KR20220045741A - Apparatus, method and computer program for providing voice recognition service - Google Patents

Abstract

A device that provides a speech recognition service may comprise: a receiving part that receives a first speech command word from a user after a call word is recognized; a determining part that determines whether or not a second speech command word satisfies a continuous situation recognition condition, when the second speech command word is received, after the speech recognition service corresponding to the first speech command word is provided; and a providing part that provides the speech recognition service corresponding to the second speech command word based on a determination result.

Description

Apparatus, method, and computer program for providing voice recognition service by judging continuous situations

The present invention relates to an apparatus, method and computer program for providing a voice recognition service by determining a continuous situation.

A general voice recognition device (eg, an artificial intelligence speaker, etc.) may start a voice recognition service based on a voice wake-up method. That is, the voice recognition apparatus does not perform voice recognition on the user's utterance before the utterance command corresponding to the call word is input, and starts voice recognition when the utterance command corresponding to the call word is input.

In this regard, the utterance of the call word required for activation of the voice recognition apparatus is not required only for the first time. Accordingly, when the same user requests an additional voice recognition service at a time interval, it is inconvenient to have to utter an utterance command corresponding to the calling word each time prior to requesting the additional voice recognition service.

Referring to FIG. 1 , after the user is provided with a voice recognition service (service to operate the fan) corresponding to a voice command including 'turn on the fan' from the activated voice recognition device, 'make the fan wind strength lower' includes In order to receive the voice recognition service corresponding to the voice command (a service for setting the wind strength of the fan to be high), the utterance command corresponding to the calling word must be uttered again.

As such, in order to receive a voice recognition service through the existing voice recognition device, it was necessary to first utter a call word each time to switch the voice recognition device to an active state.

Korean Patent Publication No. 2019-0096856 (published on August 20, 2019)

An object of the present invention is to solve the problems of the prior art, and to provide a voice recognition service that recognizes sequentially uttered voice commands even when a user utters only the first call word by recognizing continuous situations through various conditions.

However, the technical problems to be achieved by the present embodiment are not limited to the technical problems described above, and other technical problems may exist.

As a technical means for achieving the above technical problem, an apparatus for providing a voice recognition service according to a first aspect of the present invention includes: a receiver configured to receive a first voice command from a user after a call word is recognized; a determination unit configured to determine whether the second voice command satisfies a continuous context recognition condition when a second voice command is received after the voice recognition service corresponding to the first voice command is provided; and a providing unit that provides a voice recognition service corresponding to the second voice command based on the determination result.

A method of providing a voice recognition service performed by an apparatus for providing a voice recognition service according to a second aspect of the present invention comprises: receiving a first voice command from a user after a call word is recognized; receiving a second voice command from the user after a voice recognition service corresponding to the first voice command is provided; determining whether the second voice command satisfies a continuous context recognition condition; and providing a voice recognition service corresponding to the second voice command based on the determination result.

When a computer program stored in a computer readable recording medium including a sequence of instructions for providing a voice recognition service according to the third aspect of the present invention is executed by a computing device, after a call word is recognized, a first voice command is issued from the user After receiving and providing a voice recognition service corresponding to the first voice command, when a second voice command is received, it is determined whether the second voice command satisfies a continuous context recognition condition, and based on the determination result It may include a sequence of commands for providing a voice recognition service corresponding to the second voice command.

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

According to any one of the above-described problem solving means of the present invention, if the second voice command is continuously received after the voice recognition service corresponding to the first voice command is provided, whether the second voice command satisfies the continuous context recognition condition It is possible to provide a voice recognition service corresponding to the second voice command based on whether or not there is.

Through this, in the present invention, when the second voice command satisfies the continuous context recognition condition, the utterance of the call word may be omitted and a voice recognition service corresponding to the second voice command may be provided.

In addition, the present invention can provide a voice recognition service for a plurality of voice commands belonging to a continuous situation even if the user does not repeatedly utter a call word.

1 is a diagram for explaining a conventional method of providing a voice recognition service.
2 is a block diagram of a system for providing a voice recognition service according to an embodiment of the present invention.
3 is a block diagram of an apparatus for providing a voice recognition service shown in FIG. 1 according to an embodiment of the present invention.
4A to 4B are diagrams for explaining a method of mapping utterance intentions for each control category, according to an embodiment of the present invention.
5 is a diagram illustrating a dialogue analysis result for a voice command according to an embodiment of the present invention.
6 is a diagram illustrating a table including prior state information of an external environment according to an embodiment of the present invention.
7A to 7B are diagrams for explaining a method of updating a utterance intention for each control category according to a user's feedback information, according to an embodiment of the present invention.
8 is a flowchart illustrating a method of providing a voice recognition service according to an embodiment of the present invention.
9 is a flowchart illustrating a method of providing a voice recognition service by determining a continuous situation according to an embodiment of the present invention.
10 is a flowchart illustrating a method of providing a voice recognition service by determining a continuous situation according to another embodiment of the present invention.
11A to 11B are diagrams illustrating scenarios occurring when a voice recognition service omitting a call word utterance in a continuous situation is applied, according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS 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 them. However, the present invention may be embodied in several different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part is "connected" with another part, this includes not only the case of being "directly connected" but also the case of being "electrically connected" with another element interposed therebetween. . Also, when a part "includes" a certain component, it means that other components may be further included, rather than excluding other components, unless otherwise stated.

