KR20210055193A - Apparatus for recognizing voice and Method for detecting events by using the same - Google Patents

Abstract

The present invention relates to a voice recognition an apparatus and a method for detecting an event using the same. The voice recognition device according to an embodiment of the present invention comprises: a sound wave output unit for outputting a sound wave signal by superimposing an audible signal in an audible frequency band generated according to a user's request and an inaudible signal in an inaudible frequency band generated for detecting an event in the monitoring space; a sound wave receiving unit for collecting a noise signal generated in the monitoring space, including the reflected signal reflected in the monitoring space; and a control unit for extracting a target signal corresponding to the inaudible signal from the noise signal and determining whether an event has occurred in the monitoring space using the target signal. Accordingly, the present invention may detect a temperature change in the monitoring space.

Description

Voice recognition device and event detection method using the same {Apparatus for recognizing voice and Method for detecting events by using the same}

The present application relates to a voice recognition device and an event detection method using the same, and to a voice recognition device capable of simultaneously performing a sound source providing function requested by a user and an event monitoring function in a monitoring space, and an event detection method using the same.

Commonly used security systems include an IR sensor (infrared sensor), an image sensor, and a heat sensor. However, in the case of an IR sensor, it is possible to become useless due to exposure of the boundary area as it cannot detect an intruder when intruding wearing an IR shielding suit. In the case of an image sensor, misinformation due to light and shadow frequently occurs, and separate at night. Lighting equipment is required, and the heat ray sensor reacts sensitively to changes in the ambient temperature and has a high probability of generating false information, so there is a blind spot as a security system.

In addition, in order to build a security system, not only the purchase cost of various sensors mentioned above, but also the cost including time and labor cost for the construction are expended, there is a problem that the general public cannot easily access.

Therefore, when building and using a security system, time and cost are saved, no misinformation occurs, no additional device is required, and the need to develop a security system capable of detecting in any form when an intruder occurs has emerged.

The present application is to provide a voice recognition device capable of simultaneously performing a sound source providing function requested by a user and an event monitoring in a monitoring space, and an event detection method using the same.

The present application is to provide a voice recognition device capable of detecting an event such as a user activity or temperature change in a monitoring space using sound waves reflected in a monitoring space, and an event detection method using the same.

A voice recognition device according to an embodiment of the present invention relates to a voice recognition device located in a monitoring space, and an audible signal of an audible frequency band generated according to a user's request, and an audible signal generated for detecting an event in the monitoring space. A sound wave output unit for outputting a sound wave signal by superimposing the inaudible signal of the inaudible frequency band; A sound wave receiver configured to collect a noise signal generated in the monitoring space, including a reflected signal from which the sound wave signal is reflected in the monitoring space; And a control unit that extracts a target signal corresponding to the inaudible signal from the noise signal, and determines whether an event occurs in the monitoring space by using the target signal.

Here, the control unit may determine a user activity or temperature change in the monitoring space as the occurrence of the event.

Here, the control unit extracts a target signal corresponding to the inaudible signal by applying a frequency-time conversion and a high pass filter to the noise signal, and determines whether an event occurs in the monitoring space using the target signal. It may include an event detection module to determine.

Here, the control unit performs pre-processing to classify the voice into the noise signal, extracts a voice signal by applying the frequency-time conversion and a low pass filter to the pre-processed signal, and then recognizes the voice signal for the voice signal. It may further include a voice recognition module to proceed.

Here, the event detection module, when the size of the target signal is greater than or equal to the reference value, determines that the activity of the user in the monitoring space has occurred, and compares the size of the target signal with a distance reference table to calculate the distance to the user. have.

Here, the sound wave receiver further includes at least two or more microphones for collecting the noise signal, and the event detection module compares the magnitudes of target signals measured by the plurality of microphones to determine the direction of the user. And determine the location of the user using the distance and direction to the user.

Here, the event detection module may determine whether a temperature change occurs in the monitoring space by comparing the frequency shift of the target signal and the inaudible signal.

Here, the sound wave output unit may form and output the shape of the inaudible signal in a frequency domain in a sawtooth shape.

