KR20240006401A - Method of determining end point detection time and electronic device performing the method - Google Patents

Abstract

Disclosed are a method for determining an end point detection time and an electronic device for performing the same. According to an embodiment of the present invention, the electronic device includes a processor and a memory electrically connected to the processor and storing a command executed by the processor. The processor recognizes speech when the command is executed, determines an end point detection time for finishing the recognition of the speech based on speech data stored in the memory, and finishes the speech recognition in case additional user speech is not inputted within the end point detection time, wherein the speech data can include sentence data including a pattern about a sentence type of the speech and word data including a word of the speech.

Description

Method for determining end point detection time and electronic device for performing the method {METHOD OF DETERMINING END POINT DETECTION TIME AND ELECTRONIC DEVICE PERFORMING THE METHOD}

The following disclosure relates to a method for determining an endpoint detection time based on a user's individual speech history and an electronic device that performs the method.

A variety of electronic devices equipped with a voice recognition function that provides services based on user utterances are becoming popular. In voice recognition, the operation of checking whether the user's speech has ended can have a significant impact on voice recognition performance.

To determine whether the user's speech has ended, voice activity detection (VAD) using voice signals may be used.

An electronic device according to an embodiment may include a processor and a memory that is electrically connected to the processor and stores instructions executed by the processor. The processor may recognize utterance when the instruction is executed. The processor may determine an end point detection time for ending recognition of the utterance, based on utterance data stored in the memory. The processor may end speech recognition when no additional user speech is input within the endpoint detection time. The utterance data may include sentence data including a pattern related to the sentence type of the utterance and word data including words of the utterance.

An electronic device according to an embodiment may include a processor and a memory that is electrically connected to the processor and stores instructions executed by the processor. The processor may determine the category of the utterance when the instruction is executed. The processor may compare the pattern of the utterance with a pattern stored in the utterance data, based on the utterance data corresponding to the category. If the pattern of the utterance is different from the pattern stored in the utterance data, the processor may determine whether the utterance is a complete sentence, an incomplete sentence, or an ambiguous sentence. The processor determines an end point detection time (end) to end recognition of the utterance, based on at least one of a comparison result between the utterance and the pattern stored in the utterance data and whether the utterance is a complete sentence, an incomplete sentence, or an ambiguous sentence. point detection time) can be determined. The processor may end speech recognition when no additional user speech is input within the endpoint detection time. The utterance data may include sentence data including the pattern related to the sentence type of the utterance and word data including words of the utterance.

A method for determining an end point detection time according to an embodiment includes the operation of recognizing an utterance, the operation of determining an end point detection time for ending recognition of the utterance based on utterance data stored in a memory, and the end point detection time. and terminating speech recognition when no additional user utterance is input within the detection time, wherein the utterance data includes sentence data including a pattern regarding a sentence type of the utterance and word data including words of the utterance. may include.

1 is a block diagram of an electronic device in a network environment, according to various embodiments.
Figure 2 is a block diagram showing an integrated intelligence system according to an embodiment.
FIG. 3 is a diagram illustrating an operation of an electronic device determining an endpoint detection time, according to an embodiment.
Figure 4 is a flowchart of an operation method for an electronic device to determine an endpoint detection time, according to an embodiment.
Figure 5 is a flowchart of an operation method for an electronic device to determine an endpoint detection time, according to an embodiment.
Figure 6 is a diagram showing speech data according to one embodiment.
Figure 7 is a diagram illustrating a user interface for registering an utterance in utterance data according to an embodiment.
Figure 8 is a diagram showing categories of utterances according to an embodiment.
Figure 9 is a diagram showing the category proportions of utterances according to an embodiment.
Figure 10 is a diagram showing sentence data according to one embodiment.
Figure 11 is a diagram showing word data according to an embodiment.
Figure 12 is a diagram showing the endpoint detection time determined for each utterance by an electronic device according to an embodiment.
FIG. 13 is a diagram illustrating an operation of an electronic device determining an endpoint detection time according to an EPD command, according to an embodiment.

Hereinafter, embodiments will be described in detail with reference to the attached drawings. In the description with reference to the accompanying drawings, identical components will be assigned the same reference numerals regardless of the reference numerals, and overlapping descriptions thereof will be omitted.

1 is a block diagram of an electronic device 101 in a network environment 100, according to various embodiments. Referring to FIG. 1, in the network environment 100, the electronic device 101 communicates with the electronic device 102 through a first network 198 (e.g., a short-range wireless communication network) or a second network 199. It is possible to communicate with at least one of the electronic device 104 or the server 108 through (e.g., a long-distance wireless communication network). According to one embodiment, the electronic device 101 may communicate with the electronic device 104 through the server 108. According to one embodiment, the electronic device 101 includes a processor 120, a memory 130, an input module 150, an audio output module 155, a display module 160, an audio module 170, and a sensor module ( 176), interface 177, connection terminal 178, haptic module 179, camera module 180, power management module 188, battery 189, communication module 190, subscriber identification module 196 , or may include an antenna module 197. In some embodiments, at least one of these components (eg, the connection terminal 178) may be omitted or one or more other components may be added to the electronic device 101. In some embodiments, some of these components (e.g., sensor module 176, camera module 180, or antenna module 197) are integrated into one component (e.g., display module 160). It can be.

The processor 120, for example, executes software (e.g., program 140) to operate at least one other component (e.g., hardware or software component) of the electronic device 101 connected to the processor 120. It can be controlled and various data processing or calculations can be performed. According to one embodiment, as at least part of data processing or computation, the processor 120 stores commands or data received from another component (e.g., sensor module 176 or communication module 190) in volatile memory 132. The commands or data stored in the volatile memory 132 can be processed, and the resulting data can be stored in the non-volatile memory 134. According to one embodiment, the processor 120 includes a main processor 121 (e.g., a central processing unit or an application processor) or an auxiliary processor 123 that can operate independently or together (e.g., a graphics processing unit, a neural network processing unit ( It may include a neural processing unit (NPU), an image signal processor, a sensor hub processor, or a communication processor). For example, if the electronic device 101 includes a main processor 121 and a secondary processor 123, the secondary processor 123 may be set to use lower power than the main processor 121 or be specialized for a designated function. You can. The auxiliary processor 123 may be implemented separately from the main processor 121 or as part of it.

The auxiliary processor 123 may, for example, act on behalf of the main processor 121 while the main processor 121 is in an inactive (e.g., sleep) state, or while the main processor 121 is in an active (e.g., application execution) state. ), together with the main processor 121, at least one of the components of the electronic device 101 (e.g., the display module 160, the sensor module 176, or the communication module 190) At least some of the functions or states related to can be controlled. According to one embodiment, co-processor 123 (e.g., image signal processor or communication processor) may be implemented as part of another functionally related component (e.g., camera module 180 or communication module 190). there is. According to one embodiment, the auxiliary processor 123 (eg, neural network processing unit) may include a hardware structure specialized for processing artificial intelligence models. Artificial intelligence models can be created through machine learning. For example, such learning may be performed in the electronic device 101 itself on which the artificial intelligence model is performed, or may be performed through a separate server (e.g., server 108). Learning algorithms may include, for example, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but It is not limited. An artificial intelligence model may include multiple artificial neural network layers. Artificial neural networks include deep neural network (DNN), convolutional neural network (CNN), recurrent neural network (RNN), restricted boltzmann machine (RBM), belief deep network (DBN), bidirectional recurrent deep neural network (BRDNN), It may be one of deep Q-networks or a combination of two or more of the above, but is not limited to the examples described above. In addition to hardware structures, artificial intelligence models may additionally or alternatively include software structures.

The memory 130 may store various data used by at least one component (eg, the processor 120 or the sensor module 176) of the electronic device 101. Data may include, for example, input data or output data for software (e.g., program 140) and instructions related thereto. Memory 130 may include volatile memory 132 or non-volatile memory 134.

The program 140 may be stored as software in the memory 130 and may include, for example, an operating system 142, middleware 144, or application 146.

The input module 150 may receive commands or data to be used in a component of the electronic device 101 (e.g., the processor 120) from outside the electronic device 101 (e.g., a user). The input module 150 may include, for example, a microphone, mouse, keyboard, keys (eg, buttons), or digital pen (eg, stylus pen).

The sound output module 155 may output sound signals to the outside of the electronic device 101. The sound output module 155 may include, for example, a speaker or a receiver. Speakers can be used for general purposes such as multimedia playback or recording playback. The receiver can be used to receive incoming calls. According to one embodiment, the receiver may be implemented separately from the speaker or as part of it.

The display module 160 can visually provide information to the outside of the electronic device 101 (eg, a user). The display module 160 may include, for example, a display, a hologram device, or a projector, and a control circuit for controlling the device. According to one embodiment, the display module 160 may include a touch sensor configured to detect a touch, or a pressure sensor configured to measure the intensity of force generated by the touch.

The audio module 170 can convert sound into an electrical signal or, conversely, convert an electrical signal into sound. According to one embodiment, the audio module 170 acquires sound through the input module 150, the sound output module 155, or an external electronic device (e.g., directly or wirelessly connected to the electronic device 101). Sound may be output through the electronic device 102 (e.g., speaker or headphone).

The sensor module 176 detects the operating state (e.g., power or temperature) of the electronic device 101 or the external environmental state (e.g., user state) and generates an electrical signal or data value corresponding to the detected state. can do. According to one embodiment, the sensor module 176 includes, for example, a gesture sensor, a gyro sensor, an air pressure sensor, a magnetic sensor, an acceleration sensor, a grip sensor, a proximity sensor, a color sensor, an IR (infrared) sensor, a biometric sensor, It may include a temperature sensor, humidity sensor, or light sensor.

The interface 177 may support one or more designated protocols that can be used to connect the electronic device 101 directly or wirelessly with an external electronic device (eg, the electronic device 102). According to one embodiment, the interface 177 may include, for example, a high definition multimedia interface (HDMI), a universal serial bus (USB) interface, an SD card interface, or an audio interface.

The connection terminal 178 may include a connector through which the electronic device 101 can be physically connected to an external electronic device (eg, the electronic device 102). According to one embodiment, the connection terminal 178 may include, for example, an HDMI connector, a USB connector, an SD card connector, or an audio connector (eg, a headphone connector).

The haptic module 179 can convert electrical signals into mechanical stimulation (e.g., vibration or movement) or electrical stimulation that the user can perceive through tactile or kinesthetic senses. According to one embodiment, the haptic module 179 may include, for example, a motor, a piezoelectric element, or an electrical stimulation device.

The camera module 180 can capture still images and moving images. According to one embodiment, the camera module 180 may include one or more lenses, image sensors, image signal processors, or flashes.

The power management module 188 can manage power supplied to the electronic device 101. According to one embodiment, the power management module 188 may be implemented as at least a part of, for example, a power management integrated circuit (PMIC).

The battery 189 may supply power to at least one component of the electronic device 101. According to one embodiment, the battery 189 may include, for example, a non-rechargeable primary battery, a rechargeable secondary battery, or a fuel cell.

Communication module 190 is configured to provide a direct (e.g., wired) communication channel or wireless communication channel between electronic device 101 and an external electronic device (e.g., electronic device 102, electronic device 104, or server 108). It can support establishment and communication through established communication channels. Communication module 190 operates independently of processor 120 (e.g., an application processor) and may include one or more communication processors that support direct (e.g., wired) communication or wireless communication. According to one embodiment, the communication module 190 is a wireless communication module 192 (e.g., a cellular communication module, a short-range wireless communication module, or a global navigation satellite system (GNSS) communication module) or a wired communication module 194 (e.g., : LAN (local area network) communication module, or power line communication module) may be included. Among these communication modules, the corresponding communication module is a first network 198 (e.g., a short-range communication network such as Bluetooth, wireless fidelity (WiFi) direct, or infrared data association (IrDA)) or a second network 199 (e.g., legacy It may communicate with an external electronic device 104 through a telecommunication network such as a cellular network, a 5G network, a next-generation communication network, the Internet, or a computer network (e.g., LAN or WAN). These various types of communication modules may be integrated into one component (e.g., a single chip) or may be implemented as a plurality of separate components (e.g., multiple chips). The wireless communication module 192 uses subscriber information (e.g., International Mobile Subscriber Identifier (IMSI)) stored in the subscriber identification module 196 within a communication network such as the first network 198 or the second network 199. The electronic device 101 can be confirmed or authenticated.