In this specification, a "part" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. In addition, one unit may be implemented using two or more hardware, and two or more units may be implemented by one hardware.

Some of the operations or functions described as being performed by the terminal or device in this specification may be instead performed by a server connected to the terminal or device. Similarly, some of the operations or functions described as being performed by the server may also be performed in a terminal or device connected to the server.

Hereinafter, detailed contents for carrying out the present invention will be described with reference to the accompanying configuration diagram or process flow diagram.

2 is a block diagram of a system for providing a voice recognition service according to an embodiment of the present invention.

Referring to FIG. 2 , the voice recognition service providing system may include a voice recognition service providing apparatus 100 , a voice recognition server 110 , a voice analysis server 120 , and a voice synthesis server 130 . However, since the voice recognition service providing system of FIG. 1 is only an embodiment of the present invention, the present invention is not limitedly interpreted through FIG. 1 , and may be configured differently from FIG. 1 according to various embodiments of the present invention. there is.

The apparatus 100 for providing a voice recognition service may provide artificial intelligence-based functions such as voice recognition, voice search, voice translation, and voice assistant through communication with a user. For example, the apparatus 100 for providing a voice recognition service may be an AI speaker that provides the aforementioned artificial intelligence-based function, but is not limited thereto. That is, the apparatus 100 for providing a voice recognition service may be any terminal capable of providing the aforementioned artificial intelligence-based function.

When the voice recognition service providing apparatus 100 receives a call word for activation of the voice recognition service providing apparatus 100 from the user, it may prepare for the voice recognition service.

After recognizing the call word received from the user, the apparatus 100 for providing a voice recognition service may receive a first voice command from the user. Here, the call word means a preset word (eg, Genie) for the apparatus 100 for providing a voice recognition service to initiate reception of a voice command.

The voice recognition service providing apparatus 100 may transmit sound data for the received first voice command to the voice recognition server 110 .

The voice recognition server 110 may convert the sound received from the voice recognition service providing apparatus 100 into text, and transmit the converted text to the voice recognition service providing apparatus 100 .

The voice recognition service providing apparatus 100 may transmit the text received from the voice recognition server 110 to the voice analysis server 120 .

The voice analysis server 120 may analyze the corresponding text and transmit the analysis result to the voice recognition service providing apparatus 100 . The voice analysis server 120 may convert the corresponding text into a preset format and transmit information in the converted format to the service providing apparatus 100 . For example, if the text corresponding to the first voice command is 'Tell me the weather today', the voice analysis server 120 transmits information 'TYPE=weather, INTENT=question' to the voice recognition service providing apparatus 100 can

The voice recognition service providing apparatus 100 may provide a voice recognition service corresponding to the first voice command based on the analysis result by the voice analysis server 120 . For example, when it is not necessary to output a voice with respect to the first voice command (eg, changing a TV channel, outputting specific information to the TV), the apparatus 100 for providing a voice recognition service provides a specific information corresponding to the first voice command. function can be performed.

As another example, when it is necessary to output a voice with respect to the first voice command (eg, providing weather information, etc.), the apparatus 100 for providing a voice recognition service is configured to provide an external server based on the analysis result of the first voice command. It is possible to request response information (eg, weather information) corresponding to the first voice command from (not shown), and receive response information corresponding to the first voice command from an external server (not shown).

Thereafter, the voice recognition service providing apparatus 100 transmits response information corresponding to the first voice command received from an external server (not shown) to the voice synthesis server 130 , and the corresponding response from the voice recognition server 130 . A sound corresponding to the information may be received.

The voice recognition service providing apparatus 100 may provide a voice recognition service corresponding to the first voice command by outputting the sound received from the voice synthesis server 130 through a speaker.

The voice recognition service providing apparatus 100 may provide a voice recognition service corresponding to the first voice command and sequentially receive the second voice command from the user.

The apparatus 100 for providing a voice recognition service may determine whether the second voice command satisfies a continuous context recognition condition. Here, the continuous context recognition condition may include a condition in which the second voice command is received within a preset time. That is, the apparatus 100 for providing a voice recognition service continues to perform voice recognition even if a call word is not recognized for a preset time after providing a specific voice recognition service.