An event detection method according to an embodiment of the present invention relates to an event detection method of a voice recognition device, and an audible signal in an audible frequency band generated according to a user's request and a rain generated to detect an event in the monitoring space are Outputting a sound wave signal by superimposing inaudible signals in a blue frequency band; Collecting a noise signal generated in the monitoring space, including a reflected signal from which the sound wave signal is reflected in the monitoring space; And extracting a target signal corresponding to the inaudible signal from the noise signal, and determining whether an event occurs in the monitoring space using the target signal.

In addition, the solution to the above-described problem does not enumerate all the features of the present invention. Various features of the present invention and advantages and effects thereof may be understood in more detail with reference to the following specific embodiments.

According to the voice recognition device and the event detection method using the same according to an embodiment of the present invention, it is possible to simultaneously perform a sound source providing function requested by a user and an event monitoring function in a monitoring space because a signal in a non-audible band is used. In addition, an event such as a user activity or temperature change in the monitoring space may be detected by using the sound wave signal reflected in the monitoring space.

According to the voice recognition device and the event detection method using the same according to an embodiment of the present invention, it is possible to accurately determine the location of a user in a monitoring space including a plurality of microphones, and to clearly recognize a frequency shift according to a temperature change. Therefore, it is possible to accurately determine the temperature change in the monitoring space.

1 is a schematic diagram showing the detection of a user activity or temperature change in a monitoring space using a voice recognition device according to an embodiment of the present invention.
2 is a block diagram showing a speech recognition apparatus according to an embodiment of the present invention.
3 is a block diagram showing a sound wave output unit of a speech recognition apparatus according to an embodiment of the present invention.
4 is a block diagram showing a control unit of a voice recognition apparatus according to an embodiment of the present invention.
5 is a schematic diagram showing a non-audible signal for sensing temperature of a voice recognition device according to an embodiment of the present invention.
6 and 7 are flow charts illustrating an event detection method of a voice recognition device according to an embodiment of the present invention.

Hereinafter, preferred embodiments will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art may easily implement the present invention. However, in describing a preferred embodiment of the present invention in detail, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. In addition, the same reference numerals are used throughout the drawings for parts having similar functions and functions.

In addition, throughout the specification, when a part is said to be'connected' with another part, it is not only'directly connected', but also'indirectly connected' with another element in the middle. Includes. In addition, "including" a certain component means that other components may be further included rather than excluding other components unless otherwise stated. In addition, terms such as "~ unit" and "module" described in the specification mean a unit that processes at least one function or operation, which may be implemented as hardware or software, or as a combination of hardware and software.

1 is a schematic diagram showing detection of a user activity or temperature change in a monitoring space using a voice recognition device according to an embodiment of the present invention.

As shown in Fig. 1, the voice recognition device 100 may be located in a monitoring space (s) such as a living room, and output various sound sources such as news, music, weather information, audiobook, etc. by recognizing a user's voice command. , Voice guidance for the user's question can be performed.

Here, the voice recognition apparatus 100 may additionally output a sound wave signal in the monitoring space s to detect an event occurring in the monitoring space s. That is, the sound wave signal output from the voice recognition device 100 may be reflected by a structure in the monitoring space (s), and the reflected signal may generate events such as user activity, fire, or temperature change due to the operation of the air conditioner. Can be affected by Accordingly, it is possible to detect the occurrence of an event in the monitoring space (s) by comparing the reflected signal at the time of the occurrence of the event with the reflected signal at an ordinary time.

Depending on the embodiment, the voice recognition device 100 may perform the functions of event detection and sound source output at the same time, but in this case, the sound wave signal generated by the voice recognition device 100 for event detection may be a problem. have. That is, when the sound wave signal is included in the audible frequency band, the user can listen to the sound wave signal together with the sound source, which may be perceived as noise by the user, causing discomfort.

In order to solve this problem, the speech recognition apparatus 100 according to an embodiment of the present invention may utilize a signal in a non-audible band. That is, in the case of a signal in the inaudible band, since users cannot recognize it, users can use the sound source provided by the voice recognition device 100 without inconvenience, and at the same time, it is necessary to detect events in the monitoring space (s). It is possible.

