KR20200141126A - Device and method for preventing misperception of wake word - Google Patents

Abstract

A voice control device for preventing misrecognition for a wake-up word may comprise: a reception unit receiving a broadcast stream from a broadcast server; a reproducing unit reproducing sound data of the received broadcast stream inside the voice control device; an output unit outputting sound data of the broadcast stream through an external device connected to the voice control device; an input unit receiving the sound data outputted through the external device through a microphone of the voice control device; a wake-up word detection unit detecting a first wake-up word from the sound data inputted through the microphone, and detecting a second wake-up word from the sound data reproduced inside the voice control device; and a determination determining whether a wake-up word is misrecognized based on a difference between a first recognition time for the first wake-up word and a second recognition time for the second wake-up word.

Description

Device and method for preventing misrecognition of call words {DEVICE AND METHOD FOR PREVENTING MISPERCEPTION OF WAKE WORD}

The present invention relates to an apparatus and method for preventing misrecognition of a pager.

In recent years, as the demand for artificial intelligence services increases, the importance of call word recognition technology is increasing. The caller corresponds to a command that activates an artificial intelligence device.

In an artificial intelligence device, a pager recognizer capable of recognizing a pager is built into the device in order to secure security for a user interface and sound information using voice.

On the other hand, there is a problem of misrecognition due to echo as a factor causing a problem in the pager recognizer. The problem that the artificial intelligence device's self-reproducing sound is reproduced through the built-in speaker of the artificial intelligence device or externally connected speaker and re-input to the microphone built into the artificial intelligence device is called echo. Acoustic Echo Canceller (AEC) is used to cancel this echo. However, even if the echo is removed using AEC, a residual echo of a certain level or more remains.

When a pager exists in a sound source (eg, a broadcast sound source) played inside the artificial intelligence device, the pager recognizer enters a command recognition state by residual echo for the pager. In this case, the artificial intelligence device outputs an unnecessary voice recognition preparation notification message to a user who is listening to music or watching a video, and stops playing the content according to the misrecognition of the caller.

It is impossible to recognize in advance that a caller and a similar caller are included in the internally played sound with the existing caller recognition technology, so if a caller is included in a broadcast sound source or randomly played sound source, the artificial intelligence device does not want to There may be problems with responding.

Korean Patent Publication No. 2018-0107003 (published on October 1, 2018)

The present invention is to solve the problems of the prior art described above, through a recognition time for a first pager detected from sound data of a broadcast stream played inside a voice control device, and an external device connected to the voice control device. An attempt is made to determine whether or not the caller is misrecognized by comparing the recognition time for the second caller detected from sound data of the broadcast stream input through the microphone of the voice control device. However, the technical problem to be achieved by the present embodiment is not limited to the technical problems as described above, and other technical problems may exist.

As a technical means for achieving the above technical problem, a voice control apparatus for preventing misrecognition of a pager according to a first aspect of the present invention includes: a receiver for receiving a broadcast stream from a broadcast server; A reproducing unit for reproducing sound data of the received broadcast stream in the voice control device; An output unit outputting sound data of the broadcast stream through an external device connected to the voice control device; An input unit receiving sound data output through the external device through a microphone of the voice control device; A pager detection unit detecting a first pager from the sound data received through the microphone, and detecting a second pager from the sound data reproduced inside the voice control device; And a determination unit that determines whether or not a caller is recognized incorrectly based on a difference between a first recognition time for the first pager and a second recognition time for the second pager.

A method for preventing misrecognition of a page word in a voice control device according to a second aspect of the present invention includes: receiving a broadcast stream from a broadcast server; Reproducing sound data of the received broadcast stream inside the voice control device; Outputting sound data of the broadcast stream through an external device connected to the voice control device; Receiving sound data output through the external device through a microphone of the voice control device; Detecting a first pager from sound data received through the microphone; Detecting a second call word from sound data reproduced inside the voice control device; And determining whether or not a pager is recognized incorrectly based on a difference between a first recognition time for the first pager and a second recognition time for the second pager.

The above-described problem solving means are merely exemplary and should not be construed as limiting the present invention. In addition to the above-described exemplary embodiments, there may be additional embodiments described in the drawings and detailed description of the invention.