For example, when the second voice command is received within a preset time, the apparatus 100 for providing a voice recognition service may determine the second voice command as a continuous situation associated with the first voice command.

The voice recognition service providing apparatus 100 may provide a voice recognition service corresponding to the second voice command based on a result of determining whether the second voice command satisfies the continuous context recognition condition.

The apparatus 100 for providing a voice recognition service may provide a voice recognition service for the second voice command in such a way that the voice recognition service for the first voice command is provided.

Hereinafter, the operation of each component of the voice recognition service providing system of FIG. 1 will be described in more detail.

3 is a block diagram of the apparatus 100 for providing a voice recognition service shown in FIG. 2 according to an embodiment of the present invention.

Referring to FIG. 3 , the apparatus 100 for providing a voice recognition service may include a receiving unit 300 , a determining unit 310 , a providing unit 320 , an update unit 330 , and a storage unit 340 . However, the apparatus 100 for providing a voice recognition service illustrated in FIG. 3 is only one implementation example of the present invention, and various modifications are possible based on the components illustrated in FIG. 3 .

Hereinafter, FIG. 3 will be described with reference to FIGS. 4A to 7B and FIGS. 11A to 11B .

The dialog model building unit (not shown) may build different dialog models for each voice service (dialog models in which utterance intentions classified for each voice service are learned) for dialog analysis of voice commands received from the user. Here, the reason for constructing different dialog models for each service is that the utterance intention may be different for each service even though the utterance is the same. For example, a voice command of 'tell me' and a voice command of 'show me' may be identified as the same utterance intention or different utterance intentions depending on the service. For example, referring to FIG. 4A , the dialog model building unit (not shown) may build a dialog model for each voice service based on information on the utterance intention for each voice service input through the dialog model management page 40 . there is. Here, the information on the utterance intention includes selection information (eg, 'order') of the control category corresponding to the voice service to be set, the utterance intention name to be mapped to the selected control category (eg, 'pizza order'), and the utterance intention. It may include a corresponding dialogue phrase (eg, 'make a cheese pizza').

The conversation analysis unit (not shown) may analyze the user's voice command by using the established conversation model for each voice service. For example, the conversation analysis unit (not shown) may derive what the user's utterance intention for the voice command is through the conversation model for each voice service. Here, the analysis result analyzed through the dialogue analyzer (not shown) may include, for example, a utterance intention corresponding to a voice command, emotional information of a user who uttered the voice command, accuracy information on the analysis result, and the like. . For example, referring to FIG. 5 , the conversation analysis unit (not shown) derives the utterance intention 50 of 'selectLevel' as a result of analyzing the voice command of 'Choose me as an intermediate animal' through a conversation model for each voice service. can do.

The storage 340 may map and store a plurality of utterance intentions for each of a plurality of control categories.

For example, referring to FIG. 4B , the storage unit 340 stores a control category table 42 in which a plurality of utterance intentions are mapped for each of a plurality of control categories in order to determine a continuation situation with respect to the continuously received user's voice commands. ) can be stored. For example, the 'TurnOnAC' utterance intention corresponding to the voice command of 'Turn on the air conditioner' and the 'UpACTemp' utterance intention corresponding to the voice command of 'Raise the air conditioner temperature' may be mapped to the IoT control category. The 'TurnOnTV' utterance intention corresponding to the voice command of 'Turn on TV' and the 'UpTVCH' utterance intent corresponding to the voice command of 'Turn up TV channel' may be mapped to the TV control category.

Here, as a condition for determining the continuity situation for each control category, the time (that is, a preset time) after the voice recognition service for the preceding voice command is performed until the subsequent voice command is received (that is, a preset time) is the control category table (42) can be stored. For example, the preset time may be set differently for each control category. For example, the preset time may be a fixed value automatically set by the apparatus 100 for providing a voice recognition service, or may be a variable value input from a user.

The control category table 42 including the utterance intentions for each of the plurality of control categories may be used to determine whether consecutive voice commands belong to a continuous situation. For example, if the user consecutively utters the voice command of 'Turn on TV' after the voice service corresponding to the voice command of 'Turn on the air conditioner' is provided, the utterance intention corresponding to the voice command of 'Turn on the air conditioner' and 'Turn on the TV' Since the utterance intention corresponding to the voice command of ' belongs to different control categories, it may be determined that the situation is not continuous. If, after the voice recognition service corresponding to the voice command of 'Turn on the air conditioner' is provided, if the user consecutively utters the voice command of 'Raise the air conditioner temperature', the utterance intention corresponding to the voice command of 'Turn on the air conditioner' and the Since the utterance intention corresponding to the voice command of 'Upgrade' belongs to the same control category, it may be determined as a continuous situation.