Hereinafter, a speech recognition apparatus according to an embodiment of the present invention will be described.

2 is a block diagram showing a speech recognition apparatus according to an embodiment of the present invention. Referring to FIG. 2, the speech recognition apparatus 100 according to an embodiment of the present invention may include a sound wave output unit 110, a sound wave reception unit 120, and a control unit 130.

The sound wave output unit 110 may generate an audible signal in an audible frequency band generated according to a user's request, and a non-audible signal in an audible frequency band generated to detect an event in the monitoring space (s). Inaudible signals can be superimposed and output as sound wave signals. Here, the sound wave output unit 110 may include an audible signal generation module 111, an inaudible signal generation module 112, a mixer 113, and a speaker 114, as shown in FIG. 3.

The audible signal generation module 111 generates an audible signal according to a user's request, wherein the audible signal may be various sound sources requested by the user such as news, music, weather information, and audiobooks. Depending on the embodiment, a sound source provided by a streaming method through a wired or wireless Internet connected to the voice recognition apparatus 100 may be generated as an audible signal.

The inaudible signal generation module 112 may generate an inaudible signal of an inaudible frequency band that is output to detect an event in the monitoring space (s). Here, the inaudible signal is a signal in an inaudible frequency band and cannot be recognized by the user, but may be processed in various ways to detect an event.

The mixer 113 may generate a sound wave signal by superimposing each audible signal and an inaudible signal, and the speaker 114 may output the sound wave signal in the monitoring space s. In this case, a sound wave signal may be applied in the monitoring space s, and a reflected signal reflected in the monitoring space s may be generated.

The sound wave receiver 120 may collect a noise signal in the monitoring space (s). That is, sound generated by the actions or actions of users located in the monitoring space (s) can be collected through the sound wave receiving unit 120, and the noise signal includes a reflected signal reflected in the monitoring space (s). I can.

Depending on the embodiment, the sound wave receiving unit 120 may include a plurality of microphones and collect noise signals with each microphone. In addition, as shown in Fig. 2(b), it is possible to arrange the microphones to be located in all four positions on the top, bottom, left and right. In addition, the number of microphones (m) and installation locations can be selected in various ways.

The controller 130 may extract a target signal corresponding to the inaudible signal from the noise signal. Thereafter, it is possible to determine whether an event occurs in the monitoring space (s) by using the extracted target signal. Specifically, the controller 130 may include an event detection module 131 and a voice recognition module 132 as shown in FIG. 4.

The event detection module 131 can extract a target signal corresponding to an inaudible signal by applying a frequency-time conversion and a high pass filter to the collected noise signal, and generates an event in the monitoring space using the target signal. You can determine whether or not. Here, a Fast Fourier Transform (FFT) may be used as a frequency-time transformation, and a noise signal collected by the sound wave receiver 120 may be converted into a discrete signal by using the FFT. Thereafter, a high-pass filter may be applied to extract a target signal corresponding to an inaudible signal from among the converted discrete signals. That is, since the inaudible band corresponds to a relatively high frequency region, the target signal can be extracted through the high pass filter. Thereafter, it is possible to determine whether an event in the monitoring space s occurs from the target signal.

Specifically, when the magnitude of the target signal is greater than or equal to the reference value, the event detection module 131 may first determine that the user's activity in the monitoring space s has occurred. Here, the reference value may be preset as a fixed value, but may vary depending on the location where the voice recognition apparatus 100 is installed, depending on the embodiment. That is, since the environment is different for each monitoring space (s), a reference value may be set to correspond to each environment.

Thereafter, the event detection module 131 may calculate a distance from the voice recognition apparatus 100 to the user by comparing the size of the target signal with the distance reference table. Here, the distance reference table is generated in advance for each monitoring space (s), and can be generated by experimentally obtaining a target signal size when a user exists at a location spaced apart from the voice recognition device 100 by a predetermined interval. That is, by repeatedly measuring while adjusting the distance between the voice recognition device 100 and the user, the size of the target signal for each distance in the monitoring space (s) can be derived, and the result can be stored in the distance reference table. .