The wireless communication module 192 may support 5G networks after 4G networks and next-generation communication technologies, for example, NR access technology (new radio access technology). NR access technology provides high-speed transmission of high-capacity data (eMBB (enhanced mobile broadband)), minimization of terminal power and access to multiple terminals (mMTC (massive machine type communications)), or high reliability and low latency (URLLC (ultra-reliable and low latency). -latency communications)) can be supported. The wireless communication module 192 may support high frequency bands (eg, mmWave bands), for example, to achieve high data rates. The wireless communication module 192 uses various technologies to secure performance in high frequency bands, for example, beamforming, massive array multiple-input and multiple-output (MIMO), and full-dimensional multiplexing. It can support technologies such as input/output (FD-MIMO: full dimensional MIMO), array antenna, analog beam-forming, or large scale antenna. The wireless communication module 192 may support various requirements specified in the electronic device 101, an external electronic device (e.g., electronic device 104), or a network system (e.g., second network 199). According to one embodiment, the wireless communication module 192 supports Peak data rate (e.g., 20 Gbps or more) for realizing eMBB, loss coverage (e.g., 164 dB or less) for realizing mmTC, or U-plane latency (e.g., 164 dB or less) for realizing URLLC. Example: Downlink (DL) and uplink (UL) each of 0.5 ms or less, or round trip 1 ms or less) can be supported.

The antenna module 197 may transmit or receive signals or power to or from the outside (eg, an external electronic device). According to one embodiment, the antenna module 197 may include an antenna including a radiator made of a conductor or a conductive pattern formed on a substrate (eg, PCB). According to one embodiment, the antenna module 197 may include a plurality of antennas (eg, an array antenna). In this case, at least one antenna suitable for a communication method used in a communication network such as the first network 198 or the second network 199 is connected to the plurality of antennas by, for example, the communication module 190. can be selected. Signals or power may be transmitted or received between the communication module 190 and an external electronic device through the at least one selected antenna. According to some embodiments, in addition to the radiator, other components (eg, radio frequency integrated circuit (RFIC)) may be additionally formed as part of the antenna module 197.

According to various embodiments, the antenna module 197 may form a mmWave antenna module. According to one embodiment, a mmWave antenna module includes: a printed circuit board, an RFIC disposed on or adjacent to a first side (e.g., bottom side) of the printed circuit board and capable of supporting a designated high frequency band (e.g., mmWave band); And a plurality of antennas (e.g., array antennas) disposed on or adjacent to the second side (e.g., top or side) of the printed circuit board and capable of transmitting or receiving signals in the designated high frequency band. can do.

At least some of the components are connected to each other through a communication method between peripheral devices (e.g., bus, general purpose input and output (GPIO), serial peripheral interface (SPI), or mobile industry processor interface (MIPI)) and signal ( (e.g. commands or data) can be exchanged with each other.

According to one embodiment, commands or data may be transmitted or received between the electronic device 101 and the external electronic device 104 through the server 108 connected to the second network 199. Each of the external electronic devices 102 or 104 may be of the same or different type as the electronic device 101. According to one embodiment, all or part of the operations performed in the electronic device 101 may be executed in one or more of the external electronic devices 102, 104, or 108. For example, when the electronic device 101 needs to perform a certain function or service automatically or in response to a request from a user or another device, the electronic device 101 may perform the function or service instead of executing the function or service on its own. Alternatively, or additionally, one or more external electronic devices may be requested to perform at least part of the function or service. One or more external electronic devices that have received the request may execute at least part of the requested function or service, or an additional function or service related to the request, and transmit the result of the execution to the electronic device 101. The electronic device 101 may process the result as is or additionally and provide it as at least part of a response to the request. For this purpose, for example, cloud computing, distributed computing, mobile edge computing (MEC), or client-server computing technology can be used. The electronic device 101 may provide an ultra-low latency service using, for example, distributed computing or mobile edge computing. In another embodiment, the external electronic device 104 may include an Internet of Things (IoT) device. Server 108 may be an intelligent server using machine learning and/or neural networks. According to one embodiment, the external electronic device 104 or server 108 may be included in the second network 199. The electronic device 101 may be applied to intelligent services (e.g., smart home, smart city, smart car, or healthcare) based on 5G communication technology and IoT-related technology.

Figure 2 is a block diagram showing an integrated intelligence system according to an embodiment.

Referring to FIG. 2, the integrated intelligent system 10 of one embodiment may include an electronic device 101, an intelligent server 200, and a service server 300.

The electronic device 101 of one embodiment may be a terminal device (or electronic device) capable of connecting to the Internet, for example, a mobile phone, a smartphone, a personal digital assistant (PDA), a laptop computer, a TV, a white appliance, It could be a wearable device, HMD, or smart speaker.

According to the illustrated embodiment, the electronic device 101 includes an interface 177 (e.g., interface 177 in FIG. 1), an input module 150 (e.g., input module 150 in FIG. 1), and an audio output module. (155) (e.g., audio output module 155 in FIG. 1), display module 160 (e.g., display module 160 in FIG. 1), memory 130 (e.g., memory 130 in FIG. 1) , or may include a processor 120 (eg, processor 120 of FIG. 1). The components listed above may be operatively or electrically connected to each other.

The interface 177 in one embodiment may be configured to connect to an external device to transmit and receive data. The input module 150 in one embodiment may receive sound (eg, a user's speech) and convert it into an electrical signal. The sound output module 155 of one embodiment may output an electrical signal as sound (eg, voice).

The display module 160 in one embodiment may be configured to display images or videos. The display module 160 of one embodiment may also display a graphic user interface (GUI) of an app (or application program) being executed. The display module 160 in one embodiment may receive a touch input through a touch sensor. For example, the display module 160 may receive text input through a touch sensor in the on-screen keyboard area displayed within the display module 160.

The memory 130 in one embodiment may store a client module 131, a software development kit (SDK) 133, and a plurality of apps 135. The client module 131 and SDK 133 may form a framework (or solution program) for performing general functions. Additionally, the client module 131 or SDK 133 may configure a framework for processing user input (eg, voice input, text input, touch input).

The plurality of apps 135 stored in the memory 130 of one embodiment may be programs for performing designated functions. According to one embodiment, the plurality of apps 135 may include a first app 135_1 and a second app 135_2. According to one embodiment, each of the plurality of apps 135 may include a plurality of operations to perform a designated function. For example, the apps may include an alarm app, a messaging app, and/or a schedule app. According to one embodiment, the plurality of apps 135 are executed by the processor 120 to sequentially execute at least some of the plurality of operations.

The processor 120 in one embodiment may control the overall operation of the electronic device 101. For example, the processor 120 may be electrically connected to the interface 177, the input module 150, the audio output module 155, and the display module 160 to perform a designated operation.

The processor 120 of one embodiment may also execute a program stored in the memory 130 to perform a designated function. For example, the processor 120 may execute at least one of the client module 131 or the SDK 133 and perform the following operations to process user input. The processor 120 may control the operation of a plurality of apps 155 through the SDK 133, for example. The following operations described as operations of the client module 131 or SDK 133 may be operations performed by the processor 120.

The client module 131 in one embodiment may receive user input. For example, the client module 131 may receive a voice signal corresponding to a user utterance detected through the input module 150. Alternatively, the client module 131 may receive a touch input detected through the display module 160. Alternatively, the client module 131 may receive text input detected through a keyboard or visual keyboard. In addition, various types of user inputs detected through an input module included in the electronic device 101 or connected to the electronic device 101 can be received. The client module 131 may transmit the received user input to the intelligent server 200. The client module 131 may transmit status information of the electronic device 101 to the intelligent server 200 along with the received user input. The status information may be, for example, execution status information of an app.

The client module 131 of one embodiment may receive a result corresponding to the received user input. For example, when the intelligent server 200 can calculate a result corresponding to the received user input, the client module 131 may receive a result corresponding to the received user input. The client module 131 may display the received result on the display module 160. Additionally, the client module 131 may output the received result as audio through the audio output module 155.

The client module 131 of one embodiment may receive a plan corresponding to the received user input. The client module 131 may display the results of executing multiple operations of the app according to the plan on the display module 160. For example, the client module 131 may sequentially display execution results of a plurality of operations on the display module 160 and output audio through the sound output module 155. For another example, the electronic device 101 may display only some results of executing a plurality of operations (e.g., the result of the last operation) on the display module 160 and output them as audio through the sound output module 155. Can be printed.

According to one embodiment, the client module 131 may receive a request from the intelligent server 200 to obtain information necessary to calculate a result corresponding to the user input. According to one embodiment, the client module 131 may transmit the necessary information to the intelligent server 200 in response to the request.

The client module 131 in one embodiment may transmit information as a result of executing a plurality of operations according to the plan to the intelligent server 200. The intelligent server 200 can use the result information to confirm that the received user input has been processed correctly.

The client module 131 in one embodiment may include a voice recognition module. According to one embodiment, the client module 131 may recognize voice input performing a limited function through the voice recognition module. For example, the client module 131 may execute an intelligent app for processing voice input to perform an organic action through a designated input (e.g., wake up!).

The intelligent server 200 in one embodiment may receive information related to the user's voice input from the electronic device 101 through a communication network. According to one embodiment, the intelligent server 200 may change data related to the received voice input into text data. According to one embodiment, the intelligent server 200 may generate a plan for performing a task corresponding to the user's voice input based on the text data.

According to one embodiment, the plan may be generated by an artificial intelligence (AI) system. Artificial intelligence systems may be rule-based systems or neural network-based systems (e.g., feedforward neural network (FNN), recurrent neural network (RNN)). ))) It could be. Alternatively, it may be a combination of the above or a different artificial intelligence system. According to one embodiment, a plan may be selected from a set of predefined plans or may be generated in real time in response to a user request. For example, an artificial intelligence system can select at least one plan from a plurality of predefined plans.

The intelligent server 200 of one embodiment may transmit a result according to the generated plan to the electronic device 101 or transmit the generated plan to the electronic device 101. According to one embodiment, the electronic device 101 may display the results according to the plan on the display module 160. According to one embodiment, the electronic device 101 may display the results of executing an operation according to the plan on the display module 160.

The intelligent server 200 of one embodiment includes a front end 210, a natural language platform 220, a capsule DB 230, an execution engine 240, It may include an end user interface (250), a management platform (260), a big data platform (270), or an analytic platform (280).

The front end 210 of one embodiment may receive user input received from the electronic device 101. The front end 210 may transmit a response corresponding to the user input.

According to one embodiment, the natural language platform 220 includes an automatic speech recognition module (ASR module) 221, a natural language understanding module (NLU module) 223, and a planner module (223). It may include a planner module (225), a natural language generator module (NLG module) (227), or a text to speech module (TTS module) (229).

The automatic voice recognition module 221 of one embodiment may convert data related to voice input received from the electronic device 101 into text data. The natural language understanding module 223 in one embodiment may determine the user's intention using text data of voice input. For example, the natural language understanding module 223 may determine the user's intention by performing syntactic analysis or semantic analysis on user input in the form of text data. The natural language understanding module 223 in one embodiment uses linguistic features (e.g., grammatical elements) of morphemes or phrases to determine the meaning of words extracted from user input, and matches the meaning of the identified words to the user's intent. You can determine your intention. The natural language understanding module 223 can acquire intent information corresponding to the user's utterance. Intention information may be information indicating the user's intention determined by interpreting text data. Intent information may include information indicating an action or function that the user wishes to perform using the device.

The planner module 225 in one embodiment may generate a plan using the intent and parameters determined by the natural language understanding module 223. According to one embodiment, the planner module 225 may determine a plurality of domains required to perform the task based on the determined intention. The planner module 225 may determine a plurality of operations included in each of the plurality of domains determined based on the intention. According to one embodiment, the planner module 225 may determine parameters required to execute the determined plurality of operations or result values output by executing the plurality of operations. The parameters and the result values may be defined as concepts of a specified type (or class). Accordingly, the plan may include a plurality of operations and a plurality of concepts determined by the user's intention. The planner module 225 may determine the relationship between the plurality of operations and the plurality of concepts in a stepwise (or hierarchical) manner. For example, the planner module 225 may determine the execution order of a plurality of operations determined based on the user's intention based on a plurality of concepts. In other words, the planner module 225 may determine the execution order of the plurality of operations based on the parameters required for execution of the plurality of operations and the results output by executing the plurality of operations. Accordingly, the planner module 225 may generate a plan that includes association information (eg, ontology) between a plurality of operations and a plurality of concepts. The planner module 225 can create a plan using information stored in the capsule database 230, which stores a set of relationships between concepts and operations.

