KR20190100703A - Speaker with motion recognition technology using artificial intelligence and control method therefore - Google Patents

Abstract

The present invention relates to an artificial intelligent speaker capable of a movement using sound source position recognition technology, which comprises: a voice input unit having at least one microphone for receiving an externally generated voice signal; a sound source tracking unit for tracking a sound source of a received voice signal through sound source recognition technology using an arduino; a motor driving unit for rotating an artificial intelligent speaker toward a position of the tracked sound source; and a control unit for receiving direction information from the sound source tracking unit to drive the motor so as to allow the artificial intelligent speaker to be oriented in a sound source direction. The sound source tracking unit can determine a threshold of the magnitude of a voice to distinguish a voice of a speaker from other voices, and can extract a coordinate of a sound source of the voice signal exceeding the threshold to drive the motor through the control unit.

Description

Artificial speaker capable of movement using sound source position recognition technology and its control method {SPEAKER WITH MOTION RECOGNITION TECHNOLOGY USING ARTIFICIAL INTELLIGENCE AND CONTROL METHOD THEREFORE}

The present invention relates to an artificial intelligence speaker, and more particularly, to an artificial intelligence speaker capable of movement using a sound source position recognition technology for tracking the position of a sound source to move in a direction in which a user speaks.

Artificial intelligence (AI) speakers have been developed that can recognize a user's voice, play songs easily, or control household appliances connected to a communication network. As shown in FIG. 1, AI speakers are sold as a product in the field of intelligent personal assistants (IPAs), and many global companies are striving to gain an edge in this field. Further, as shown in FIG. 2, market research firm TrendForce predicts that the global voice recognition market will reach about 16 billion in 2021 from about $ 2.6 billion in 2016, and research consultant Gartner Forecasts that the AI speaker market will grow from about $ 3.6 billion in 2015 to about $ 2.1 billion in 2020.

As described above, the AI speaker market is expected to continue to grow in the era of the fourth industrial revolution, but most of the conventional AI speakers merely output voice. In fact, when talking, non-verbal elements, such as actions or facial expressions, may contain much more information than verbal elements in speech, so users of conventional AI speakers, which mostly output only voice, are familiar with AI speakers. There was a problem that it is difficult to feel that they are communicating. As shown in Fig. 3, the 7-38-55 law published by Albert Meravion, professor of psychology at the University of California, Los Angeles (UCLA), is more expressive than the verbal content (7%) contained in speech. Visual images such as gestures (55%) have a much greater impact on communication. Therefore, the conventional artificial intelligence speaker that communicates only by voice is excluded from movements such as facial expressions and gestures, so that the width of the human speaker can be significantly reduced.

In addition, the conventional artificial intelligence speaker uses a sound source location recognition function used in an intelligent control service, a noise measuring device, or a voice tracking camera as shown in FIG. 4, so that a user calls the name of the artificial intelligence speaker, Speaking of a specific function such as a weather check, the function was limited to the way that the AI speaker executes the function. Therefore, many users of the artificial intelligence speaker often evaluate that the result of using the conventional artificial intelligence speaker is 'the answer is good but the conversation is not good'. 'Conversation' here can be said to 'go through' when communication between the two speakers is smooth. When the user and the AI speaker are talking to each other, the user tries to talk, but most of the conventional AI speaker performs a simple answer task rather than communicating. That is, the conventional artificial intelligence speaker uses only hearing, which is one of the five senses of humans, and in many cases, does not use the other movement, there is a problem in that the interaction between the user and the artificial intelligence speaker is difficult.

Conventional patent document WO 2014/109422 A1 discloses a voice tracking device and a control method thereof as shown on the right side of FIG. 4. However, this prior art discloses only a camera that tracks a sound source using a microphone and rotates toward the tracked sound source, and a method in which an artificial intelligence speaker makes contextual actions and sounds to communicate better with a user is disclosed. It is not.

In particular, since the AI speaker often spends most of its time with the user at home, the sound source is output from an electronic device such as a CD player, a computer, a mobile phone, and the like. There is a problem in that only by listening to the sound source, the user's sensitivity that requires familiarity such as pets cannot be satisfied.