According to any one of the above-described problem solving means of the present invention, the present invention provides a recognition time for the first call word detected from sound data of a broadcast stream played inside the voice control device, and an external device connected to the voice control device. By comparing the recognition time for the second call word detected from the sound data of the broadcast stream input through the microphone of the voice control device through the comparison, it is possible to determine whether the call word is misrecognized.

Through this, the present invention minimizes an erroneous speech recognition rate, and in the case of a call word for a self-reproducing sound, a more stable speech recognition environment can be established by removing a response to the call word. For example, the present invention can eliminate a phenomenon such as stopping playback of a broadcast stream due to a pager and a similar pager included in the playback sound while the voice control device plays the broadcast stream.

1 is a block diagram of a voice control device according to an embodiment of the present invention.
2 is a diagram for explaining a problem of misrecognition of a call word.
3 is a flowchart illustrating a method of preventing misrecognition of a pager in a voice control device according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art can easily implement the present invention. However, the present invention may be implemented in various different forms and is not limited to the embodiments described herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and similar reference numerals are assigned to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only "directly connected" but also "electrically connected" with another element interposed therebetween. . In addition, when a part "includes" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

In the present specification, the term "unit" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Further, one unit may be realized using two or more hardware, or two or more units may be realized using one hardware.

In the present specification, some of the operations or functions described as being performed by the terminal or device may be performed instead by a server connected to the terminal or device. Likewise, some of the operations or functions described as being performed by the server may also be performed by a terminal or device connected to the server.

Hereinafter, with reference to the accompanying configuration diagram or processing flow chart, specific details for the implementation of the present invention will be described.

When a broadcast sound source containing a call word is reproduced inside a conventional artificial intelligence device, a phenomenon as shown in FIG. 2 occurs. Referring to FIG. 2, assuming that the call word for activating the artificial intelligence device is "Jiniya", the name of the person appearing in the broadcast sound source 200 is "Jin-hee" or "Jin-ya" 210 is included, or , When a person's conversation includes "it's my family," or the like, the sound reproduced by the broadcast sound source 200 may be reproduced similarly to the sound of the caller of the artificial intelligence device.

In preparation for such a case, in the conventional artificial intelligence device, the echo for the call word included in the reproduced sound of the broadcast sound source 200 is removed by using Acoustic Echo Cancellation (AEC) ("Jin Hee-ya" 220 attenuated after AEC). ). However, even if the echo is removed by using AEC, a certain level of residual echo is present, and a false recognition is generated in a pager recognizer that uses a microphone of an artificial intelligence device as an input.

The existing various attempts to modify the AEC and the pager recognizer to improve the performance of false caller detection by such echo lead to a decrease in the caller recognition rate, and false detection attenuation using voice processing correction reduces the desired recognition rate and false detection. Can't achieve In addition, when sound is decoded by an external player (eg, a TV or an AV receiver), such audio processing is very difficult or impossible.

1 is a block diagram of a voice control apparatus 100 according to an embodiment of the present invention.

Referring to FIG. 1, the voice control device 100 includes a receiving unit 110, a playback unit 120, an output unit 130, an input unit 140, an audio channel conversion unit 150, a pager detection unit 160, A derivation unit 170, a determination unit 180, and an adjustment unit 190 may be included. However, the voice control apparatus 100 shown in FIG. 1 is only an example of implementation of the present invention, and various modifications are possible based on the components shown in FIG. 1.

The receiver 110 may receive a broadcast stream from a broadcast server.

The playback unit 120 may reproduce sound data of the received broadcast stream inside the voice control device 100.

The output unit 130 may output sound data of a broadcast stream through an external device (eg, a TV or an AV receiver) connected to the voice control device 100.

The input unit 140 may receive sound data output through an external device through a microphone of the voice control device 100.

The sound channel converter 150 may divide and decode sound data of a broadcast stream into multiple channels, and convert the decoded multiple channels into one sound channel.

For example, the sound channel conversion unit 150 may decode sound data of a broadcast stream output through a speaker of an external device such as a TV or an AV receiver.

For example, the sound channel conversion unit 150 extracts sound data from a broadcast stream (audio/video stream) output from the voice control device 100, and separates the extracted sound data into multiple channels according to an audio format. After that, it can be decoded.

Since it is difficult to predict which channel among the channels of the decoded sound data has a reproducible sound similar to the call word of the voice control device 100, the sound channel conversion unit 150 down-mixes the decoded multiple channels. It can be converted to one sound channel. In this case, it is possible to prevent the occurrence of similar call words by combining each phoneme existing in the multiple channels into one channel.