The storage unit 340 may further map and store pre-state information of an external environment preset for each utterance intention in the control category table. Alternatively, the storage unit 340 may map and store pre-state information of a preset external environment for each of a plurality of utterance intentions.

For example, referring to FIG. 6 , the storage unit 340 is an external device to which pre-state information of an external environment preset for each of a plurality of utterance intentions is mapped in order to determine a continuation situation with respect to the continuously received user's voice commands. The environment dictionary state table 60 may be stored. For example, the 'UpACTemp' utterance intention corresponding to the voice command of 'Turn up the air conditioner temperature' and the 'TurnOffAC' utterance intention corresponding to the voice command of 'Turn off the air conditioner' are mapped to the pre-state information of the external environment of 'air conditioner on'. can be

In addition, the 'UpTVCH' utterance intention corresponding to the voice command of 'Turn up TV channel' and the 'TurnOffTV' utterance intent corresponding to the voice command of 'Turn off TV' may be mapped to the prior state information of the external environment of 'TV on'. there is.

The external environment dictionary state table 60 including preset external environment prior state information mapped for each utterance intention may be used to determine whether successive voice commands belong to a continuous context.

For example, if the user utters the voice command of 'Raise the air conditioner temperature' after the air conditioner is turned on according to the voice command of 'Turn on the air conditioner', the state of the external environment at the time the voice command of 'Raise the air conditioner temperature' is ignited Since the air conditioner is on, it is judged as a continuous situation because it satisfies the pre-state information of the external environment mapped to the utterance intention ('UpACTemp') corresponding to the voice command of 'Raise the air conditioner temperature', and the voice of 'Raise the air conditioner temperature' A voice recognition service for commands may be provided.

According to the present embodiment, the voice recognition service providing apparatus 100 may start the voice recognition service without a call word when a voice command related to a utterance intention corresponding to an external environment activated according to the prior state information of the external environment is received. can

The receiver 300 may receive a first voice command from the user after a call word for activating the voice recognition service providing apparatus 100 is recognized.

The receiver 300 may receive the second voice command from the user after the voice recognition service corresponding to the first voice command is provided.

The determination unit 310 may determine whether the second voice command satisfies a continuous context recognition condition. Here, the continuous context recognition condition may include a condition in which the second voice command is received within a preset time, as described above. According to the present embodiment, the apparatus 100 for providing a voice recognition service may start a voice recognition service without a call word for a preset time after the voice recognition service for a preceding voice command is performed.

In addition, the continuous context recognition information includes the second voice command within a preset time set in the control category to which the utterance intention of the first voice command belongs, and the utterance intention of the first voice command belongs to the continuous context recognition information. It may include a condition that the command is received.

When the second voice command is received within a preset time, the determination unit 310 may determine the second voice command as a continuous situation associated with the first voice command.

The determination unit 310 includes the second voice command within a preset time set in the control category to which the utterance intention of the first voice command belongs, and the utterance intention of the second voice command is included in the control category to which the utterance intention of the first voice command belongs. If it is determined that the second voice command has been received, it may be determined as a continuous situation associated with the first voice command.

For example, referring to FIG. 4B , when the second voice command of 'Raise the air conditioner temperature' is continuously received from the user after the voice service corresponding to the first voice command of 'Turn on the air conditioner' is provided (that is, the voice service is After being provided, the second voice command is received within a preset time), the determination unit 310 'Raise the air conditioner temperature' to the IoT control category to which the ignition intention ('TurnOnAC') corresponding to the first voice command of 'Turn on the air conditioner' belongs Since the utterance intention ('UpACTemp') corresponding to the second voice command of ' belongs to, the second voice command of 'Raise the air conditioner temperature' can be determined as a continuous situation associated with the first voice command of 'Turn on the air conditioner'.

In addition, the determination unit 310 is a second voice command of 'Raise the air conditioner temperature' belonging to the IoT control category to which the utterance intention ('TurnOnAC') corresponding to the first voice command of 'Turn on the air conditioner' is a preset first time If received within (eg, 5 seconds), it may be determined that the second voice command of 'raise the air conditioner temperature' is a continuous situation associated with the first voice command of 'turn on the air conditioner'.

When the utterance intention of the second voice command is not included in the control category to which the utterance intention of the first voice command belongs, the determination unit 310 may determine the second voice command as a voice command unrelated to the first voice command. there is.