On the other hand, among the artificial intelligence speakers of the prior art there is a device to deliver the life of music to the user by adjusting a plurality of LED lights when playing music. However, such a configuration also excludes various movements of the artificial intelligence speaker, and thus cannot express the atmosphere to be transmitted by the rotation and the rotation speed suitable for the beat and mood, and thus there is a limitation as a static communication method. In addition, the conventional artificial intelligence speaker has a problem in that it does not express various movements to help communication with the user while most of the cases merely convey the meaning of sound as text of a messenger.

An object of the present invention is to provide an artificial intelligence speaker that communicates with the user by movement and sound after detecting the user by the sound source location recognition technology for effective communication. In particular, the present invention uses a sound source position recognition technology that can effectively track the position of the sound source through the sound signal of the sound source, it was possible to switch the direction of the artificial intelligence speaker in the sound direction.

You can also interact with the user by gesturing for different situations, such as bowing your head when you greet the user, answering yes or no, nodding your head, playing music, dancing, etc. This is to provide a much better AI speaker.

An artificial intelligence speaker capable of moving using a sound source position recognition technology according to a first embodiment of the present invention for solving the above problems is a voice input unit having at least one microphone for receiving an externally generated voice signal, Arduino ( arduino), the sound source tracking unit that tracks the sound source of the received voice signal through the sound source position recognition technology, the motor driver that rotates the AI speaker toward the tracked sound source, and the sound source tracking unit receives the direction information from the motor. It may include a control unit for driving the AI speaker in the direction of the sound source, the sound source tracking unit determines a threshold of the loudness of the voice to distinguish the voice of the talker and other voices, and exceeds the threshold The motor may be driven through the control unit by extracting the coordinates of the sound source of the voice signal.

In the artificial intelligence speaker capable of moving using the sound source position recognition technology according to the second embodiment of the present invention, the sound source tracker stores voice signal coordinates equal to or greater than a threshold, so that the most frequent azimuth and elevation ) May be set to reflect the value of the AI speaker when implementing the movement.

According to another embodiment of the present invention, an artificial intelligence speaker capable of moving using a sound source position recognition technology may distinguish a received voice signal from a sound source tracker and selectively implement different actions and sounds according to the type of the divided voice signals. have.

According to another embodiment of the present invention, an artificial intelligence speaker capable of moving using a sound source position recognition technology may further include a rotatable arm, and the motor may include a first motor and a second motor. The first motor rotates the AI speaker about the vertical axis, and the second motor can implement the movement of the arm.

According to an embodiment of the present invention, a method of controlling an artificial speaker capable of moving using a sound source position recognition technology includes a voice input unit, a sound source tracking unit, a motor driving unit, and a controller. Receiving a voice signal, tracking a sound source of the received voice signal, rotating the artificial speaker to the position of the tracked sound source, and setting a threshold value of the voice size to distinguish the user voice from other voices. And extracting the coordinates of the sound source of the voice signal exceeding the threshold value to determine whether the voice is the voice of the talker.

According to another embodiment of the present invention, a method of controlling an artificial speaker capable of moving using sound source position recognition technology may include determining whether a voice of a talker is a voice by storing sound coordinates of a threshold value or more, so that azimuth and altitude values are most frequently used. It may further comprise setting to reflect when the artificial intelligence speaker implements the movement.

According to another exemplary embodiment of the present invention, a method of controlling an artificial speaker capable of moving using sound source position recognition technology may include different actions and sounds according to types of voice signals in which the artificial intelligence speaker is divided by classifying received voice signals. The method may further include a motion implementing step for selectively implementing.

The artificial intelligence speaker according to the present invention can implement non-verbal elements such as actions or facial expressions according to the situation, and thus, there is an effect that can effectively communicate with and interact with a person. The present invention can change the direction of the artificial intelligence speaker in a sound direction by using a sound source position recognition technology that can effectively track the position of the sound source through the sound signal of the sound source.