The natural language generation module 227 of one embodiment may change specified information into text form. The information changed to the text form may be in the form of natural language speech. The text-to-speech conversion module 229 in one embodiment can change information in text form into information in voice form.

According to one embodiment, some or all of the functions of the natural language platform 220 may be implemented in the electronic device 101.

The capsule database 230 may store information about the relationship between a plurality of concepts and operations corresponding to a plurality of domains. A capsule according to one embodiment may include a plurality of action objects (action objects or action information) and concept objects (concept objects or concept information) included in the plan. According to one embodiment, the capsule database 230 may store a plurality of capsules in the form of CAN (concept action network). According to one embodiment, a plurality of capsules may be stored in a function registry included in the capsule database 230.

The capsule database 230 may include a strategy registry in which strategy information necessary for determining a plan corresponding to a voice input is stored. The strategy information may include standard information for determining one plan when there are multiple plans corresponding to user input. According to one embodiment, the capsule database 230 may include a follow up registry in which information on follow-up actions is stored to suggest follow-up actions to the user in a specified situation. The follow-up action may include, for example, follow-up speech. According to one embodiment, the capsule database 230 may include a layout registry that stores layout information of information output through the electronic device 101. According to one embodiment, the capsule database 230 may include a vocabulary registry where vocabulary information included in capsule information is stored. According to one embodiment, the capsule database 230 may include a dialogue registry in which information about dialogue (or interaction) with a user is stored. The capsule database 230 can update stored objects through a developer tool. The developer tool may include, for example, a function editor for updating operation objects or concept objects. The developer tool may include a vocabulary editor for updating the vocabulary. The developer tool may include a strategy editor that creates and registers a strategy for determining the plan. The developer tool may include a dialogue editor that creates a dialogue with the user. The developer tool may include a follow up editor that can edit follow-up utterances to activate follow-up goals and provide hints. The subsequent goal may be determined based on currently set goals, user preferences, or environmental conditions. In one embodiment, the capsule database 230 may also be implemented within the electronic device 101.

The execution engine 240 of one embodiment may calculate a result using the generated plan. The end user interface 250 may transmit the calculated result to the electronic device 101. Accordingly, the electronic device 101 may receive the result and provide the received result to the user. The management platform 260 of one embodiment can manage information used in the intelligent server 200. The big data platform 270 in one embodiment may collect user data. The analysis platform 280 of one embodiment may manage quality of service (QoS) of the intelligent server 200. For example, the analytics platform 280 can manage the components and processing speed (or efficiency) of the intelligent server 200.

The service server 300 in one embodiment may provide a designated service (eg, food ordering or hotel reservation) to the electronic device 101. According to one embodiment, the service server 300 may be a server operated by a third party. The service server 300 in one embodiment may provide the intelligent server 200 with information for creating a plan corresponding to the received voice input. The provided information may be stored in the capsule database 230. Additionally, the service server 300 may provide result information according to the plan to the intelligent server 200.

In the integrated intelligence system 10 described above, the electronic device 101 can provide various intelligent services to the user in response to user input. The user input may include, for example, input through a physical button, touch input, or voice input.

In one embodiment, the electronic device 101 may provide a voice recognition service through an internally stored intelligent app (or voice recognition app). In this case, for example, the electronic device 101 may recognize a user utterance or voice input received through the microphone and provide a service corresponding to the recognized voice input to the user. .

In one embodiment, the electronic device 101 may perform a designated operation alone or together with the intelligent server and/or service server based on the received voice input. For example, the electronic device 101 may run an app corresponding to a received voice input and perform a designated operation through the executed app.

In one embodiment, when the electronic device 101 provides a service together with the intelligent server 200 and/or the service server, the user terminal detects a user utterance using the input module 150, A signal (or voice data) corresponding to the detected user utterance may be generated. The user terminal may transmit the voice data to the intelligent server 200 using the interface 177.

In response to a voice input received from the electronic device 101, the intelligent server 200 according to one embodiment provides a plan for performing a task corresponding to the voice input, or an operation according to the plan. can produce results. The plan may include, for example, a plurality of operations for performing a task corresponding to a user's voice input, and a plurality of concepts related to the plurality of operations. The concept may define parameters input to the execution of the plurality of operations or result values output by the execution of the plurality of operations. The plan may include association information between a plurality of operations and a plurality of concepts.

The electronic device 101 in one embodiment may receive the response using the interface 177. The electronic device 101 uses the audio output module 155 to output a voice signal generated inside the electronic device 101 to the outside, or uses the display module 160 to output a voice signal generated inside the electronic device 101. Images can be output externally.

FIG. 3 is a diagram illustrating an operation of an electronic device (e.g., the electronic device 101 of FIG. 2) to determine an endpoint detection time, according to an embodiment.

Referring to FIG. 3, the electronic device 101 includes an automatic voice recognition module 221, an end point detection (EPD) determination module 110, a sentence type determination module 111, and speech data 112 shown in FIG. 3. Alternatively, it may include at least one of the voice section detection module 116.

For example, the speech data 112 may include at least one of sentence data 113, word data 114, or EPD command 115. For example, the electronic device 101 may generate speech data 112 according to sentences and words generated based on the input user speech 195. For example, the electronic device 101 may generate speech data 112 according to the generated sentence or word based on the user input. The electronic device 101 may store the generated speech data 112 in a memory (eg, memory 130 in FIG. 2).

As an example, the sentence data 113 may include a sentence generated according to the user utterance 195. For example, word data 114 may include words generated according to user utterance 195. For example, EPD command 115 may include words and/or sentences generated according to user input.

For example, the speech data 112 is generated according to the user speech 195 and/or user input, and may be generated according to the user's speech history.

For example, the automatic voice recognition module 221 may generate text using the user utterance 195. The electronic device 101 may generate speech data 112 using text generated by the automatic speech recognition module 221.

For example, the electronic device 101 may generate speech data 112 using the final text generated by the automatic speech recognition module 221. For example, the final text may represent text generated after recognition of the user utterance 195 is completed.

According to one embodiment, the voice section detection module 116 may detect a voice section (eg, speech mode) and a non-speech section (eg, non-speech mode) using the input user speech 195. For example, user speech 195 may refer to an audio signal input to the electronic device 101.

For example, the voice section detection module 116 may include at least one of a voice section determination module 117 or a voice-based model 118. For example, the voice section detection module 116 may input the input user utterance 195 to the voice section determination module 117.

For example, the voice section determination module 117 may input the input user utterance 195 (eg, audio signal) into the voice-based model 118. The voice-based model 118 can distinguish between voice sections and non-voice sections using the input user utterance 195.

For example, the voice-based model 118 may transmit voice activity detection (VAD) to the voice section determination module 117 depending on whether voice is detected in the input user utterance 195 or when voice is not detected. there is.

For example, the voice-based model 118 may transmit a VAD of 1 to the voice section determination module 117 during the time when voice is detected. For example, the voice-based model 118 may transmit a VAD of 0 to the voice section determination module 117 during times when voice is not detected.

For example, the voice section determination module 117 may determine a voice section or a non-voice section based on the VAD. For example, for a section where VAD is 1, the voice section determination module 117 may determine the section to be a voice section. For example, for a section where VAD is 0, the voice section determination module 117 may determine the section to be a non-voice section.

For example, the voice section determination module 117 may transmit an end point detection (EPD) to the EPD determination module 110 according to the voice detection result. For example, during the time when voice is detected, the voice section determination module 117 may transmit an EPD of 1 to the EPD determination module 110. For example, during a time when no voice is detected, the voice section determination module 117 may transmit an EPD of 0 to the EPD determination module 110.

As an example, the EPD determination module 110 may determine an endpoint detection time to end recognition of the user utterance 195. For example, the end point detection time may mean a hangover time for determining the end of the utterance 195 in end point detection (EPD). For example, if the non-voice section lasts longer than the end point detection time or hangover time, the electronic device 101 may determine that the speech 195 has ended. For example, when the utterance 195 ends, the electronic device 101 may end recognition of the user utterance 195.

For example, EPD determination module 110 may determine the endpoint detection time based on utterance data 112 or sentence type. For example, the EPD determination module 110 can use the utterance data 112 to determine the endpoint detection time. For example, the EPD determination module 110 can determine the endpoint detection time using the sentence type.

As an example, the EPD determination module 110 may receive text from the automatic speech recognition module 221. For example, the EPD determination module 110 may receive partial text from the automatic speech recognition module 221. For example, the partial text may be part of the text corresponding to the user utterance 195 generated by the automatic speech recognition module 221. For example, the EPD determination module 110 uses the speech data 112 and text or partial text to determine whether the pattern of the user speech 195 is substantially the same as the pattern stored in the speech data 112. You can. For example, the EPD determination module 110 may use the speech data 112 and text or partial text to determine whether the pattern of the user speech 195 is similar to the pattern stored in the speech data 112. .

For example, when the pattern of the user's speech 195 is substantially the same as the pattern stored in the speech data 112, the EPD determination module 110 may determine the endpoint detection time according to the pattern. For example, the initially set endpoint detection time may be about 600 ms. For example, if the pattern of the user utterance 195 is substantially the same as the pattern stored in the utterance data 112, the EPD determination module 110 may determine an endpoint detection time of a length such as about 100 ms or about 2,000 ms depending on the pattern. can be decided.

For example, if the user's utterance intention of the pattern is the intention to end utterance, the EPD determination module 110 may determine the end point detection time to be about 100 ms. For example, if the user utterance intent of the pattern is the subsequent utterance intent, the EPD determination module may determine the endpoint detection time to be about 2,000 ms. For example, if the pattern of the user utterance 195 is the intent of the user utterance 195 In the case of a pattern for early termination of recognition, the EPD determination module 110 may determine an endpoint detection time (e.g., about 100 ms) that is shorter than the initially set endpoint detection time (e.g., about 600 ms).

For example, if the pattern of the user utterance 195 is a pattern for inputting a subsequent utterance after the user utterance 195, the EPD determination module 110 detects an endpoint longer than the initially set endpoint detection time (e.g., about 600 ms). The detection time (e.g., approximately 2,000 ms) can be determined.

In the above example, the initially set endpoint detection time of approximately 600 ms, the endpoint detection time determined by the EPD determination module 110 of approximately 100 ms, and 2,000 ms are exemplary and are not limited to the above example. For example, the initially set endpoint detection time may be approximately 900 ms. For example, the EPD determination module 110 may determine the endpoint detection time to be about 50 ms, about 150 ms, about 200 ms, about 1,500 ms, about 2,500 ms, etc., depending on the user's speech intention. For example, the EPD determination module 110 may determine the endpoint detection time to be shorter than the initially set endpoint detection time (fast) or longer than the initially set endpoint detection time (slow), depending on the user's speech intention.

As above, when the user's utterance 195 substantially matches a specific pattern, the electronic device 101 considers the user's intention to end the utterance 195 for voice recognition and shortens the initially set end point detection time. The endpoint detection time can be determined. The electronic device 101 can shorten the waiting time for voice recognition and improve user convenience through a short endpoint detection time.

If a new user utterance 195 is not input within a short endpoint detection time, the electronic device 101 may end recognition of the user utterance 195.

For example, ending recognition of the user utterance 195 may mean that the automatic voice recognition module 221 ends recognizing the additionally input user utterance 195. For example, when recognition of the user's utterance 195 is completed, the electronic device 101 may use the text generated by the automatic voice recognition module 221 to perform an operation that matches the user's intention.

For example, if a new user utterance 195 is recognized after a long waiting time has elapsed since the pattern, the EPD determination module 110 may determine an endpoint detection time (e.g., about 600 ms) longer than the initially set endpoint detection time (e.g., about 600 ms). 1,000 ms or about 2,000 ms, etc.) can be determined.

In the case of a pattern in which the user's speech 195 is additionally input after a long waiting time has elapsed, the electronic device 101 may determine the endpoint detection time to be longer than the initially set endpoint detection time in consideration of the user's intention. Through a long endpoint detection time, the electronic device 101 can receive the user's subsequent utterance by taking into account the user's intention not to end the utterance 195 when a specific pattern is input.