For example, the voice recognition apparatus 100 may guide a user to repeat movement in a certain distance unit (eg, 1m unit) within the monitoring space (s). Here, the voice recognition apparatus 100 may collect target signals according to the user's movement, and among the target signals collected through the user's repetitive movement at the same location, valid data within 95% of the normal distribution are collected. It can be discriminated and classified and saved. Thereafter, when the valid data for the corresponding location is successfully stored, the voice recognition device 100 can guide the movement to repeat while increasing the distance, and store the valid data for each distance while increasing the distance to create a distance reference table. Configurable. Accordingly, the event detection module 131 can accurately determine the distance between the voice recognition apparatus 100 and the user by comparing the target signal and the distance reference table. Additionally, when a plurality of microphones are included, the distance may be measured by comparing the average value of target signals collected from each microphone with a distance reference table.

In addition, the sound wave receiving unit 120 may include a plurality of microphones, and as shown in FIG. 2(b), each of the microphones may point in different directions. Accordingly, the event detection module 131 may determine a user's direction with respect to the voice recognition apparatus 100 from the magnitude of each target signal collected by a plurality of microphones. For example, when microphones are provided on the front, left, and right sides of the voice recognition apparatus 100, respectively, the size of a target signal measured by each microphone may be different according to a user's location. That is, when the user is located on the left, the size of the target signal may be relatively large on the left and the front side, and the size of the target signal measured on the right side may be relatively small. On the other hand, when the user is located on the right, the size of the target signal may be relatively large in the right and the front, and the size of the target signal measured in the left may be relatively small. Therefore, it is possible to specify the direction in which the user is located by utilizing this.

Furthermore, the event detection module 131 may specify the location of the user from the distance and direction of the user extracted from the target signal. Specifically, when it is determined that the user's activity occurs, the event detection module 131 may first align the target signals received from the plurality of microphones according to the size to match the size order and the arrangement direction of each microphone. That is, the direction to which the microphones having a large target signal are directed can be specified as the user's direction. In addition, after obtaining the average of the magnitude values of each target signal, the distance to the user may be calculated by comparing the result with a distance reference table. Thereafter, the location of the user based on the voice recognition apparatus 100 may be specified using the distance and direction to the user.

On the other hand, the event detection module 131 may also detect a temperature change such as a fire in the monitoring space (s). Here, the event detection module 131 may detect a temperature change in the monitoring space using a change in the propagation speed of the target signal according to the temperature. That is, when the temperature in the monitoring space s increases, the propagation speed of the sound wave increases, so that the target signal moves in the high-frequency direction without changing the shape. Accordingly, the event detection module 131 may measure the degree of frequency shift according to the temperature change, and then determine the relative temperature change through the measurement. For example, when the target signal is received, the event detection module 131 can compare the frequency with the inaudible signal, and when a frequency shift of the received target frequency occurs, whether or not the temperature change is based on the frequency change amount of the target signal. Can be discriminated.

However, in the case of using a flat non-audible signal having a constant size for each frequency, there is a problem in that it is difficult to recognize a frequency transition due to temperature change. That is, when a flat type of inaudible signal is used, since the target signal is also displayed in a flat shape having a constant size for each frequency, it is difficult to recognize whether a frequency shift occurs with respect to the target signal. Accordingly, here, as shown in FIG. 5, by using a sawtooth-shaped non-audible signal having a different size for each frequency, it can be implemented so that a frequency shift according to a temperature change can be easily recognized.

That is, as shown in Fig. 5(a), the inaudible signal can be formed in a sawtooth shape in the frequency domain, and if the temperature in the monitoring space (s) increases afterwards, as shown in Fig. 5(b) The frequency can be shifted in the high frequency direction. At this time, since the inaudible signal has a sawtooth shape, the frequency shift can be easily recognized, and the amount of change due to the frequency shift can also be clearly identified.