In addition, the present invention can implement various movements of the artificial intelligence speaker. In other words, the user and the AI speaker can be used to implement gestures suitable for various situations, such as bowing the head while greeting the user, answering yes or no, nodding the head, playing music, and dancing. Has a much better effect of interaction.

1 is a diagram illustrating a conventional artificial intelligence speaker.
2 is a graph showing the global voice recognition market and the global smart speaker market forecast.
3 is a graph illustrating the law of Meravion.
4 is a flowchart schematically illustrating a process of a conventional intelligent control service and a noise measuring instrument and a voice tracking camera.
5 is a schematic block diagram of an artificial intelligence speaker of the present invention.
FIG. 6 is a diagram schematically illustrating a method in which an artificial intelligence speaker of the present invention communicates with a user when there is no movement and when there is no movement.
7 is a diagram illustrating a set up code for creating an environment in which the artificial intelligence speaker of the present invention can operate.
8 is a diagram illustrating a threshold code for maximally filtering out unnecessary noise in an artificial intelligence speaker according to the first and second embodiments of the present invention.
FIG. 9 is a diagram illustrating a signal check code for checking whether a voice signal is input to the artificial intelligence speaker of the present invention and coping with a situation.
10 is a view schematically showing the movement of the artificial intelligence speaker of the present invention.
11 is a view schematically showing the expected effect of the artificial intelligence speaker of the present invention play with the companion animal.
12 is a flowchart illustrating a method of controlling an artificial intelligence speaker according to an embodiment of the present invention.

The foregoing objects, features, and advantages will become more apparent from the following examples taken in conjunction with the accompanying drawings.

Specific structural to functional descriptions are merely illustrated for the purpose of describing embodiments in accordance with the inventive concept, and embodiments in accordance with the inventive concept may be embodied in various forms and are described in the specification of the present application. It should not be construed as limited to these.

Embodiments according to the concept of the present invention can be variously modified and have a variety of forms specific embodiments will be illustrated in the drawings and described in detail in the specification of the present application. However, this is not intended to limit the embodiments in accordance with the concept of the present invention to a particular disclosed form, it should be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the present invention.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. As used herein, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, action, component, part, or combination thereof that is described, and that one or more other features or numbers, It is to be understood that it does not exclude in advance the possibility of the presence or addition of steps, actions, components, parts or combinations thereof.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Like reference numerals in the drawings denote like elements.

As shown in FIG. 5, the voice tracking device 100 of the artificial intelligence speaker 1 includes a voice input unit 110, a sound source tracking unit 120, a motor driver 130, a controller 140, and an image photographing unit ( 150).

The artificial intelligence speaker (1) has a portion corresponding to the human head, body, arms, etc. in order to induce a sense of familiarity similar to the shape of the human being, so that the artificial speaker (1) Can be moved as a whole. For example, the artificial intelligence speaker 1 may be provided with a rotatable arm. The rotatable arm may be placed on a rotatable arm, and a plastic hemisphere having a diameter of about 12 cm may be attached thereto to form a face. In addition, in order to prevent the internal skeleton structure from being exposed to the outside, it may be configured to cover the inside of the hardboard paper by covering it with a round and cover the whole with a cloth. The AI speaker 1 may be formed in various forms, such as a dog, a ghost, as well as a person, and since a gesture is included in actual movement, a fixed arm may be made of paper and attached to both sides. The expression of the face can be implemented using an LED or a screen switching display, and the shape of the eye can be attached so that the user can stare at the conversation. The LED may be configured to rotate 360 degrees to move the voice position in the recognized direction. Through this, it is possible to satisfy aesthetics while allowing various expressions.

The voice input unit 110 is for receiving an externally generated voice signal and may include at least one microphone. The microphone may be provided with various devices for removing noise generated from an external voice signal. In addition, since the motor rotates the AI speaker 1 to the position of the tracked sound source so that the AI speaker 1 is directed in the direction of the sound source, the microphone is the AI speaker that will be directed in the sound source direction inside the AI speaker 1. It may be provided toward the front of (1). The microphone is preferably located in the head portion of the artificial intelligence speaker that the user is looking at, but is not limited thereto. On the other hand, the microphone may be built in a camera such as a webcam provided in the artificial intelligence speaker may have a voice recognition function in addition to the image recognition function. The microphone may be made of at least one.