For example, if the pattern of the user's speech 195 is similar to the pattern stored in the speech data 112, the EPD determination module 110 may determine the endpoint detection time according to the pattern. For example, the initially set endpoint detection time may be approximately 600 ms. For example, if the pattern of the user's speech 195 is similar to the pattern stored in the speech data 112, the EPD determination module 110 may determine the endpoint detection time according to the pattern. For example, if the user's speech intention of the pattern is to end the user's speech, the EPD determination module 110 may determine the end point detection time to be about 100 ms. For example, if the user's utterance intention of the pattern is a follow-up utterance intention, the EPD determination module 110 may determine the endpoint detection time to be about 2,000 ms.

When the pattern of the user utterance 195 is similar to the pattern stored in the utterance data 112, the EPD determination module 110 determines whether the pattern of the user utterance 195 is substantially the same as the pattern stored in the utterance data 112. In the same way, the endpoint detection time can be determined.

For example, the EPD determination module 110 determines whether the pattern of the user utterance 195 is substantially the same as the stored pattern based on at least one of the sentence data 113 or the word data 114 stored in the utterance data 112. You can determine whether or not it is similar. For example, the EPD determination module 110 may compare the text received from the automatic speech recognition module 221 with the sentence stored in the sentence data 113. For example, the EPD determination module 110 may compare the sentence data 113 with the sentence type, composition, words, etc. of the text, and compare the pattern of the user utterance 195 with the stored pattern. For example, the EPD determination module 110 may compare the word data 114 with words in the text to compare the pattern of the user utterance 195 with the stored pattern.

As an example, the EPD determination module 110 may transmit text to the sentence type determination module 111. For example, the EPD determination module 110 may transmit partial text to the sentence type determination module 111. As an example, the EPD determination module 110 may receive a sentence type corresponding to the partial text from the sentence type determination module 111 in response to the partial text.

For example, sentence types may include incomplete sentence types, complete sentence types, and ambiguous sentence types. For example, an incomplete sentence type may mean a sentence that is not finished. For example, the incomplete sentence type may refer to a sentence that is not complete, such as “I eat rice,” “tomorrow’s weather,” “singer A song,” “tomorrow’s schedule,” etc.

For example, a complete sentence type may mean a sentence that ends completely. For example, the complete sentence type may mean a complete sentence, such as “Tell me the weather tomorrow,” “Tell me the schedule for tomorrow,” “Play singer A’s song,” etc.

For example, an ambiguous sentence type may refer to a type of sentence that does not belong to either an incomplete sentence type or a complete sentence type.

For example, the electronic device 101 may classify or determine the sentence type of the user utterance 195. For example, the sentence type determination module 111 may determine the sentence type of the received text.

For example, the sentence type determination module 111 may determine the sentence type by considering language grammar and/or special uses (e.g., usability of incomplete sentences) or constraints of the received text.

For example, the sentence type determination module 111 may determine the sentence type of the text by considering the purpose of the text. For example, texts such as “today’s weather,” “10 o’clock alarm,” and “calling mom,” may correspond to an incomplete sentence type in which the sentence does not end completely according to language grammar. Texts such as “today’s weather,” “10 o’clock alarm,” and “mom’s phone call” correspond to an incomplete sentence type according to language grammar, but the sentence type determination module 111 considers the purpose of the text and “today’s weather.” Usable incomplete sentence types such as ", "10 o'clock alarm", and "Mom's phone call" (e.g. noun text) can be classified as complete sentence types. For example, usable incomplete sentence type text may refer to text that corresponds to an incomplete sentence according to language grammar, but can be used to determine the user's intention, command, and actions according to the user's intention and command. .

For example, the sentence type determination module 111 may determine an inverted sentence to be a complete sentence. For example, the sentence “Send a message to my sister” may be classified as an incomplete sentence in terms of language grammar, but the sentence type determination module 111 may determine the sentence to be a complete sentence.

For example, the sentence type determination module 111 may determine the sentence type of the text by considering a constraint situation (e.g., ambiguity in language grammar). For example, in the case of a text that may be classified as a complete sentence type or an incomplete sentence type depending on the interpretation, such as “what is the weather tomorrow,” the sentence type determination module 111 considers the constraint situation and Text can be determined to be an ambiguous sentence type.

For example, the character type determination module 111 may determine the text as an ambiguous sentence type when the grammar type cannot be determined, such as a typo (misprint) text or a text with a single-character token. For example, when the automatic speech recognition module 221 receives text containing typos or grammatical errors, when the sentence is short and it is difficult to determine whether it is a complete or incomplete sentence, or in the case of an ambiguous sentence, the character type determination module ( 111) can determine the sentence as an ambiguous sentence type. For example, the sentence type determination module 111 may determine the sentence type of the user utterance 195 using the text and speech data 112 received from the EPD determination module 110. For example, the sentence type determination module 111 may determine the sentence type of the user utterance 195 using at least one of the sentence pattern and/or word stored in the utterance data 112.

For example, the EPD determination module 110 may determine the endpoint detection time depending on the sentence type. For example, when the sentence type is a complete sentence, the EPD determination module 110 may determine an endpoint detection time (e.g., about 300 ms) shorter than the initially set endpoint detection time (e.g., about 600 ms).

For example, when the sentence type is an incomplete sentence, the EPD determination module 110 may determine the endpoint detection time (e.g., about 1,000 ms) to be longer than the initially set endpoint detection time (e.g., about 600 ms).

For example, if the sentence type is an ambiguous sentence type, the EPD determination module 110 may determine the endpoint detection time using an initially set endpoint detection time (eg, about 600 ms).

The electronic device 101 may determine the endpoint detection time by considering the user's intention to input an additional utterance 195 according to the sentence type of the user utterance 195.

For example, the EPD determination module 110 may cause the automatic voice recognition module 221 to end recognition of the user utterance 195 according to the end point detection time. For example, if the voice section is not detected within the determined endpoint detection time, the EPD determination module 110 may cause the automatic voice recognition module 221 to end recognition of the user utterance 195. For example, when recognition of the user utterance 195 is completed, the automatic voice recognition module 221 transfers the text of the recognized user utterance 195 to a natural language understanding module (e.g., the natural language understanding module 223 in FIG. 2). It can be delivered.

For example, the automatic voice recognition module 221 may end recognition of the user utterance 195 if the voice section is not detected within the determined end point detection time. For example, the automatic voice recognition module 221 may receive a result of detecting a voice section or a non-voice section from the voice section detection module 116.

Figure 3 shows that the electronic device 101 according to one embodiment includes an automatic speech recognition module 221, an EPD determination module 110, a sentence type determination module 111, speech data 112, and a speech section detection module 116. This is a diagram showing the operation of determining the endpoint detection time using .

According to one embodiment, the processor 120 of the electronic device 101 includes the automatic speech recognition module 221, the EPD determination module 110, the sentence type determination module 111, or the speech section detection module shown in FIG. 3 ( 116), at least one operation can be performed. For example, the operation of each component shown in FIG. 3 (e.g., automatic voice recognition module 221, EPD determination module 110, character type determination module, and voice section detection module 116) is performed by the electronic device 101. It may be performed by a processor (e.g., processor 120 of FIG. 2).

In FIG. 3 above, speech data 112 may refer to the user's individual speech history. The electronic device 101 can determine an endpoint detection time suitable for an individual user by determining the endpoint detection time according to the user's individual speech history. The electronic device 101 determines the end point detection time according to the user's speech history to terminate recognition of the user speech 195 early, or delay the end of recognition of the user speech 195, and adjust the user's voice to suit the user's intent. It can be recognized.

For example, the speech data 112 may be initially set using standardized data including the speech history of a plurality of users. For example, the speech history of a plurality of users may be classified according to the users' individual information such as their residential area, gender, language used, and age. For example, speech data may be set using user information of the electronic device 101. For example, the electronic device 101 may initially set the speech data 112 using speech histories corresponding to user information in standardized data.

For example, the electronic device 101 updates the speech data 112 using the user speech history in the speech data 112 initially set based on standardized data, or updates the speech data 112 according to the user speech history. It can be replaced.

FIG. 4 is a flowchart of an operation method for an electronic device (eg, the electronic device 101 of FIG. 2 ) to determine an endpoint detection time, according to an embodiment.

In the following embodiments, each operation may be performed sequentially, but is not necessarily performed sequentially. For example, the order of each operation may be changed, and at least two operations may be performed in parallel.

According to one embodiment, operations 310 to 330 may be understood as being performed by a processor (e.g., processor 120 of FIG. 2) of an electronic device (e.g., electronic device 101 of FIG. 2).

Referring to FIG. 4 , the electronic device 101 according to an embodiment may identify whether a user utterance (eg, user utterance 195 of FIG. 3 ) is input in operation 310 . For example, the processor of the electronic device 101 (e.g., the processor 120 in FIG. 2) uses an automatic voice recognition module (e.g., the automatic voice recognition module 221 in FIGS. 2 and 3) to recognize a user's utterance ( 195) can be determined whether it is entered.

The electronic device 101 according to one embodiment may determine the EPD time based on speech data (eg, speech data 112 of FIG. 3) in operation 320. As an example, EPD time may mean endpoint detection time.

For example, in operation 320, the electronic device 101 determines the endpoint detection time based on at least one of whether the pattern of the user utterance 195 is substantially the same as or similar to the pattern stored in the utterance data 112. You can decide.

For example, the electronic device 101 may determine the endpoint detection time according to the sentence type of the user utterance 195 in operation 320. Character types may include complete sentence types, incomplete sentence types, or ambiguous sentence types.

The electronic device 101 according to one embodiment may determine whether the user utterance 195 is input within the EPD time in operation 330. For example, when a new user utterance 195 is input within the determined endpoint detection time, a new endpoint detection time may be determined according to operation 320. For example, if the user utterance 195 is not input within the determined endpoint detection time, the electronic device 101 may end recognition of the user utterance 195. For example, when recognition of the user utterance 195 ends, the electronic device 101 uses the text corresponding to the input user utterance 195 to perform an operation according to the intention included in the user utterance 195. can do.

FIG. 5 is a flowchart of an operation method for an electronic device (eg, the electronic device 101 of FIG. 2 ) to determine an endpoint detection time, according to an embodiment.

In the following embodiments, each operation may be performed sequentially, but is not necessarily performed sequentially. For example, the order of each operation may be changed, and at least two operations may be performed in parallel.

According to one embodiment, operations 405 to 460 may be understood as being performed by a processor (e.g., processor 120 of FIG. 2) of an electronic device (e.g., electronic device 101 of FIG. 2).

Referring to FIG. 5, the electronic device 101 according to one embodiment may identify whether a user utterance 195 is input in operation 405. The description of operation 310 of FIG. 4 can be applied substantially identically to operation 405.

According to one embodiment, the electronic device 101 may determine the category of the user's utterance in operation 410. For example, the processor of the electronic device 101 (eg, processor 120 in FIG. 2) may determine the category of the user's utterance using the K-means algorithm. The K-mean algorithm is an algorithm that groups given data into k clusters, and can operate by minimizing the variance of the distance difference between each cluster. For example, the processor 120 of the electronic device 101 may determine a category corresponding to a user utterance (eg, user utterance 195 of FIG. 3) from among a plurality of clustered categories.

In order to determine the category of the user's utterance, the electronic device 101 clusters a plurality of categories using a known algorithm different from the above example (e.g., Mean Shift, Gaussian Mixture Model, DBSCAN, Agglomerative Clustering, etc.), and the user The category corresponding to the utterance 195 can be determined.

According to one embodiment, in operation 415, the electronic device 101 determines whether the pattern of the user's speech 195 is substantially the same as the pattern stored in speech data (e.g., speech data 112 of FIG. 3). can do.

According to one embodiment, in operation 420, the electronic device 101 may determine whether the pattern of the user's speech 195 is similar to the pattern stored in the speech data 112.

For example, the electronic device 101 converts a sentence corresponding to the user utterance 195 and a word included in the user utterance 195 into sentence data (e.g., sentence data 113 in FIG. 3) and/or word data (e.g. : Similarity can be determined by comparing with the word data 114 of FIG. 3. For example, the electronic device 101 may determine similarity based on a similarity model. For example, a similarity model may refer to a pre-trained, tuned model. For example, the similarity model can use an input sentence to determine whether it is similar to stored sentences and words.