For example, referring to FIG. 4B , when the second voice command of 'Turn on TV' is continuously received from the user after the voice service corresponding to the first voice command of 'Turn on the air conditioner' is provided, the determination unit 310 is Since the utterance intent ('TurnOnTV') corresponding to the second voice command of 'Turn on TV' does not belong to the IoT control category to which the utterance intent ('TurnOnAC') corresponding to the first voice command of 'Turn on the air conditioner' belongs, 'Turn on TV' The second voice command of ' may be determined as a voice command that is not continuous with the first voice command of 'Turn on the air conditioner'.

The determination unit 310 includes a preset time set in the control category to which the utterance intention of the second voice command belongs, but the second voice command is included in the control category to which the utterance intention of the first voice command belongs. If it is not received within, it may be determined that the second voice command is a voice command that is not related to the first voice command.

For example, referring to FIG. 4B , a utterance intention corresponding to the second voice command of 'Raise the air conditioner temperature' in the IoT control category to which the utterance intention ('TurnOnAC') corresponding to the first voice command of 'Turn on the air conditioner' belongs ('UpACTemp') belongs to, but the second voice command of 'Raise the air conditioner temperature' is not received within the first preset time (eg, 5 seconds) set in the IoT control category, the determination unit 310 'Air conditioner The second voice command of 'Raise the temperature' may not be determined as a continuous situation associated with the first voice command of 'Turn on the air conditioner'.

The determination unit 310 is configured to combine the second voice command with the first voice command when there is a prior state of the external environment related to the utterance intention for the second voice command among the pre-state information of the external environment preset for each utterance intention of the user. It can be judged as a series of related situations.

For example, referring to FIG. 6 , if the first voice command of 'turn on the air conditioner' is received and the air conditioner is turned on, the determination unit 310 recognizes 'air conditioner on' as pre-state information of the external environment. , it is possible to recognize a voice command corresponding to the utterance intention corresponding to the prior state information of the external environment of 'air conditioner on' without a call.

For example, after the air conditioner is turned on according to the first voice command of 'Turn on the air conditioner', if a second voice command of 'Raise the air conditioner temperature' is received from the user, the determination unit 310 determines that the second voice command is 'Air conditioner Since it is one of the utterance intentions corresponding to the prior state information of the external environment of 'on', it can be determined that the second voice command of 'raise the air conditioner temperature' is a continuous situation associated with the first voice command of 'turn on the air conditioner'.

According to the present embodiment, the voice recognition service providing apparatus 100 may start the voice recognition service without a call word when a voice command included in an utterance intention corresponding thereto is received in a state in which prior state information of a specific external environment is activated. can In this case, there may be no particular limitation on the reception time of the subsequent voice command.

In an embodiment, the determination unit 310 compares the similarity between the first sentence in which the first voice command is converted into text and the second sentence in which the second voice command is converted into text (eg, cosine similarity, jacquard similarity, similarity). Through the Clidian similarity, Manhattan similarity, etc.), it may be determined whether the second voice command belongs to a continuous situation associated with the first voice command. For example, when the similarity value between the first sentence corresponding to the first voice command and the second sentence corresponding to the second voice command exceeds a preset threshold, the determination unit 310 may transmit the second voice command to the first It can be determined by the continuous situation related to the voice command.

In another embodiment, when the user's second voice command is received after the voice recognition service corresponding to the user's first voice command is provided, the determination unit 310 is recognized through a customized personal conversation model only for the user The final continuous situation obtained by summing the first continuous situation determination value for the user's second voice command and the second continuous situation determination value for the user's second voice command recognized through the entire conversation model for an unspecified number of users Based on the determination value, it is possible to determine the second voice command as a continuous situation associated with the first voice command. For example, the final continuous situation determination value may be calculated through [Equation 1].

[Equation 1]

,

,

는 최종 연속 상황 판단값이고,

는 전체 대화 모델의 가중치이고,

는 전체 대화 모델에 의한 연속 상황 판단값이고,

는 맞춤형 개인 대화 모델의 가중치이고,

는 맞춤형 개인 모델에 의한 연속 상황 판단값이다. here,

is the final continuous situation judgment value,

is the weight of the entire dialog model,

is a continuous situation judgment value by the entire dialogue model,

is the weight of the customized personal conversation model,

is a continuous situation judgment value by the customized personal model.

In another embodiment, the output unit (not shown) determines that the second voice command additionally received from the user is a continuation situation of the first voice command after the voice recognition service corresponding to the user's first voice command is provided A specific LED color may be output through the LED bulb of the voice recognition service providing apparatus 100 in order to notify the user of the current state. Through this, the user can confirm that the voice recognition service providing apparatus 100 recognizes a subsequent voice command associated with a preceding voice command within a continuous context.

In addition, the output unit (not shown) calls the user when a preset time (time set to determine the continuous situation) has elapsed after the voice recognition service is provided by recognizing a subsequent voice command related to a preceding voice command in a continuous situation. An effect sound (eg, a beep sound) indicating that only words can be recognized may be output through the speaker of the apparatus 100 for providing a voice recognition service.