The sound source tracking unit 120 may track the sound source of the voice signal using the received voice signal. The sound source tracking unit 120 may have a sound source position recognition technology using Arduino. Arduino refers to a board complete with an open source, single board microcontroller and related development tools and environments. For example, the sound source tracking unit 120 using one of Arduino's OpenCM 9.04 receives direction information to control the movement of the AI speaker. In addition, the sound source tracking unit 120 may determine the received voice signal and drive the motor through the controller 140 to implement the movement of the artificial intelligence speaker. Furthermore, the sound source tracking unit 120 may be configured to recognize who speaks among those who have registered their voices by using different waveforms of voices for each person, and to give an answer that is appropriate for the user. The voice recognition may be configured only at a location where the user's voice is heard.

On the other hand, as shown in Figure 6, for example, a simple model of the artificial intelligence speaker 1, as shown in the figure shown on the left of Figure 6 when the user asks a question, the function of the artificial intelligence speaker is executed without any movement It may be assumed that a situation in which the user asks a question and adds a slight shake when the user asks a question and the AI speaker answers, as shown in the drawing on the right side of FIG. 6. In this case, the AI speaker commonly answers a question such as 'how about the weather today', but the user may not use the AI speaker even with the simple movement of the AI speaker as shown in the right side of FIG. 6. You can feel the improved communication with the.

7 to 9 illustrate a code for using the MATLAB to allow the artificial intelligence speaker 1 to communicate with the user through movement and sound after detecting the user with the sound source location recognition technology for effective communication.

7 shows code for creating an environment in which the AI speaker 1 can operate. According to this, the position of the sound is determined by using an array of microphones, and then an environment in which the position coordinates are received by TCP is created. In addition, the OpenCM environment for using the two actuators (actuator) and the video recording unit 150, such as a webcam, the audio input unit 110, such as a microphone is set to be used. And initialize the variables to be used later.

8 shows a code representing a first embodiment and a second embodiment for maximally filtering out unwanted noise.

In the first exemplary embodiment, the sound source tracking unit 120 determines a threshold of a loudness of a voice to distinguish between a talker's voice and other voices, and extracts coordinates of a sound source of a voice signal exceeding the threshold. 140 may drive the motor. In other words, if the loudness of the voice does not exceed the threshold by setting a threshold, the coordinates of the voice received at that time are ignored.

In the second embodiment, the sound source tracking unit 120 stores voice signal coordinates of a threshold value or higher to reflect the most frequent azimuth and elevation values when the AI speaker implements movement. will be. Since the likelihood of noise such as ambient noise is repeated at the same position is low, the coordinates of a specific position are repeatedly input when the user speaks, but sounds like noise are less likely to have repeated coordinate values. Thus, the sound source tracker 120 according to the first embodiment accepts the voice signal coordinate values due to the loud noise that may actually occur between, before, or after the user speaks, but in the end, Accordingly, since the most frequent voice signal coordinate values in the stored data rarely change, the artificial intelligence speaker according to the second embodiment can continuously look at the user. On the other hand, the first embodiment and the second embodiment can be implemented without restriction in order, respectively.

In summary, as shown in FIG. 12, the method of controlling the artificial intelligence speaker 1 includes receiving an externally generated voice signal (S1), tracking a voice of the received voice signal (S2), and tracked the Rotating the artificial intelligence speaker 1 to the position of the sound source (S3), to determine the threshold of the loudness of the voice to distinguish the user's voice and other voices, and extract the coordinates of the sound source of the voice signal exceeding the threshold And checking whether the voice is the voice of the talker (S4). In addition, the step S4 of confirming whether the voice of the talker is stored may further include storing sound coordinates equal to or greater than a threshold value to reflect the most frequent azimuth and altitude values when the AI speaker implements the movement. have.