For example, the electronic device 101 may compare the text corresponding to the user utterance 195 and the sentence stored in the sentence data 113 in operations 415 and 420 . For example, the electronic device 101 may determine whether the sentence pattern of the text is substantially the same as the sentence pattern stored in the sentence data 113. For example, the electronic device 101 may determine whether the sentence pattern of the text is similar to the sentence pattern stored in the sentence data 113.

For example, in operations 415 and 420, the electronic device 101 determines whether the pattern of the user speech 195 is substantially the same or similar to the pattern stored in the speech data 112, Various known algorithms can be used to determine the similarity of sentences.

According to one embodiment, the electronic device 101 may determine whether the sentence type of the user utterance 195 is a complete sentence type in operation 425.

According to one embodiment, the electronic device 101 may determine whether the sentence type of the user utterance 195 is an incomplete sentence type in operation 430.

For example, in operations 425 and 430, the electronic device 101 may determine the sentence type of the user utterance 195 according to the type of the last word of the user utterance 195. For example, depending on the user's speech history, each word included in the word data 114 may be set to one of an obvious complete sentence type, a complete sentence type, an obvious incomplete sentence type, an incomplete sentence type, and an ambiguous sentence type.

For example, when the utterance 195 in which no additional voice is input after word A is repeatedly input, the electronic device 101 may set word A as a complete sentence type. For example, when the utterance 195 in which no additional voice is input after word A is input beyond a specified threshold, the electronic device 101 may set word A as an obvious complete sentence type.

For example, when the utterance 195 in which an additional voice is input after word B is repeatedly input, the electronic device 101 may set word B as an incomplete sentence type. For example, when the utterance 195 in which additional speech is input after word B is input beyond a specified threshold, the electronic device 101 may set word B as an obvious incomplete sentence type.

For example, when an utterance 195 in which an additional voice is input after word C and an utterance 195 in which no additional voice is input are input, the electronic device 101 may set word C as an ambiguous sentence type.

In the above example, the electronic device 101 may set a sentence type for each word by comparing the number of times an additional utterance 195 is input after each word with a threshold. The sentence type set for each word can be set according to the user's speech pattern. For example, in the above example, since there is a tendency for no additional speech to be input after inputting word A due to the user's speech pattern, the electronic device 101 may set the type of word A to an obvious complete sentence type.

For example, the electronic device 101 may determine the sentence type of the user utterance 195 according to the sentence type corresponding to the last text of the user utterance 195.

For example, in operations 425 and 430, if the sentence type of the user utterance 195 does not correspond to a complete sentence or an incomplete sentence, the electronic device 101 determines the sentence type of the user utterance 195 to be ambiguous. It can be judged from sentences.

The method for the electronic device 101 to determine the sentence type of the user utterance 195 is not limited to the examples described in operations 425 and 430 above. For example, the electronic device 101 may use various known methods to determine whether the user utterance 195 corresponds to a complete sentence or an incomplete sentence and determine the user's intention to input additional voice or not to input voice. there is.

In the above operations 415, 420, 425, and 430, the electronic device 101 may use speech data 112 corresponding to the category of the user speech. For example, if the category of the user utterance is A, the electronic device 101 uses the utterance data 112 corresponding to category A among the utterance data 112 in operation 415 to create the user utterance 195. It can be determined whether the pattern is the same as the stored pattern.

For example, in operation 415, the electronic device 101 stores the pattern of the user utterance 195 using sentence data 113 and/or word data 114 corresponding to category A of the user utterance. It can be determined whether it is substantially the same as .

Substantially the same as the above example, in at least one of operation 420, operation 425, or operation 430, the electronic device 101 generates speech data 112 of a category substantially the same as the category of the user speech. ) can be used to operate.

For example, in operations 435, 440, 445, 450, and 455 below, the length of the very short endpoint detection time is the first length, and the length of the short endpoint detection time is may correspond to the second length, the length of the initially set endpoint detection time may correspond to the third length, the length of the long endpoint detection time may correspond to the fourth length, and the length of the very long endpoint detection time may correspond to the fifth length. For example, the size of the length of each endpoint detection time may be set as (first length) < (second length) < (third length) < (fourth length) < (fifth length).

According to one embodiment, the electronic device 101 may determine the EPD time according to the matching pattern in operation 435. For example, the electronic device 101 may determine the endpoint detection time according to the matching pattern in operation 435.

For example, the electronic device 101 may determine the endpoint detection time by considering the user's utterance intention. For example, if the pattern of the user utterance 195 substantially matches the pattern stored in the utterance data 112, and the user utterance intention of the pattern is the intention to end the utterance, the electronic device 101 may The endpoint detection time of a short length (eg, the endpoint detection time of the first length) may be determined.

For example, if the pattern of the user utterance 195 substantially matches the pattern stored in the utterance data 112, and the user utterance intention of the pattern is the intention to end the utterance, it may mean that the user's utterance 195 is completed. there is. If the pattern of the user's utterance 195 substantially matches the pattern stored in the utterance data 112 and the user's utterance intention of the pattern is the intention to end the utterance, the electronic device 101 determines that the user's utterance 195 is completed. , the endpoint detection time of the first length, which is the shortest length among the plurality of set endpoint detection time lengths, can be determined.

For example, if the pattern of the user utterance 195 matches the pattern stored in the utterance data 112, and the user utterance intention of the pattern is the intention of a subsequent utterance, the electronic device 101 The endpoint detection time (e.g., the endpoint detection time of the fifth length) can be determined. For example, in the case of a pattern in which the user's utterance intention is the intention of a subsequent utterance, the user may delay the end of the utterance 195, and the user's utterance 195 may not be completed for a long time. If the pattern of the user's utterance 195 matches a specific pattern stored in the utterance data 112 and the user's utterance intention of the pattern is a subsequent utterance intention, the electronic device 101 determines that the user's utterance 195 is not complete. By making this determination, the endpoint detection time of the fifth length, which is the longest length among the set plurality of endpoint detection times, may be determined. For example, a specific pattern corresponding to the endpoint detection time of the fifth length may be determined according to user input. . For example, a specific pattern corresponding to a very long endpoint detection time may be determined based on the user's speech history.

According to one embodiment, the electronic device 101 may determine the EPD time according to a similar pattern in operation 440. For example, the electronic device 101 may determine the endpoint detection time according to a similar pattern in operation 440.

For example, in operation 440, the electronic device 101 may detect an endpoint detection time longer than the initially set endpoint detection time (e.g., an endpoint of the fourth length or the fifth length) according to the similar pattern and/or the user's utterance intention of the similar pattern. detection time) or an endpoint detection time shorter than an initially set endpoint detection time (e.g., endpoint detection time of the first length or second length) may be determined. With respect to operation 440, the description of operation 435 may be applied substantially identically.

For example, if the pattern of the user's utterance 195 is similar to the pattern stored in the utterance data 112 and the user's utterance intention of the similar pattern is the intention to end the utterance, the electronic device 101 may Upon determining that it is complete, the endpoint detection time may be determined to be a second length shorter than the initially set endpoint detection time.

For example, if the pattern of the user's utterance 195 is similar to the pattern stored in the utterance data 112 and the user's utterance intention of the similar pattern is the subsequent utterance intention, the electronic device 101 may If it is determined that it is not completed, the endpoint detection time may be determined to be a third length longer than the initially set endpoint detection time.

According to the 112-day embodiment, the electronic device 101 may set an endpoint detection time corresponding to each sentence pattern. For example, the electronic device 101 may set the end point detection times of the first and second lengths in sentence pattern 1, and set the end point detection times of the fifth length and fourth length in sentence pattern 2.

For example, if the pattern of the user utterance 195 in operation 415 is substantially the same as sentence pattern 1, the electronic device 101 may determine the endpoint detection time of the first length in operation 435. For example, if the pattern of the user utterance 195 in operation 420 is similar to sentence pattern 1, the electronic device 101 may determine the endpoint detection time of the second length in operation 440.

For example, if the pattern of the user utterance 195 in operation 415 is substantially the same as sentence pattern 2, the electronic device 101 may determine the endpoint detection time of the fifth length in operation 435. For example, if the pattern of the user's utterance 195 is similar to sentence pattern 2 in operation 420, the electronic device 101 may determine the endpoint detection time of the fourth length in operation 440.

In the above example, the end point detection time of the first length set in sentence pattern 1 may correspond to the end point detection time set in the user utterance having substantially the same pattern as sentence pattern 1. For example, the endpoint detection time of the second length set in sentence pattern 1 may correspond to the endpoint detection time set in a user utterance having a similar pattern to sentence pattern 1. The above description can be equally applied to the end point detection time of the fifth length and the end point detection time of the fourth length set in sentence pattern 2.

According to one embodiment, the electronic device 101 may determine the EPD time according to the full mode in operation 445. For example, if the sentence type of the user utterance 195 is a complete sentence type or an obvious complete sentence type, the electronic device 101 may detect an endpoint detection time shorter than the initially set endpoint detection time (e.g., the first length or the second length). 2 length endpoint detection time) can be determined.

According to one embodiment, the electronic device 101 may determine the EPD time according to the incomplete mode in operation 450. For example, if the sentence type of the user utterance 195 is an incomplete sentence type or an obviously incomplete sentence type, the electronic device 101 may set an endpoint detection time longer than the initially set endpoint detection time (e.g., fourth length or second length). 5 length endpoint detection time) can be determined.

According to one embodiment, the electronic device 101 may determine the endpoint detection time according to a preset EPD time in operation 455. For example, when the sentence type of the user utterance 195 is an ambiguous sentence type, the electronic device 101 may determine an initially set endpoint detection time.

The above division of the first to fifth lengths is illustrative and is not limited to the above examples. For example, the length of each endpoint detection time is divided into a first length to a third length, and the length can be set as (first length) > (second length) > (third length).

In FIGS. 4 and 5 above, the electronic device 101 may determine the endpoint detection time using speech data 112 generated based on the user's speech history. The electronic device 101 determines the end point detection time based on the utterance 195 pattern frequently input by the user, taking into account the user's intention to end the utterance or the intention to make a subsequent utterance when the user inputs the utterance 195, The end point of utterance 195 can be detected.

For example, when the user ends the utterance 195, the electronic device 101 sets an end point detection time shorter or longer than the initially set end point detection time to end recognition of the utterance 195 early. , it can operate according to the input utterance 195.

4 and 5 above are examples showing operations performed by the electronic device 101 to determine the endpoint detection time, and are not limited to the order of operations shown in FIGS. 4 and 5, and some operations may be omitted. there is.

FIG. 6 is a diagram illustrating speech data (eg, speech data 112 of FIG. 3) according to an embodiment.

For example, an electronic device (e.g., the electronic device 101 of FIG. 2) may generate speech data 112 shown in FIG. 6 using a previously entered user speech history. For example, the electronic device 101 includes the number of utterances 510, the input date 520, the category 530 of the utterance, and the number of categories of the previously entered user utterance (e.g., the user utterance 195 in FIG. 3). Speech data 112 can be generated according to 540. For example, the number of utterances 510 may refer to the number of inputs of the corresponding utterance 195, and the number of categories 540 may refer to the number of inputs of the corresponding category. For example, in Figure 6, the music category 530-1 may be entered 33 times, the weather category 530-2 may be entered 20 times, and the phone category 530-3 may be entered 10 times.

For example, the input date 520 may indicate the date on which the corresponding utterance 195 was input or the date on which the corresponding utterance 195 was last input.

FIG. 7 is a diagram illustrating a user interface for registering an utterance (e.g., user utterance 195 of FIG. 3) to utterance data (e.g., utterance data 112 of FIG. 3) according to an embodiment.

For example, an electronic device (e.g., the electronic device 101 of FIG. 2) may provide a user interface for registering a user utterance 195 as shown in FIG. 7. For example, the electronic device 101 includes a display module (e.g., the display module 160 of FIG. 2) and may provide a user interface 700 as shown in FIG. 7 using the display module 160. .

For example, the electronic device 101 may generate speech data 112 based on user input. For example, using the user speech registration interface 710 in FIG. 7, “today’s weather” (710-1), “play A song” (710-2), “tomorrow’s weather” (710-3), You can enter sentences such as “Play song B” (710-4) or “Latest news” (710-5). The electronic device 101 may generate speech data 112 by registering the sentences 710-1 to 710-5 shown in the user speech 195 registration item in FIG. 7 as the user speech 195. For example, the electronic device 101 may generate speech data 112 such as the speech data 112 shown in FIG. 6 using the registered user speech 195.