The providing unit 320 may provide a voice recognition service corresponding to the second voice command based on the determination result.

When the second voice command satisfies the continuous context recognition condition, the providing unit 320 may provide a voice recognition service corresponding to the second voice command.

The providing unit 320 includes a second voice command within a preset time set in the control category to which the utterance intention of the second voice command belongs, and the utterance intention of the first voice command belongs to the control category. When is received, a voice recognition service corresponding to the first voice command and the second voice command determined as a continuation situation may be provided.

The providing unit 320 determines that the situation is continuous with the first voice command when there is a prior state of the external environment related to the utterance intention for the second voice command among the pre-state information of the external environment preset for each utterance intention of the user. A voice recognition service corresponding to the second voice command may be provided.

For example, referring to FIG. 11A , after a voice recognition service corresponding to the first voice command of 'turn on the fan' (a function of driving the state of the fan into an ON operation) is provided, the user receives the In addition, when the second voice command of 'lower the wind strength of the fan' is received, the providing unit 320 determines the second voice command as a continuous situation related to the first voice command and responds to the second voice command without a call word A voice recognition service (a function to lower the wind strength of the fan) may be provided.

When the second voice command does not satisfy the continuous context recognition condition, the providing unit 320 ignores the second voice command and does not provide a voice recognition service corresponding to the second voice command. For example, referring to FIG. 11B , after a voice recognition service corresponding to the first voice command of 'turn on the fan' (a function of driving the state of the fan to an ON operation) is provided, the above conditions are not satisfied. When the second voice command of 'hello' is received from the user in the state, the providing unit 320 determines that the second voice command of 'hello' is not a continuous situation related to the first voice command and provides the second voice command can be ignored

The receiver 300 may receive feedback information from the user after the voice recognition service corresponding to the second voice command is provided. At this time, the second voice command analyzed as a continuous situation by the user's feedback information or the second voice commands analyzed as not in the continuous situation are separately managed, and only the second voice commands that are managed separately are preceded by the first voice A similarity value with the command can be derived. The derived similarity value may be used to update the user's utterance intention mapped to a plurality of control categories.

For example, assuming that the voice recognition service for the first voice command of 'Turn on the air conditioner' is provided and then the voice recognition service for the second voice command of 'Turn up the air conditioner temperature' uttered by the user is provided, the user When is uttered to the voice recognition service providing apparatus 100 feedback information confirming that the second voice command is not the utterance of the first voice command and the continuous situation (eg, 'Ginny, I didn't say that to you') , the receiving unit 300 may receive the corresponding feedback information.

Also, when it is determined that the second voice command is not in a continuous situation and thus a voice recognition service corresponding to the second voice command is not provided, the receiver 300 may receive feedback information from the user. For example, after the voice recognition service for the first voice command of 'Turn on the air conditioner' is provided, the second voice command of 'Turn on TV' uttered by the user is not recognized by the voice recognition service providing apparatus 100 In this case, the receiving unit 300 may receive feedback information including 'Ginny said to you' from the user.

The updater 330 may update the user's utterance intention mapped to a plurality of control categories based on the user's feedback information.

For example, when the analysis result of the user's feedback information received from the user in a situation where the voice recognition service for the second voice command is not provided is analyzed that the second voice command is a continuous situation associated with the first voice command , the updater 330 may update the control category table 42 by merging the control category to which the utterance intention of the first voice command belongs with the control category to which the utterance intention of the second voice command belongs. Alternatively, when the analysis result of the user's feedback information is analyzed that the second voice command is not a continuous situation associated with the first voice command, the updater 330 divides the control category to which the utterance intention of the first voice command belongs. to update the control category table 42 .

The updater 330 may add a user-customized category to the plurality of control categories.

For example, referring to FIGS. 4B and 7A together, after the voice recognition service for the first voice command 701 of 'Turn on the air conditioner' is provided, the second voice of 'Turn up TV channel' uttered by the user additionally When the command 703 is not recognized by the apparatus 100 for providing a voice recognition service (ie, when continuous situation recognition fails), the update unit 330 sets the IoT control category to which the utterance intention of the first voice command 701 belongs. The control category table 42 may be updated by adding a user-customized category that merges the TV control category to which the utterance intention of the second voice command 703 and the second voice command 703 belongs to the control category table 42 .

For example, referring to FIGS. 4B and 7B together, after the voice recognition service for the first voice command 701 of 'Turn on the air conditioner' is provided, the second voice of 'Raise the air conditioner temperature' further uttered by the user When it is analyzed that the command 705 is recognized by the voice recognition service providing apparatus 100 and the voice recognition service for the second voice command 705 is provided from the user's feedback information, it is analyzed that the continuous situation is misrecognized, the update unit 330 ) is controlled by adding a user-customized category obtained by dividing the IoT control category to which the utterance intention of the first voice command 701 belongs and the IoT control category to which the utterance intention of the second voice command 703 belongs to the control category table 42 . The category table 42 may be updated.