9 is a code illustrating a third embodiment for determining whether a voice signal has been received and allowing a situation to be dealt with according to a situation.

The third embodiment determines a signal in a function called signal_check of the code shown in FIG. For example, when the AI speaker 1 greets, the signal variable value is 1, and when the answer is yes or no, the value is 2, and the user's fashion is taken and recognized. Change the chord to 3 if you are playing a song, or 4 if you are playing a song and dancing. Also, the value of the signal variable when simply turning to the position of the sound is zero. Through this, the artificial intelligence speaker (1) can make a behavior and sound in accordance with the situation. Accordingly, the movement of various emoticons to help the communication with the user can also be implemented in the artificial intelligence speaker (1) has the effect of giving a user friendly.

That is, as shown in FIG. 12, the control method of the AI speaker may include: a motion implementing step of dividing the received voice signal so that the AI speaker may make different actions and sounds according to the type of the divided voice signal ( S5) may be further included.

Through these embodiments, as shown in FIG. 10, the AI speaker 1 can communicate with the user not only by voice, but also by actions made by devices such as software code and two motors. For example, you can implement gestures for different situations, such as bowing your head in greetings, answering yes or no, nodding or intercepting, playing music and dancing, and expecting all of these actions. As you can see and do it, the interaction between the user and the AI speaker 1 can be much better. Furthermore, since the user can implement various functions such as viewing the user's voice position and nodding or shaking the head, the AI speaker 1 feels as if the pet is welcoming the owner. There is an effect that can convey the feelings such as niceness to the user. In addition, even in the music playback function, the artificial intelligence speaker 1 is not simply a static communication using the LED light, etc. when playing music, the atmosphere that the music is intended to deliver by adjusting the rotation and rotation speed according to the mood, according to the beat Can be expressed well.

On the other hand, the motor driver 130 may rotate the microphone to the position of the sound source tracked through the sound source tracking unit 120. The motor driver 130 may be provided with at least one motor. The motor driver 130 may rotate the artificial intelligence speaker 1 to rotate the microphone of the voice input unit 110 included in the artificial intelligence speaker 1 in the direction of the sound source, and the microphone may capture the image capture unit ( When attached to the camera of 150, the camera may be rotated in the direction of the sound source so that the microphone may also be rotated in the direction of the sound source. In addition, the motor driver 130 may rotate the microphone or the camera in the vertical, horizontal, left and right directions, without being limited thereto, and may express movement in various directions.

In addition, the motor of the motor driver 130 may include a first motor and a second motor. The first motor may rotate the AI speaker 1 about the vertical axis, and the second motor may be configured to implement the movement of the rotatable arm provided in the AI speaker 1. Through this, the artificial intelligence speaker 1 can not only rotate the body but also rotate the arm, and can implement various movements for interacting with the user, such as greeting and dancing.

The controller 140 is for driving a motor, and receives a direction information from a sound source tracking unit having a sound source position recognition technology using an arduino to drive the motor so that the microphone can be directed in the sound source direction. In addition, it may be made to control the overall operation of the voice tracking device 100. Meanwhile, the controller 14 may receive a voice signal through a microphone array and then track the position of the sound source in real time using a beam forming technique. Depending on how the microphone array is arranged, the range that can be tracked varies. If it is arranged on a plane, it can track two directions of X, Y, and three directions of X, Y, and Z. It is also possible to use a spherical array to enclose a sphere in a plane, tracking a full 360 degrees using a planar technique rather than a stereoscopic technique. Furthermore, the artificial intelligence speaker 1 combines the data received through various sensors with various Internet of Things (IoT) technologies that exchange data with the Internet in real time to receive information on the situation to the controller 140 via the Internet. It can also be implemented to provide appropriate answers and actions for the user.