As an example of the display module 160, the electronic device 101 may update the speech data 112 according to a preset update standard. For example, the update criteria may include criteria such as a certain period of time and the number of newly input user utterances 195 (e.g., the number of utterances 510 in FIG. 6. For example, the electronic device 101 may The input user utterances 195 are stored in a storage space (e.g., memory 130 in FIG. 2), and are stored every certain period (e.g., one week) or as set by the number 510 of newly stored user utterances 195. If the number exceeds the number, the stored user utterance 195 can be processed and the utterance data 112 can be updated.

For example, the electronic device 101 may display a display object 720 on the display module that can set user inputs such as “run now” and “run with reservation” shown in FIG. 7 . The electronic device 101 may update the speech data 112 based on user input for settings such as “execute immediately” or “execute with reservation.” For example, when receiving a “run now” input from a user, the electronic device 101 may process the user utterance 195 stored in the storage space and update the utterance data 112. For example, when receiving a “reservation execution” input from the user, the electronic device 101 may process the user utterance 195 stored in the storage space at a set time and update the utterance data 112.

Figure 8 is a diagram showing categories of utterances according to an embodiment.

For example, an electronic device (e.g., electronic device 101 of FIG. 2) may use a previously input user utterance (e.g., user utterance 195 of FIG. 3) to cluster it into a plurality of categories. For example, the electronic device 101 divides the pre-entered user utterances 195 into a plurality of categories, such as music, weather, alarm, and call, as shown in FIG. It can be clustered.

For example, the electronic device 101 may determine the category of the input user utterance. For example, the electronic device 101 may determine the category of the newly input user utterance 195 “Play song A” as music among the categories as shown in FIG. 8 .

For example, the category of utterances shown in FIG. 8 represents categories of utterances clustered by the electronic device 101 according to one embodiment using the K-means algorithm. For example, the electronic device 101 may generate K random centroids and include each point in the cluster of the nearest centroid. The electronic device 101 may calculate a new center point using the average of the points included in each cluster.

For example, the electronic device 101 may perform sentence embedding clustering using the K-means algorithm. For example, the electronic device may include a sentence embedding clustering module.

For example, according to a certain number of utterance categories (e.g., K), initial K center points can be created. When a set number or more of user speech data (e.g., speech data 112 of FIG. 2) is stored, the electronic device 101 may update the speech category using the sentence embedding clustering module. For example, the electronic device 101 may include points according to the stored speech data 112 in a cluster whose center point is closest and update the center point of each cluster. For example, the electronic device 101 may use a pre-learned sentence embedding vector table to express a sentence corresponding to the speech data 112 as a vector. The electronic device 101 may use a vector in addition to the K-means algorithm described above. It is possible to classify utterance categories and categories of utterance data 112 using various methods, and the utterance categories are not limited to the utterance categories shown in FIG. 8.

Figure 9 is a diagram showing the category proportions of utterances according to an embodiment.

For example, an electronic device (e.g., electronic device 101 in FIG. 2) clusters a pre-entered user utterance (e.g., user utterance 195 in FIG. 3) as shown in FIG. The specific gravity can be expressed as shown in Figure 9. For example, in Figure 9, the proportion of the music category (Music in Fig. 9) is about 49%, the proportion of the phone category (Call in Fig. 9) is about 29%, and the proportion of the weather category (Weather in Fig. 9) is about 15%. %, This is a diagram showing an example where the proportion of the alarm category (Alarm in FIG. 9) is about 7%.

For example, the electronic device 101 may determine the category of the user's utterance based on the weight of the category. For example, as a result of the clustering of the newly input user utterance 195 in FIG. 8, when the distances between the center points of the music category and the alarm category are the same, the electronic device 101 performs clustering according to the proportion of each category shown in FIG. 9. By applying weights, the category of the user's utterance can be determined. For example, the electronic device 101 may determine the category of the user's utterance by assigning greater weight to a category with a greater weight.

For example, in the category proportions shown in FIG. 9, as a result of clustering of the user utterance 195, when the distance between the center point of the music category and the alarm category is the same, the proportion of the music category (about 49%) is the proportion of the alarm category (about 49%). Since it is greater than about 7%), the electronic device 101 may determine the category of the user utterance 195 to be the music category.

FIG. 10 is a diagram illustrating sentence data (eg, sentence data 113 in FIG. 3) according to an embodiment.

Referring to FIG. 10, an electronic device (e.g., electronic device 101 of FIG. 2) according to an embodiment generates sentence data 113 based on speech data (e.g., speech data 112 of FIG. 3). can do. For example, the electronic device 101 may generate sentence data 113 shown in FIG. 10 using the speech data 112 shown in FIG. 6 . For example, the sentence data 113 may be classified according to utterance, count of utterance, and category.

FIG. 11 is a diagram illustrating word data (eg, word data 114 in FIG. 3) according to an embodiment.

Referring to FIG. 11, an electronic device (e.g., electronic device 101 of FIG. 2) according to an embodiment generates word data 114 based on speech data (e.g., speech data 112 of FIG. 3). can do. For example, the electronic device 101 may generate word data 114 shown in FIG. 11 using the speech data 112 shown in FIG. 6 . For example, word data 114113 may be classified according to utterance, count of utterance, category, and type.

For example, the type corresponding to a word can be set to complete, clear complete, incomplete, clear incomplete, and ambiguous.

For example, the electronic device 101 uses the last text and word data 114 of the text corresponding to the user utterance (e.g., the user utterance 195 in FIG. 3) to determine the sentence type of the user utterance 195. You can decide. For example, text corresponding to the user utterance 195 may be generated according to the user utterance 195 in the automatic speech recognition module 221.

For example, if the type of word corresponding to the last text is complete, the electronic device 101 may determine the sentence type of the user utterance 195 as a complete sentence. For example, if the type of the word corresponding to the last text is a clear complete sentence, the electronic device 101 may determine the sentence type of the user utterance 195 as a clear complete sentence.

For example, if the type of word corresponding to the last text is incomplete, the electronic device 101 may determine the sentence type of the user utterance 195 as an incomplete sentence. For example, if the type of word corresponding to the last text is clearly incomplete, the electronic device 101 may determine the sentence type of the user utterance 195 to be a clear incomplete sentence.

For example, if the type of word corresponding to the last text is ambiguous, the electronic device 101 may determine the sentence type of the user utterance 195 to be an ambiguous sentence. For example, the electronic device 101 may determine the endpoint detection time according to the sentence type of the user utterance 195. For example, when the sentence types of the user utterance 195 are a clear complete sentence, a complete sentence, an incomplete sentence, and a clear incomplete sentence, the electronic device 101 has a very short endpoint detection time (e.g., about 100 ms) and a short endpoint detection time. This can be determined by detection time (e.g., approximately 300 ms), long endpoint detection time (e.g., approximately 800 ms), or very long endpoint detection time (e.g., approximately 1,000 ms).

For example, the electronic device 101 may determine the type of word using the speech data 112. For example, when another utterance 195 is always input after word A, the electronic device 101 may determine the type of word A as a clear incomplete. For example, if no other utterance 195 is always input after word B, the electronic device 101 may determine the type of word B as a clear complete.

For example, the electronic device 101 may determine the type of word C according to the frequency with which another utterance 195 is input after word C. For example, the electronic device 101 may determine the type of word C by comparing the ratio of cases where another utterance 195 is input after word C and cases where no other utterance 195 is input with a set ratio. .

For example, using a plurality of preset ratios, the ratio of cases where another utterance 195 is input after word C and cases where no other utterance 195 is input, and a plurality of preset ratios of word C The type can be determined as at least one of obvious complete, complete, incomplete, obviously incomplete, and ambiguous.

For example, referring to FIGS. 6 to 11 , the electronic device 101 may determine whether the user utterance 195 is substantially the same as the pattern stored in the utterance data 112. When the user speech 195 is substantially the same as the pattern stored in the speech data 112, the electronic device 101 may determine the endpoint detection time according to the pattern.

For example, when the user utterance 195 “Play singer A’s song” is input, the electronic device 101 may determine the category of the utterance as “music” as shown in FIG. 8 .

For example, the electronic device 101 uses the user utterance 195 “Play singer A’s song” and the sentence data 113 of the “music” category of FIG. 10 to create a pattern of the user utterance 195. The patterns stored in the utterance data 112 can be compared. For example, the electronic device 101 may determine that the pattern of the user utterance 195 “Play a song by singer A” and the pattern of “Play a song XX” of the sentence data 113 of FIG. 10 are substantially the same. there is.

For example, if the pattern of the user utterance 195 “Play a song by singer A” and the pattern of “Play a song XX” stored in the utterance data 112 are substantially the same, the electronic device 101 may operate according to the matching pattern. The endpoint detection time can be determined. For example, the electronic device 101 may determine an endpoint detection time to be very short (eg, about 100 ms), based on the matching pattern.

For example, referring to FIGS. 6 to 11 , the electronic device 101 may determine whether the user utterance 195 is similar to the pattern stored in the utterance data 112. If the user's speech 195 is similar to the pattern stored in the speech data 112, the electronic device 101 may determine the endpoint detection time according to the pattern.

For example, when the user utterance 195 “City A weather” is input, the electronic device 101 may determine the category of the utterance as “weather” as shown in FIG. 8 .

For example, the electronic device 101 uses the user utterance 195 "City A Weather" and the sentence data 113 of the "weather" category in FIG. 10 to determine the pattern and utterance of the user utterance 195. Patterns stored in data 112 can be compared. For example, the electronic device 101 may determine that the pattern of the user utterance 195 “city A weather” and the pattern of “XX weather” of the sentence data 113 of FIG. 10 are similar.

For example, the electronic device 101 uses the match probability between the pattern of the user utterance 195 and the pattern stored in the utterance data 112 to determine the pattern of the user utterance 195 and the pattern stored in the utterance data 112. It can be determined whether they are substantially identical, similar, or dissimilar. For example, the electronic device 101 may calculate a match probability between the pattern of the user's speech 195 and the pattern stored in the speech data 112. For example, the match probability may indicate the degree to which the pattern of the user's speech 195 and the pattern stored in the speech data 112 are similar.

For example, when the match probability exceeds a preset first threshold, the electronic device 101 determines that the pattern of the user speech 195 and the pattern stored in the speech data 112 are substantially the same. You can.

For example, when the match probability is below the preset first threshold or above the preset second threshold, the electronic device 101 determines that the pattern of the user speech 195 and the pattern stored in the speech data 112 are similar. You can judge. The size of the second threshold may be set to be less than the first threshold.

For example, when the match probability is less than or equal to a preset second threshold, the electronic device 101 may determine that the pattern of the user speech 195 and the pattern stored in the speech data 112 are dissimilar.

For example, if the pattern of the user utterance 195 “city A weather” and the pattern of “XX weather” stored in the utterance data 112 are similar, the electronic device 101 may determine the endpoint detection time according to the matching pattern. there is. For example, the electronic device 101 may determine the endpoint detection time to be short (eg, about 200 ms) based on a similar pattern.

For example, when the user utterance 195 "singer A sings" is input, the electronic device 101 may, similar to the operation described above, convert the pattern of the user utterance 195 "singer A sing" into sentence data ( It can be judged to be similar to the pattern of “Play XX song” or “XX song” stored in 113). When the user utterance 195 “Singer A sings” is input, the electronic device 101 may determine a short endpoint detection time according to a similar pattern.

When the above user utterance (195) “Singer A song” is input, even if the sentence type of the input user utterance (195) is incomplete, the electronic device 101 Intention can be determined. Even when an incomplete sentence is input, the electronic device 101 can determine the intention to end speech or follow-up speech according to the pattern of the user speech 195 and determine the end point detection time.

FIG. 12 is a diagram illustrating an endpoint detection time determined by an electronic device (e.g., the electronic device 101 of FIG. 2) for each utterance (e.g., the user utterance 195 of FIG. 3) according to an embodiment.

The endpoint detection time shown in FIG. 12 indicates that the electronic device 101 is determined using the user speech 195 and speech data (eg, speech data 112 in FIG. 3).

