height of the talker at the top. It can be sensed that it is near the means 110, and accordingly, the LED provided near the first input means 110 provided at the top lights up or the speaker 180 moves to the first input means 110 provided at the top. The speaker can output and guide a voice requesting the desired information.

As described in detail above, the speech recognition kiosk according to the present invention is capable of custom speech recognition for people of all ages. The speech recognition kiosk according to the present invention is equipped with a microphone as a plurality of input means on the front of the kiosk main body 100. The plurality of input means is provided in a line by forming a predetermined distance from the front top to the bottom of the kiosk body 100. That is, the heights of the plurality of input means are different from each other. Accordingly, children of short height can input their own information to the kiosk through the microphone at the bottom, and teenagers who are taller than children can input their desired information to the kiosk through the microphone of the middle, youth Taller adults can enter their desired information into the kiosk through the microphone at the top. Therefore, speech recognition is possible according to the height of people of all ages.

In addition, the user's needs can be quickly applied. In the voice recognition kiosk according to the present invention, it is possible for a user to input information desired by the user by touching the display 170, such as order purchase, ticketing, etc. , It is also possible to enter information. Accordingly, the user can quickly apply his / her desired order, door purchase, ticketing, and the like.

The above-described preferred embodiments of the present invention are disclosed for the purpose of illustration, and those skilled in the art having ordinary knowledge of the present invention will be able to make various modifications, changes and additions within the spirit and scope of the present invention. And additions should be considered to fall within the scope of the claims of the present invention.

100: kiosk body
110: first input means
120: second input means
130: noise pickup unit
140: speech recognition unit
150: control unit
160: output unit
170: display
180: speaker
190: database

Claims (9)

In the speech recognition kiosk equipped with a plurality of input means,
A kiosk body provided with a display in a predetermined size;
A plurality of first input means provided on a front surface of the kiosk body to receive a voice signal of a talker;
A noise pickup unit connected to the plurality of first input means to remove ambient noise of a speaker; And
Includes a; output unit for outputting information requested by the speaker on the display based on the voice signal of the speaker from which the ambient noise has been removed.
The noise pickup unit is provided in hardware on the back of the kiosk body,
The rear surface of the kiosk body is an opposite surface to the front surface on which the plurality of first input means are disposed,
The plurality of first input means is a microphone,
The microphone is provided at regular intervals from the top to the bottom of the kiosk body,
A sensor unit is provided around the microphone,
The sensor unit detects the key height of the talker, and guides to request micro information that is close to the key height of the talker among the microphones.
The guidance is a voice recognition kiosk with a plurality of input means, characterized in that the LED provided near the microphone is lit.
delete According to claim 1,
The microphone is provided one at each of the top, middle and bottom of the kiosk body,
The microphone is connected to one noise pickup,
A speech recognition kiosk with a plurality of input means, characterized in that the noise pickup unit removes ambient noise of the microphone.
According to claim 1,
The microphone is a voice recognition kiosk equipped with a plurality of input means, characterized in that using an omni-directional microphone array.
According to claim 1,
When the plurality of first input means are arranged in a row on the left or right side of the kiosk main body, a voice recognition kiosk with a plurality of input means, characterized in that an interval of 5 cm or more is provided within a vertical length of the kiosk main body. .
According to claim 1,
It includes a voice recognition unit for recognizing the voice of the talker to determine the information requested by the talker,
The speech recognition unit recognizes the speech of the talker using a Recurrent Neural Networks (RNN) model, thereby determining and transmitting the information requested by the talker, the speech recognition kiosk with a plurality of input means.
delete delete According to claim 1,
The sensor unit is a voice recognition kiosk having a plurality of input means, characterized in that at least one of an ultrasonic sensor, an infrared sensor and an optical sensor.

Images