The image capturing unit 150 may capture a user as an image, and may include at least one or more cameras according to a usage environment. The camera of the image capturing unit 150 may be configured to incorporate a microphone of the voice input unit 110, or may be separated from each other. The camera of the image capturing unit 150 is not limited to a function of capturing a picture, and may be provided with various functions for communicating with a user, such as a video capturing function. In addition, the image capturing unit 150 rotates the camera in a direction in which the voice is generated through the voice position recognition technology and the movement provided in the artificial intelligence speaker 1, and uploads and analyzes the photo data on the Internet. It may be.

Through this configuration, the AI speaker 1 may have a function of moving according to the situation with an emphasis on communication with the user, thereby satisfying the needs of users who need to communicate with the AI speaker 1 intimately. Can be. In addition, as shown in Figure 11, it can be used for the purpose of playing with children and pets. Children can be a great play buddy of children if the artificial intelligence speaker (1) that adds various movements according to the child's reaction, for example, play or play (hand play). In the case of pets, for example, in the case of cats, people often play with moving toys. In addition, if you use the moving artificial intelligence speaker to recognize the position of the cat's footsteps can also be a good ride.

 As mentioned above, although preferred embodiments of the present invention have been described with reference to the accompanying drawings, the present invention is not limited thereto, and those skilled in the art to which the present invention pertains may have the technical idea of the present invention. Substitution, modification and / or modification are possible without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the above-described embodiment, but should be defined by the claims below and equivalents thereof.

1: artificial intelligence speaker 110: voice input unit
120: sound source tracking unit 130: motor drive unit
140: control unit 150: image capturing unit

Claims (7)

A voice input unit 110 provided with at least one microphone for receiving an externally generated voice signal;
A sound source tracking unit 120 for tracking a sound source of the received voice signal through a sound source position recognition technology using an arduino;
A motor driver 130 for rotating the artificial intelligence speaker 1 toward the position of the tracked sound source; And
It receives a direction information from the sound source tracking unit 120 includes a control unit 140 to drive the motor to the AI speaker 1 in the direction of the sound source,
The sound source tracking unit 120 sets a threshold of the loudness of the voice to distinguish the voice of the talker and other voices, extracts the coordinates of the sound source of the voice signal exceeding the threshold, and controls the controller 140. To drive the motor through
AI speaker that can move using sound source location technology.
The method of claim 1,
The sound source tracking unit 120 stores voice signal coordinates equal to or greater than a threshold so that the AI speaker 1 reflects the values of azimuth and elevation most frequently when the AI speaker 1 implements the movement. Set
AI speaker that can move using sound source location technology.
The method according to claim 1 or 2,
The sound source tracking unit 120 classifies the received voice signals, and may selectively implement different actions and sounds according to the type of the divided voice signals.
AI speaker that can move using sound source location technology.
The method according to claim 1 or 2,
The artificial intelligence speaker 1 further includes a rotatable arm,
The motor consists of a first motor and a second motor,
The first motor rotates the artificial intelligence speaker 1 about the vertical axis,
The second motor to implement the movement of the arm
AI speaker that can move using sound source location technology.
In the control method of the artificial intelligence speaker 1 including the voice input unit 110, the sound source tracking unit 120, the motor driver 130 and the control unit 140,
Receiving an externally generated voice signal (S1);
Tracking (S2) a sound source of the received voice signal;
Rotating the artificial intelligence speaker (1) to the position of the tracked sound source (S3); And
Determining a threshold of a loudness of the voice to distinguish the user's voice from other voices, and extracting coordinates of the sound source of the voice signal exceeding the threshold to determine whether the voice is the voice of the talker (S4).
A method for controlling an artificial intelligence speaker using sound source position recognition technology.
The method of claim 5,
The step (S4) of confirming whether the voice of the talker is voice may further include storing sound coordinates equal to or greater than a threshold value to reflect the most frequent azimuth and altitude values when the AI speaker 1 implements the movement. doing
A method for controlling an artificial intelligence speaker using sound source position recognition technology.
The method of claim 5,
Further comprising a motion implementing step (S5) to distinguish the received voice signal so that the artificial intelligence speaker 1 can selectively implement different actions and sounds according to the type of the divided voice signal.
A method for controlling an artificial intelligence speaker using sound source position recognition technology.

Images