On the other hand, as shown in Fig. 5(c), when the temperature in the monitoring space (s) decreases, it can be confirmed that the inaudible signal transitions in the low frequency direction, through which the air conditioner in the monitoring space (s) Motion or inflow of outside air can be detected.

On the other hand, the speech recognition module 132 performs pre-processing to classify speech into a noise signal, and extracts a speech signal by applying an FFT and a low-pass filter to the preprocessed signal, and then performs speech recognition on the extracted speech signal. I can. That is, through the voice recognition module 132, the voice recognition apparatus 100 may recognize a voice command or the like that is applied by a user, and perform a control operation accordingly. Here, speech recognition can be implemented through various algorithms.

6 and 7 are flowcharts illustrating an event detection method of a voice recognition device according to an embodiment of the present invention.

Referring to FIG. 6, the voice recognition device outputs a sound wave signal by superimposing an audible signal in an audible frequency band generated according to a user's request and an inaudible signal in an inaudible frequency band generated to detect an event in the monitoring space. Can be (S10). Here, the audible signal may be a variety of sound sources requested by the user, such as news, music, weather information, and audio books, and may be sound sources provided in a streaming method through wired or wireless Internet connected to a voice recognition device according to embodiments. . The inaudible signal is a signal in the inaudible frequency band and cannot be recognized by the user, but can be processed in various ways to detect an event.

Thereafter, a noise signal generated in the monitoring space may be collected, and at this time, a reflected signal from which the sound wave signal is reflected in the monitoring space may also be included in the noise signal (S20). Depending on the embodiment, the speech recognition device may include a plurality of microphones, and it is also possible to collect noise signals with each microphone. At this time, the number and installation positions of the microphones provided in the voice recognition device can be selected in various ways, and according to an embodiment, the number of microphones provided in the voice recognition device may be arranged so as to be positioned in all four up, down, left, and right sides.

Upon receiving the noise signal, the speech recognition apparatus may extract a target signal corresponding to the inaudible signal from the noise signal, and determine whether an event occurs in the monitoring space using the target signal (S30).

Specifically, referring to FIG. 7, the speech recognition apparatus can extract a target signal (S31), and can check whether the extracted target signal is greater than or equal to a reference value (S32). Here, when the target signal is less than the reference value, since user activity is not detected, the target signal corresponding to the inaudible signal output in the next cycle may be extracted.

On the other hand, when the target signal is greater than or equal to the reference value, the direction and distance of the user can be determined, respectively. That is, the voice recognition apparatus may compare the size of the target signal with the distance reference table (S33), and calculate the distance from the voice recognition apparatus to the user (S34). Here, the distance reference table is generated in advance for each monitoring space, and may be generated by experimentally obtaining a target signal size when a user exists at a location spaced apart from the voice recognition device. Accordingly, by comparing the target signal and the distance reference table, it is possible to determine the distance between the voice recognition device and the user.

In addition, when a plurality of microphones are included in the voice recognition device, each microphone can be directed in a different direction, and the voice recognition device is a voice recognition device based on the size of each target signal collected from the plurality of microphones. You can determine the user's direction for. That is, it is possible to compare the magnitudes of the target signals measured by the plurality of microphones (S35), and the direction to which the microphones having the large target signal are directed among the plurality of microphones may be determined as the user's direction (S36). Accordingly, the voice recognition device can specify the location of the user based on the voice recognition device by collecting the distance and direction to the user.

Meanwhile, the voice recognition device can also detect temperature changes in the monitoring space. That is, since the propagation speed of the target signal changes according to the temperature, it is possible to detect the temperature change in the monitoring space by utilizing this. Specifically, after measuring the degree of frequency shift of the target signal according to the temperature change, the relative temperature change may be determined through the measurement. Depending on the embodiment, it is possible to use this to determine whether a fire occurs in the monitoring space or whether the air conditioner is operating.