Meanwhile, for those skilled in the art, each of the receiving unit 300, the determining unit 310, the providing unit 320, the update unit 330, and the storage unit 340 may be implemented separately, or one or more of them may be integrated. You will fully understand that

8 is a flowchart illustrating a method of providing a voice recognition service according to an embodiment of the present invention.

Referring to FIG. 8 , in step S801 , the apparatus 100 for providing a voice recognition service may receive a first voice command from a user after a call word is recognized.

In step S803 , the apparatus 100 for providing a voice recognition service may receive a second voice command from the user after the voice recognition service corresponding to the first voice command is provided.

In step S805, the apparatus 100 for providing a voice recognition service may determine whether the second voice command satisfies a continuous context recognition condition. Here, the continuous context recognition condition may include a condition in which the second voice command is received within a preset time.

In operation S807, the apparatus 100 for providing a voice recognition service may provide a voice recognition service corresponding to the second voice command based on the determination result. For example, when the second voice command is received within a preset time, the apparatus 100 for providing a voice recognition service determines that the second voice command is a continuous situation related to the first voice command, and corresponds to the second voice command. A voice recognition service may be provided. If the second voice command is not received within the preset time, the voice recognition service providing apparatus 100 does not determine the second voice command as a continuous situation related to the first voice command, but does not determine that the second voice command corresponds to the second voice command It does not provide voice recognition service.

In the above description, steps S801 to S807 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 if necessary, and the order between the steps may be changed.

9 is a flowchart illustrating a method of providing a voice recognition service by determining a continuous situation according to an embodiment of the present invention.

Referring to FIG. 9 , in step S901 , the apparatus 100 for providing a voice recognition service may receive a first voice command from a user after a call word is recognized.

In step S903, the voice recognition service providing apparatus 100 may provide a voice recognition service corresponding to the first voice command.

In step S905, the apparatus 100 for providing a voice recognition service may receive a second voice command from the user.

In step S907, the apparatus 100 for providing a voice recognition service determines that the utterance intention of the second voice command is included in the control category belonging to the utterance intention of the first voice command, and is set in the control category to which the utterance intention of the first voice command belongs. When the second voice command is input within the time, it may be determined that the second voice command is a continuous situation associated with the first voice command.

If it is determined in step S909 that the second voice command is a continuous situation associated with the first voice command, the voice recognition service providing apparatus 100 may provide a voice recognition service corresponding to the second voice command.

In step S911 , the apparatus 100 for providing a voice recognition service may receive feedback information about a voice recognition service corresponding to the second voice command from the user.

In operation S913 , the apparatus 100 for providing a voice recognition service may update the user's utterance intention mapped to a plurality of control categories based on the user's feedback information.

In the above description, steps S901 to S913 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 if necessary, and the order between the steps may be changed.

10 is a flowchart illustrating a method of providing a voice recognition service by determining a continuous situation according to another embodiment of the present invention.

Referring to FIG. 10 , in step S1001 , the apparatus 100 for providing a voice recognition service may receive a first voice command from a user after a call word is recognized.

In step S1003, the apparatus 100 for providing a voice recognition service may provide a voice recognition service corresponding to the first voice command.

In step S1005, the apparatus 100 for providing a voice recognition service may receive a second voice command from the user.

In step S1007 , the apparatus 100 for providing a voice recognition service performs the second voice when there is prior state information of the external environment related to the utterance intention for the second voice command among the prior state information of the external environment preset for each user's utterance intention. The command may be determined as a continuous situation associated with the first voice command.

When it is determined that the second voice command is a continuous situation associated with the first voice command in step S1009, the voice recognition service providing apparatus 100 may provide a voice recognition service corresponding to the second voice command.

In operation S1011 , the apparatus 100 for providing a voice recognition service may receive feedback information about a voice recognition service corresponding to the second voice command from the user.

In operation S1013 , the apparatus 100 for providing a voice recognition service may update the user's utterance intention mapped to a plurality of control categories based on the user's feedback information.

In the above description, steps S1001 to S1013 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 if necessary, and the order between the steps may be changed.

An embodiment of the present invention may 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. Also, computer-readable media 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 description of the present invention described above is for illustration, and those of ordinary skill in the art to which the present invention pertains can understand that it can be easily modified into other specific forms 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 restrictive. For example, each component described as a single type may be implemented in a dispersed form, and likewise components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention. .