The pager detection unit 160 may detect the first pager from sound data input through a microphone. Specifically, when sound data of a broadcast stream is output through an external device, and the output sound data is received by the microphone of the voice control device 100, the caller detection unit 160 first receives the sound data input through the microphone. The caller can be detected.

The pager detection unit 160 may detect the second pager from sound data reproduced in the voice control device 100. At this time, the time when sound data is reproduced and recognized inside the voice control device 100 precedes the time when sound data is input and recognized through the microphone of the voice control device 100.

In this case, the second call word may be detected in two ways according to the required calculation amount in the call word recognizer of the voice control device 100 and the extra calculation amount of the voice control device 100 itself.

When the amount of extra computation of the voice control device 100 itself is not larger than the required computation amount of the call word recognizer, the call word detection unit 160 uses the sound reproduced inside the voice control device 100 through one down-mixed sound channel. The second pager can be detected from the data.

When the amount of extra computation of the voice control device 100 itself exceeds the required computational amount of the caller recognizer, and multiple operations are possible, the caller detection unit 160 uses decoded multiple channels and one downmixed sound channel. The second call word may be detected from the sound data reproduced inside the voice control device 100.

The derivation unit 170 derives a path delay time based on the time when the sound data is output through the external device and the time when the sound data output through the external device is input through the microphone, and the sound data output through the external device The processing delay time for can be derived. Here, the path delay time may be calculated as a difference (r1time-p1time) between a time (r1time) when sound data output through an external device is received through a microphone and a time (p1time) when sound data is output through an external device. . The processing delay time is a time at which sound data of a broadcast stream is divided into multiple channels and decoded inside the voice control device 100 (delay_d) and a time at which the decoded multiple channels are downmixed by the voice control device 100 (delay_m). It can be calculated as the sum (delay_d + delay_m).

In this case, the processing delay time should be smaller than the path delay time as shown in [Equation 1].

[Equation 1]

r1time-p1time >= delay_d + delay_m

The adjustment unit 190 may adjust the processing delay time to be smaller than the path delay time through an output delay time for sound data output through an external device. Here, the output delay time may be calculated as a sum (delay_o + delay_o') between a basic output delay time (delay_o) and an additional output delay time (delay_o') that sound data is output from the voice control device 100 to an external device. .

For example, when the processing delay time exceeds the path delay time, the adjustment unit 190 increases the output delay time for the sound data output through the external device to increase the processing delay time for the sound data output through the external device. Can be adjusted.

Conditions for the new path delay time (r1time-p1time + delay_o') and the processing delay time (delay_d + delay_m) based on the additional output delay time (delay_o') that the sound data is output from the voice control device 100 to the external device are It is derived as in [Equation 2], and the minimum additional output delay time (delay_o') can be derived as in [Equation 3].

[Equation 2]

r1time-p1time + delay_o' >= delay_d + delay_m

[Equation 3]

delay_o' = delay_d + delay_m-(r1time-p1time)

In this case, if the minimum additional output delay time (delay_0') is negative, no additional delay is required, but if the minimum additional output delay time is positive, an additional delay equal to the minimum additional output delay time is required.

The determination unit 180 may determine whether or not the caller recognition is misrecognized based on a difference between the first recognition time for the first pager and the second recognition time for the second pager. For example, the second recognition time for the second call word detected from the sound data for each of the multiple channels played inside the voice control apparatus 100 is recorded for each channel by Time2 [channel number], and the microphone is Assuming that the first recognition time for the first pager detected from the sound data received through is recorded as Time2, the determination unit 180 determines the second recognition time and the first pager for each channel. It is possible to determine whether or not a page word recognition is misrecognized by comparing the first recognition time for.

For example, a method of discriminating misrecognition by echo through comparison between a first recognition time for a first pager and a second recognition time for a second pager is as follows.

Time_Def[channel 1] = Time2-Time1[channel 1],

Time_Def[channel 2] = Time2-Time1[channel 2]

...

Time_Def[channel x] = Time2-Time1[channel x]

Result = Delay_thresold> MAX( Time_Def[1: x])

When the difference between the first recognition time and the second recognition time is less than a predefined threshold, the determination unit 180 may determine the recognition of the caller as a false recognition due to an echo. For example, when the maximum value for the difference between the first recognition time and the second recognition time per channel is less than a predefined threshold, the determination unit 180 may determine the recognition of the caller as a misrecognition by echo. have.