Referring to (a) of FIG. 12 , when a user utterance 195 such as “Call mom” is input, the electronic device 101 may determine the endpoint detection time using the utterance data 112.

For example, if the patterns and words of the user utterance 195 "Call mom" are substantially the same as the patterns and words stored in the utterance data 112, the electronic device 101 may have a very short endpoint detection time ( Example: approximately 100 ms) can be determined.

For example, in (a), the electronic device 101 determines that the user's utterance 195 has ended according to the user's intention using the user utterance 195 and the utterance data 112, and determines the end point detection time can be set very short.

Referring to (b) and (c) of FIG. 12, when a user utterance 195 such as “what is the” or “lunch today” is input, the electronic device 101 detects the end point using the utterance data 112. You can decide the time.

For example, by comparing the patterns and words of the user utterance 195 "what is the" or "lunch today" with the patterns and words stored in the utterance data 112, the electronic device 101 determines the long endpoint detection time, e.g. : approximately 1,000 ms) can be determined.

For example, the electronic device 101 may predict that a subsequent utterance 195 will occur from a pattern stored in the utterance data 112 similar to the pattern of “what is the” or “lunch today” and determine the endpoint detection time. You can decide for a long time.

For example, the electronic device 101 may determine the sentence type of the user utterance 195 using the type of “lunch” and the last word “the” of “what is the” or “lunch today.” For example, if the types of words “the” and “lunch” stored in the word data 114 are incomplete or clearly incomplete, the electronic device 101 may determine that the user utterance 195 has not ended.

For example, if the additional user utterance 195 "abc" or "what is" is input within the determined endpoint detection time (e.g., about 1,000 ms), the electronic device 101 recognizes the additionally input user utterance 195. can do.

For example, when an additional user utterance 195 is input, the electronic device 101 may re-determine the endpoint detection time using the newly input utterance 195. For example, in (b), the electronic device 101 uses the pre-input utterance 195 “what is the” and the additional user utterance 195 “abc” to match the pattern of the user utterance 195 and the stored pattern. can be compared. For example, when the pattern of the user's speech 195 is substantially the same as the stored pattern, the electronic device 101 may determine a short endpoint detection time (e.g., about 200 ms), as shown in (b) of FIG. 12.

For example, the electronic device 101 may use the newly input utterance 195 “abc” to determine the end point detection time according to the type of the last word “abc”.

For example, in (c), the electronic device 101 uses the pre-entered utterance 195 “lunch today” and the additional user utterance 195 “what is it?” to create a pattern of the user utterance 195 and the stored pattern. can be compared. For example, if the pattern of the user's speech 195 is similar to the stored pattern, the electronic device 101 may determine a short endpoint detection time (eg, about 200 ms), as shown in (b) of FIG. 12.

For example, the electronic device 101 may receive all user utterances 195 using the pre-entered utterances 195 and additional user utterances 195. For example, the electronic device 101 may identify user utterances 195 such as “what is the abc” and “what is for lunch today” in (b) and (c) of FIG. 12 , respectively.

In Figures 12 (b) and 12 (c), the electronic device 101 uses the patterns of the user's utterances 195 "what is the abc" and "what is for lunch today", which are substantially the same as, similar to, the stored patterns. Alternatively, it can be determined whether they are dissimilar. The electronic device 101 may determine the endpoint detection time depending on whether the pattern of the user speech 195 and the stored pattern are substantially the same, similar, or dissimilar and the user's speech intention of the stored pattern.

(c) of FIG. 12 may represent a case where the user utterance 195 corresponds to an incomplete sentence and the pattern of the user utterance 195 is similar to the stored pattern. For example, in (c) of FIG. 12, the user utterance intention of the stored pattern similar to the user utterance 195 “What is lunch today” may be the intention to end the utterance. The electronic device 101 may determine the user's intention (e.g., intention to end utterance or intention to follow up utterance) according to the stored pattern, and determine a short endpoint detection time (e.g., approximately 200 ms).

For example, in (c) of FIG. 12, when the user utterance intention of the stored pattern similar to the user utterance 195 “What is for lunch today” is a follow-up utterance intention, the electronic device 101 makes a long utterance according to the follow-up utterance intention. The endpoint detection time (e.g., approximately 1,000 ms) can be determined.

Referring to (d) and (e) of FIG. 12, by comparing the pattern of the user's speech 195 with the stored pattern, the electronic device 101 can determine a very long endpoint detection time. For example, by comparing the pattern in which the subsequent utterances 195 are input after “Baby” and “Hey” stored in the utterance data 112 and the user utterances 195 “Baby” and “Hey”, , the electronic device 101 may determine a very long endpoint detection time (e.g., approximately 2,000 ms).

For example, the electronic device 101 determines the endpoint detection time using the frequency or number of patterns in which subsequent utterances 195 are input after “Baby” and “Hey” stored in the utterance data 112. You can. For example, the higher the frequency of the pattern in which the subsequent utterance 195 is input after “Baby” and “Hey” stored in the utterance data 112, the longer the electronic device 101 increases the length of the endpoint detection time. You can decide for a long time.

Referring to (d) and (e) of FIG. 12, when the user utterance 195 is an obviously incomplete sentence, the electronic device 101 may determine a very long endpoint detection time. For example, in (d) and (e), when the types of words “Baby” and “Hey” stored in the word data 114 are obviously incomplete, the electronic device 101 has a very long endpoint detection time. (e.g. approximately 2,000 ms) can be determined.

For example, if the pattern of the user's utterance 195 is an incomplete sentence with a high utterance frequency or an obviously incomplete sentence, the electronic device 101 may determine a very long endpoint detection time (e.g., about 2,000 ms). For example, based on the speech pattern of the user speech 195, the electronic device 101 may determine the user's speech intention as the subsequent speech intention in the case of an incomplete word or a follow-up word with a high frequency of speech. The electronic device 101 may determine a long end point detection time (e.g., about 2,000 ms) in the case of incomplete words or follow-up words with high utterance frequency.

In Figures 12 (d) and (e), when additional user utterances 195 "call mom" and "Call mom" are entered within the end point detection time (about 2,000 ms), the new end point detection time ( Example: approximately 100 ms) can be determined. The operation of determining a new endpoint detection time in consideration of the additional user utterance 195 can be substantially applied in the same manner as described in (b) and (c) of FIG. 12 .

In Figures 12(f) and 12(g), the pattern of the user utterance 195 "Suwon" or "hmm" may not be substantially the same as or similar to the stored pattern. User utterance (195) The sentence type "Suwon" or "hmm" may correspond to an ambiguous sentence. If the sentence type of the user utterance 195 is an ambiguous sentence, the electronic device 101 may determine the end point detection time using a preset end point detection time (eg, about 600 ms).

In Figures 12 (f) and (g), after the user utterance 195 "Suwon" or "hmm", an additional user utterance 195 "weather ( weather)" and "youtube" are entered, the new endpoint detection time (e.g., about 100 ms) can be determined.

12(f) and 12(g), the electronic device 101 according to one embodiment may determine the endpoint detection time based on the user's speech intention. The entire user utterances (195) “Suwon weather” and “hmm Youtebe” in (f) and (g) of Figure 12 may correspond to obvious incomplete sentences in terms of language grammar.

For example, the electronic device 101 may determine the user's utterance intention from a pattern that is substantially the same or similar to the user utterance 195 even when the user utterance 195 is an obviously incomplete sentence in terms of language grammar. Based on the user's utterance intention, the electronic device 101 may determine a short endpoint detection time or a very short endpoint detection time even when the user utterance 195 is an obviously incomplete sentence. For example, as shown in (f) and (g) of FIGS. 12, even when the user utterance 195 is an obviously incomplete sentence, the electronic device 101 considers the user's utterance intention and detects a very short endpoint detection time (e.g. : approximately 100ms) can be determined.

In (f) and (g) of FIG. 12, the user's utterance intention of the user utterance 195 may be the intention to end the utterance. For example, the electronic device 101 may determine the user's utterance intention of the user utterance 195 based on the utterance data 112. For example, in (f) of FIG. 12, when the user's utterance intention with a pattern that is substantially the same or similar to the user's utterance 195 "Suwon weather" is the intention to end the utterance, the electronic device 101 User utterance (195) The user's utterance intention of "Suwon weather" can be determined as the intention to end the utterance.

For example, the electronic device 101 may use the user's speech pattern of the speech data 112 to determine the user's speech intention of the pattern. For example, when pattern A is repeatedly input without a subsequent utterance being input, the electronic device 101 may determine that the user's utterance intention corresponding to pattern A is an intention to end the utterance. For example, when pattern B is repeatedly input as a follow-up utterance, the electronic device 101 may determine that the user's utterance intention corresponding to pattern B is a follow-up utterance intention.

For example, the electronic device 101 may determine the user's speech intention of the pattern by comparing the number of times the pattern stored in the speech data 112 has been input with a preset threshold. For example, if pattern A is input more than a set threshold (e.g., 10 times) without a follow-up utterance being input, the electronic device 101 may determine that the user's utterance intention corresponding to pattern A is the intention to end the utterance. there is.

For example, if pattern A is input a times without a follow-up utterance being input and b times with a follow-up utterance, the electronic device 101 determines a number of times pattern A has been input without a follow-up utterance being input, a follow-up utterance The user's utterance intention corresponding to pattern A can be determined using the number b of pattern A input and the set threshold value. For example, when (a-b) is greater than or equal to a set threshold, the electronic device 101 may determine that the user's utterance intention corresponding to pattern A is the intention to end the utterance. For example, if (b-a) is greater than or equal to a set threshold, the electronic device 101 may determine that the user's utterance intention corresponding to pattern A is the subsequent utterance intention.

FIG. 13 is a diagram illustrating an operation in which an electronic device (e.g., the electronic device 101 of FIG. 2) determines an endpoint detection time according to an EPD command (e.g., the EPD command 115 of FIG. 3) according to an embodiment. .

For example, the EPD command 115 may include at least one of a termination command 1310 (forced EPD command), a delay command 1320 (long pause command), or an initialization command 1330 (reset command).

For example, end instructions 1310 (e.g., utterances “end”, “end”, “end”, etc.) and delay instructions 1320 (e.g., utterances “wait”, “long”, “wait”, “hold”) ", etc.) recognizes an utterance (e.g., the user utterance 195 in FIG. 3) according to an end point detection time of a preset length when the end command 1310 or delay command 1320 is input to the electronic device 101. It may mean a command to terminate.

For example, the length of the endpoint detection time corresponding to the end command 1310 is set very short (e.g., about 100 ms), and the length of the endpoint detection time corresponding to the delay command is set very long (e.g., about 2,000 ms). ) can be.

Referring to (a) of FIG. 13, when the end command 1310 along with utterance 1 is input to the electronic device 101, the electronic device 101 sets the end point detection time to a preset length (e.g., about 100 ms). You can decide. In (a) of FIG. 13, the electronic device 101 may determine the user's intention to end speech input based on the end command 1310. The electronic device 101 may end recognition of the utterance 195 faster than the initially set endpoint detection time (e.g., about 600 ms) according to the intention to end the utterance, and operate quickly according to the input user utterance 195.

Referring to (b) of FIG. 13, when the delay command 1320 along with utterance 1 is input to the electronic device 101, the electronic device 101 sets the endpoint detection time to a preset length (e.g., about 2,000 ms). You can decide. In (b) of FIG. 13, the electronic device 101 may determine the user's intention to input a follow-up speech based on the delay command 1320. The electronic device 101 may delay the end of recognition of the utterance 195 beyond the initially set endpoint detection time (eg, about 600 ms) depending on the intention of the subsequent utterance. The electronic device 101 may identify utterance 2 and pre-entered utterance 1 as user utterance 195 within a preset endpoint detection time (e.g., about 2,000 ms) according to the intention of subsequent utterance.

For example, the reset command 1330 may mean a command to ignore the utterance 195 before the reset command 1330 is input to the electronic device 101.

For example, in (c) of FIG. 13, when utterance 1, initialization command 1330, and utterance 2 are input, the electronic device 101 uses only utterance 2 excluding utterance 1 input before the initialization command 1330. , the endpoint detection time can be determined.