100: voice recognition service providing device
300: receiver
310: judgment unit
320: providing unit
330: update unit
340: storage

Claims (20)

A device for providing a voice recognition service
a receiver configured to receive a first voice command from a user after the call word is recognized;
a determination unit configured to determine whether the second voice command satisfies a continuous context recognition condition when a second voice command is received after the voice recognition service corresponding to the first voice command is provided; and
and a providing unit providing a voice recognition service corresponding to the second voice command based on a result of the determination.
The method of claim 1,
The server for providing a voice recognition service, further comprising a storage unit for mapping and storing a plurality of utterance intentions for each of a plurality of control categories.
3. The method of claim 2,
The continuous situation recognition condition is
Including a condition that the second voice command is received within a preset time,
The voice recognition service providing server that the preset time is set differently for each of the plurality of control categories.
4. The method of claim 3,
The judging unit
When the second voice command is received within the preset time, the apparatus for providing a voice recognition service is to determine the second voice command as a continuous situation associated with the first voice command.
4. The method of claim 3,
The judging unit
The utterance intention of the second voice command is included in a control category to which the utterance intention of the first voice command belongs, and the second voice command is input within a preset time set in a control category to which the utterance intention of the first voice command belongs. If it is, the second voice command will be determined as a continuous situation associated with the first voice command, the voice recognition service providing apparatus.
3. The method of claim 2,
The apparatus for providing a voice recognition service, wherein the storage unit further includes pre-state information of an external environment preset for each utterance intention.
8. The method of claim 7,
The judging unit
If there is prior state information of the external environment related to the utterance intention for the second voice command among the pre-state information of the external environment preset for each utterance intention, the second voice command is continuously related to the first voice command which is to be determined as, a voice recognition service providing device.
3. The method of claim 2,
the receiving unit
After the voice recognition service corresponding to the second voice command is provided, the apparatus for providing a voice recognition service is to receive feedback information from a user.
9. The method of claim 8,
The apparatus for providing a speech recognition service further comprising an updater configured to update the user's utterance intention mapped to the plurality of control categories based on the user's feedback information.
10. The method of claim 9,
The update unit will add a user-customized category to the plurality of control categories, the voice recognition service providing apparatus.
A method of providing a voice recognition service performed by a voice recognition service providing apparatus,
after the call word is recognized, receiving a first voice command from a user;
receiving a second voice command from the user after a voice recognition service corresponding to the first voice command is provided;
determining whether the second voice command satisfies a continuous context recognition condition; and
and providing a voice recognition service corresponding to the second voice command based on a result of the determination.
12. The method of claim 11,
The method further comprising the step of mapping and storing a plurality of utterance intentions for each of a plurality of control categories.
13. The method of claim 12,
The continuous situation recognition condition is
Including a condition that the second voice command is received within a preset time,
The method for providing a voice recognition service, wherein the preset time is set differently for each of the plurality of control categories.
14. The method of claim 13,
The step of determining whether the second voice command satisfies the continuous situation recognition condition includes:
When the second voice command is received within the preset time, determining the second voice command as a continuous situation associated with the first voice command, the voice recognition service providing method.
14. The method of claim 13,
The step of determining whether the second voice command satisfies the continuous situation recognition condition includes:
The utterance intention of the second voice command is included in a control category to which the utterance intention of the first voice command belongs, and the second voice command is input within a preset time set in a control category to which the utterance intention of the first voice command belongs. If it is, the method of providing a voice recognition service comprising the step of determining the second voice command as a continuous situation associated with the first voice command.
13. The method of claim 12,
The saving step is
The method of providing a voice recognition service comprising the step of further storing pre-state information of an external environment preset for each utterance intention.
17. The method of claim 16,
The step of determining whether the second voice command satisfies the continuous situation recognition condition includes:
If there is prior state information of the external environment related to the utterance intention for the second voice command among the pre-state information of the external environment preset for each utterance intention, the second voice command is continuously related to the first voice command A method of providing a voice recognition service comprising the step of determining as.
13. The method of claim 12,
After the voice recognition service corresponding to the second voice command is provided, the method further comprising the step of receiving feedback information from a user.
19. The method of claim 18,
The method further comprising the step of updating the user's utterance intention mapped to the plurality of control categories based on the user's feedback information.
In a computer program stored on a computer-readable recording medium comprising a sequence of instructions for providing a voice recognition service,
When the computer program is executed by a computing device,
After the call word is recognized, receiving a first voice command from the user,
When a second voice command is received after the voice recognition service corresponding to the first voice command is provided, it is determined whether the second voice command satisfies a continuous context recognition condition;
A computer program stored in a computer-readable recording medium comprising a sequence of instructions for providing a voice recognition service corresponding to the second voice command based on a result of the determination.

Images