Meanwhile, for those skilled in the art, the receiver 110, the reproduction unit 120, the output unit 130, the input unit 140, the sound channel conversion unit 150, the caller detection unit 160, the derivation unit 170, the determination unit It will be fully understood that each of the 180 and the adjustment unit 190 may be implemented separately, or one or more of them may be integrated and implemented.

3 is a flowchart illustrating a method of preventing misrecognition of a pager in the voice control apparatus 100 according to an embodiment of the present invention.

Referring to FIG. 3, in step S301, the voice control device 100 may receive a broadcast stream from a broadcast server.

In step S303, the voice control device 100 may reproduce sound data of the received broadcast stream inside the voice control device 100.

In step S305, the voice control device 100 may output sound data of a broadcast stream through an external device connected to the voice control device 100.

In step S307, the voice control device 100 may receive sound data output through an external device through a microphone of the voice control device 100.

In step S309, the voice control device 100 may detect the first call word from sound data input through the microphone of the voice control device 100.

In step S311, the voice control device 100 may detect a second call word from sound data reproduced in the voice control device 100.

In step S315, the voice control apparatus 100 may determine whether or not to recognize the page word incorrectly based on the difference between the first recognition time for the first page word and the second recognition time for the second page word.

In the above description, steps S301 to S315 may be further divided into additional steps or combined into fewer steps, according to an embodiment of the present invention. In addition, some steps may be omitted as necessary, and the order between steps may be changed.

An embodiment of the present invention may also be implemented in the form of a recording medium including instructions executable by a computer, such as a program module executed by a computer. Computer-readable media can be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Further, the computer-readable medium may include all computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

The above description of the present invention is for illustrative purposes only, and those of ordinary skill in the art to which the present invention pertains will be able to understand that other specific forms can be easily modified without changing the technical spirit or essential features of the present invention will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. .

100: voice control device
110: receiver
120: reproduction unit
130: output
140: input unit
150: sound channel conversion unit
160: pager detection unit
170: lead-out unit
180: judgment unit
190: adjustment unit

Claims (7)

In the voice control device for preventing misrecognition of a pager,
A receiver configured to receive a broadcast stream from a broadcast server;
A reproducing unit for reproducing sound data of the received broadcast stream in the voice control device;
An output unit outputting sound data of the broadcast stream through an external device connected to the voice control device;
An input unit receiving sound data output through the external device through a microphone of the voice control device;
A pager detection unit detecting a first pager from the sound data received through the microphone, and detecting a second pager from the sound data reproduced inside the voice control device; And
A determination unit that determines whether or not a pager is misrecognized based on a difference between a first recognition time for the first pager and a second recognition time for the second pager
Including a voice control device.
The method of claim 1,
An acoustic channel conversion unit that divides and decodes the audio data of the broadcast stream into multiple channels, and converts the decoded multiple channels into one acoustic channel
The voice control device further comprising a.
The method of claim 2,
The sound channel converter converts the decoded multi-channels into one sound channel by down-mixing,
The call word detection unit detects the second call word from sound data reproduced inside the voice control device through the one sound channel.
The method of claim 1,
A path delay time is derived based on a time when the sound data is output through the external device and a time when the sound data output through the external device is input through the microphone, and based on the sound data output through the external device Derivation unit for deriving the processing delay time for
The voice control device further comprising a.
The method of claim 4,
The voice control device further comprising an adjustment unit for adjusting the processing delay time to be smaller than the path delay time through an output delay time for the sound data output through the external device.
The method of claim 1,
The determination unit, when the difference between the first recognition time and the second recognition time is less than a predefined threshold value, to determine the page word recognition as a false recognition due to an echo.
In a method for preventing misrecognition of a pager in a voice control device,
Receiving a broadcast stream from a broadcast server;
Reproducing sound data of the received broadcast stream inside the voice control device;
Outputting sound data of the broadcast stream through an external device connected to the voice control device;
Receiving sound data output through the external device through a microphone of the voice control device;
Detecting a first pager from sound data received through the microphone;
Detecting a second call word from sound data reproduced inside the voice control device; And
Determining whether or not a pager is recognized incorrectly based on a difference between a first recognition time for the first pager and a second recognition time for the second pager
That includes, misrecognition prevention method.

Images