However, in the case of using a flat non-audible signal having a constant size for each frequency, there is a problem in that it is difficult to recognize a frequency transition due to temperature change. Accordingly, in this case, the speech recognition device may detect whether a temperature change occurs by forming an inaudible signal in a sawtooth shape in the frequency domain. In this case, when the frequency of the inaudible signal is shifted, it can be easily recognized, and the amount of change due to the frequency shift can be clearly confirmed.

The present invention described above can be implemented as a computer-readable code on a medium on which a program is recorded. The computer-readable medium may be one that continuously stores a program executable by a computer, or temporarily stores a program for execution or download. In addition, the medium may be a variety of recording means or storage means in a form in which a single piece of hardware or several pieces of hardware are combined. The medium is not limited to a medium directly connected to a computer system, but may be distributed on a network. Examples of media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, And there may be ones configured to store program instructions, including ROM, RAM, flash memory, and the like. In addition, examples of other media include an app store that distributes applications, a site that supplies or distributes various software, and a recording medium or a storage medium managed by a server. Therefore, the detailed description above should not be construed as restrictive in all respects and should be considered as illustrative. The scope of the present invention should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of the present invention are included in the scope of the present invention.

The present invention is not limited by the above-described embodiments and the accompanying drawings. It will be apparent to those of ordinary skill in the art to which the present invention pertains, that components according to the present invention can be substituted, modified, and changed within the scope of the technical spirit of the present invention.

100: voice recognition device 110: sound wave output unit
111: audible signal generation module 112: inaudible signal generation module
113: mixer 114: speaker
120: sound wave receiver 130: control unit
131: event detection module 132: voice recognition module

Claims (9)

In the voice recognition device located in the monitoring space,
A sound wave output unit for outputting a sound wave signal by superimposing an audible signal in an audible frequency band generated according to a user's request and an inaudible signal in an audible frequency band generated to detect an event in the monitoring space;
A sound wave receiver configured to collect a noise signal generated in the monitoring space, including a reflected signal from which the sound wave signal is reflected in the monitoring space; And
And a control unit for extracting a target signal corresponding to the inaudible signal from the noise signal, and determining whether an event occurs in the monitoring space using the target signal.
The method of claim 1, wherein the control unit
Voice recognition device, characterized in that to determine the occurrence of the event the user activity or temperature change in the monitoring space.
The method of claim 1, wherein the control unit
An event detection module that extracts a target signal corresponding to the inaudible signal by applying a frequency-time conversion and a high pass filter to the noise signal, and determines whether an event occurs in the monitoring space using the target signal Voice recognition device comprising a.
The method of claim 3, wherein the control unit
A voice recognition module further comprises a voice recognition module for performing pre-processing for classifying voice on the noise signal, extracting a voice signal by applying the frequency-time conversion and low-pass filter to the pre-processed signal, and then performing voice recognition for the voice signal. Voice recognition device, characterized in that.
The method of claim 3, wherein the event detection module
If the size of the target signal is greater than or equal to a reference value, it is determined that a user activity in the monitoring space has occurred, and a distance to the user is calculated by comparing the size of the target signal with a distance reference table.
The method of claim 3,
The sound wave receiver
Further comprising at least two or more microphones (microphone) for collecting the noise signal,
The event detection module
Comparing the magnitude of each target signal measured by the plurality of microphones to determine the direction of the user,
The voice recognition device, characterized in that to determine the location of the user by using the distance and direction to the user.
The method of claim 3, wherein the event detection module
And comparing the frequency shift of the target signal and the inaudible signal to determine whether or not a temperature change in the monitoring space has occurred.
The method of claim 7, wherein the sound wave output unit
A speech recognition device, characterized in that the shape of the inaudible signal in the frequency domain is formed in a sawtooth shape and output.
In the event detection method of a voice recognition device,
Outputting a sound wave signal by superimposing an audible signal in an audible frequency band generated according to a user's request and an inaudible signal in an audible frequency band generated to detect an event in the monitoring space;
Collecting a noise signal generated in the monitoring space, including a reflected signal from which the sound wave signal is reflected in the monitoring space; And
Extracting a target signal corresponding to the inaudible signal from the noise signal, and determining whether an event occurs in the monitoring space using the target signal.

Images