For example, in (c) of FIG. 13, the electronic device 101 may exclude utterance 1 input before inputting the initialization command 1330 from the subsequent voice recognition operation. For example, in (c) of FIG. 13, the automatic speech recognition module (e.g., the automatic speech recognition module 221 of FIG. 2) uses only the text corresponding to utterance 2 to the natural language understanding module (e.g., the natural language understanding module (e.g., the natural language understanding module of FIG. 2) 223)), the subsequent voice recognition operation can be performed.

For example, the automatic speech recognition module 221 may transmit the user utterance 195 to the natural language understanding module 223 based on the delay command 1320 or the initialization command 1330.

For example, in (b) of FIG. 13, the automatic voice recognition module 221 may transmit utterance 1 and utterance 2 to the natural language understanding module 223.

For example, in (c) of FIG. 13, the automatic speech recognition module 221 may not transmit utterance 1 input before the initialization command 1330 to the natural language understanding module 223. For example, the automatic speech recognition module 221 may transmit utterance 2 input after the initialization command 1330 to the natural understanding module 223.

An electronic device (e.g., electronic device 101 of FIG. 2) according to an embodiment is electrically connected to a processor (e.g., processor 120 of FIG. 2) and the processor 120, and is connected to the processor 120. It may include a memory (eg, memory 130 in FIG. 2) that stores instructions executed by the memory. The processor 120 may recognize an utterance (eg, user utterance 195 in FIG. 3) when the instruction is executed. The processor 120 sets an end point detection time to end recognition of the utterance 195 based on utterance data (e.g., utterance data 112 in FIG. 3) stored in the memory 130. time) can be determined. If no additional user utterance 195 is input within the endpoint detection time, the processor 120 may end recognition of the utterance 195. The utterance data 112 includes sentence data including a pattern related to the sentence type of the utterance 195 (e.g., sentence data 113 in FIG. 3) and word data including words of the utterance 195 ( For example, it may include word data 114 in FIG. 3).

The processor 120 may determine the category of the utterance. The processor 120 may determine the endpoint detection time using the sentence data 113 and the word data 114 corresponding to the category.

The processor 120 may use the utterance 195 and the utterance data 112 to determine whether the utterance 195 matches the pattern stored in the utterance data 112. The processor 120 may determine the endpoint detection time depending on whether the utterance 195 matches the utterance 195 pattern.

The processor 120 may use the utterance 195 and the utterance data 112 to determine whether the utterance 195 is similar to the pattern stored in the utterance data 112. The processor 120 may determine the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern are similar.

The processor 120 may determine the endpoint detection time based on the frequency of the pattern.

The processor 120 may use the utterance 195 and the utterance data 112 to determine whether the utterance 195 is a complete sentence. The processor 120 may determine the endpoint detection time depending on whether the utterance 195 is a complete sentence.

The processor 120 may use the utterance 195 and the utterance data 112 to determine whether the utterance 195 is an incomplete sentence. The processor 120 may determine the endpoint detection time depending on whether the utterance 195 is an incomplete sentence.

The processor 120 may identify whether the utterance 195 includes a preset EPD instruction (eg, the EPD instruction 115 of FIG. 3). The processor 120 may determine the endpoint detection time according to the EPD instruction 115.

The processor 120 may update the speech data 112 according to a preset update standard.

An electronic device (e.g., electronic device 101 of FIG. 2) according to an embodiment is electrically connected to a processor (e.g., processor 120 of FIG. 2) and the processor 120, and is connected to the processor 120. It may include a memory (eg, memory 130 in FIG. 2) that stores instructions executed by the memory. The processor 120 may determine the category of the utterance when the instruction is executed. The processor 120 determines the pattern of the utterance (e.g., the user utterance 195 of FIG. 3) based on the utterance data corresponding to the category (e.g., the utterance data 112 of FIG. 3) to the utterance data. You can compare patterns stored in (112). If the pattern of the utterance 195 is different from the pattern stored in the utterance data 112, the processor 120 may determine whether the utterance 195 is a complete sentence, an incomplete sentence, or an ambiguous sentence. The processor 120 recognizes the utterance 195 based on at least one of a comparison result between the utterance 195 and a pattern stored in the utterance data 112 and whether the utterance 195 is a complete sentence. The end point detection time for terminating can be determined. If no additional user utterance 195 is input within the endpoint detection time, the processor 120 may end recognition of the utterance 195. The utterance data 112 is sentence data including the pattern related to the sentence type of the utterance 195 (e.g., sentence data 113 in FIG. 3) and word data including words of the utterance 195. (e.g., word data 114 in FIG. 3).

The processor 120 may use the utterance 195 and the utterance data 112 to determine whether the utterance 195 matches the pattern stored in the utterance data 112. The processor 120 may determine the endpoint detection time depending on whether the utterance 195 matches the utterance 195 pattern.

The processor 120 may use the utterance 195 and the utterance data 112 to determine whether the utterance 195 is similar to the pattern stored in the utterance data 112. The processor 120 may determine the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern are similar.

The processor 120 may determine the endpoint detection time based on the frequency of the pattern.

A method for determining an endpoint detection time according to an embodiment includes an operation of recognizing an utterance (e.g., the user utterance 195 in FIG. 3) and utterance data stored in a memory (e.g., the memory 130 in FIG. 2) (e.g., the user utterance 195 in FIG. 3). An operation of determining an end point detection time for ending recognition of the utterance 195 based on the utterance data 112) and if no additional user utterance 195 is input within the end point detection time. If not, it includes an operation to end recognition of the utterance 195, and the utterance data 112 is sentence data (e.g., sentence data 113 in FIG. 3) including a pattern related to the sentence type of the utterance 195. )) and word data including words of the utterance 195 (e.g., word data 114 in FIG. 3).

The method for determining the endpoint detection time may further include determining the category of the utterance. The operation of determining the endpoint detection time may be performed by using the sentence data 113 and the word data 114 corresponding to the category.

The operation of determining the endpoint detection time includes using the speech 195 and the speech data 112 to determine whether the speech 195 matches the pattern stored in the speech data 112. may include. The operation of determining the endpoint detection time may include determining the endpoint detection time depending on whether the utterance 195 matches the utterance 195 pattern.

The operation of determining the endpoint detection time includes using the speech 195 and the speech data 112 to determine whether the speech 195 is similar to the pattern stored in the speech data 112. may include. The operation of determining the endpoint detection time may include determining the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern are similar.

The operation of determining the endpoint detection time may include the operation of determining whether the utterance 195 is a complete sentence using the utterance 195 and the utterance data 112. The operation of determining the endpoint detection time may include determining the endpoint detection time depending on whether the utterance 195 is a complete sentence.

The operation of determining the end point detection time may include the operation of determining whether the utterance 195 is an incomplete sentence using the utterance 195 and the utterance data 112. It may include determining the endpoint detection time depending on whether the utterance 195 is an incomplete sentence.

Electronic devices according to various embodiments disclosed in this document may be of various types. Electronic devices may include, for example, portable communication devices (e.g., smartphones), computer devices, portable multimedia devices, portable medical devices, cameras, wearable devices, or home appliances. Electronic devices according to embodiments of this document are not limited to the above-described devices.

The various embodiments of this document and the terms used herein are not intended to limit the technical features described in this document to specific embodiments, and should be understood to include various changes, equivalents, or replacements of the embodiments. In connection with the description of the drawings, similar reference numbers may be used for similar or related components. The singular form of a noun corresponding to an item may include one or more of the above items, unless the relevant context clearly indicates otherwise. In this document, “, “A and B”, “at least one of A and B”, “at least one of A or B”, “A, B and C”, “at least one of A, B and C” Each of the phrases "one", "and "at least one" of A, B, or C may include any one of the items listed together in that phrase, or any possible combination thereof." . Terms such as "" 1", "" 2", ""the first" and ""the second" may be used simply to distinguish one component from another and refer to those components in different aspects (e.g. : importance or order). Any (e.g. first) component is different from another (e.g. second) component with the terms “, “function,” “or “communication,” or these terms. When a component is referred to as “connected,” it means that any component can be connected to another component directly (e.g., wired), wirelessly, or through a third component. It means that you can.

As used in various embodiments of this document, the term “module” may include a unit implemented in hardware, software, or firmware, and is interchangeable with terms such as logic, logic block, component, or circuit, for example. It can be used negatively. A module may be an integrated part or a minimum unit of the parts or a part thereof that performs one or more functions. For example, according to one embodiment, the module may be implemented in the form of an application-specific integrated circuit (ASIC).

Various embodiments of the present document are software (e.g., software containing one or more instructions stored in a storage medium (e.g., internal memory or external memory) that can be read by a machine (e.g., electronic device 101). program). For example, a processor (e.g., processor 120) of a device (e.g., electronic device 101) calls at least one instruction among one or more instructions stored from a storage medium, and , it can be executed. This enables the device to be operated to perform at least one function according to the at least one instruction called. The one or more instructions can be executed by code generated by a compiler or an interpreter. A storage medium that can be read by a device may be provided in the form of a non-transitory storage medium, where 'non-transitory' refers to a device in which the storage medium is tangible. , it only means that it does not include signals (e.g. electromagnetic waves), and this term does not distinguish between cases where data is semi-permanently stored in a storage medium and cases where data is stored temporarily.

According to one embodiment, methods according to various embodiments disclosed in this document may be included and provided in a computer program product. Computer program products are commodities and can be traded between sellers and buyers. The computer program product may be distributed in the form of a machine-readable storage medium (e.g. compact disc read only memory (CD-ROM)) or through an application store (e.g. Play StoreTM) or on two user devices (e.g. It can be distributed (e.g. downloaded or uploaded) directly between smart phones) or online. In the case of online distribution, at least a portion of the computer program product may be at least temporarily stored or temporarily created in a machine-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

According to various embodiments, each component (e.g., module or program) of the above-described components may include a single or plural entity, and some of the plurality of entities may be separately placed in other components. there is. According to various embodiments, one or more of the components or operations described above may be omitted, or one or more other components or operations may be added. Alternatively or additionally, multiple components (eg, modules or programs) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each component of the plurality of components in the same or similar manner as those performed by the corresponding component of the plurality of components prior to the integration. . According to various embodiments, operations performed by a module, program, or other component may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, or omitted. Alternatively, one or more other operations may be added.

Claims (19)

In the electronic device 101,
processor 120; and
A memory 130 electrically connected to the processor 120 and storing instructions executed by the processor 120.
Including,
The processor 120,
When the command is executed, recognize utterance 195;
Based on the utterance data 112 stored in the memory 130, determine an end point detection time for ending recognition of the utterance 195;
If no additional user utterance 195 is input within the endpoint detection time, utterance 195 recognition is terminated,
The speech data 112 is,
Containing sentence data 113 containing a pattern related to the sentence type of the utterance 195 and word data 114 containing the words of the utterance 195,
Electronic device (101). According to paragraph 1,
The processor 120,
determine the category of the utterance;
Determining the endpoint detection time using the sentence data 113 and the word data 114 corresponding to the category,
Electronic device (101). According to any one of paragraphs 1 and 2,
The processor 120,
Using the utterance 195 and the utterance data 112, determine whether the utterance 195 substantially matches the pattern stored in the utterance data 112;
determining the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern substantially match;
Electronic device (101). According to any one of claims 1 to 3,
The processor 120,
Using the utterance 195 and the utterance data 112, determine whether the utterance 195 is similar to the pattern stored in the utterance data 112;
Determining the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern are similar,
Electronic device (101). According to any one of claims 1 to 4,
The processor 120,
Based on the frequency of the pattern, determining the endpoint detection time,
Electronic device (101). According to any one of claims 1 to 5,
The processor 120,
Using the utterance 195 and the utterance data 112, determine whether the utterance 195 is a complete sentence;
Depending on whether the utterance 195 is a complete sentence, determining the endpoint detection time,
Electronic device (101). According to any one of claims 1 to 6,
The processor 120,
Using the utterance 195 and the utterance data 112, determine whether the utterance 195 is an incomplete sentence;
Depending on whether the utterance 195 is an incomplete sentence, determining the endpoint detection time,
Electronic device (101). According to any one of claims 1 to 7,
The processor 120,
identify whether the utterance (195) includes a preset EPD command (115);
According to the EPD command 115, determining the endpoint detection time,
Electronic device (101). According to any one of claims 1 to 8,
The processor 120,
Updating the utterance data 112 according to preset update standards,
Electronic device (101). In the electronic device 101,
processor 120; and
A memory 130 electrically connected to the processor 120 and storing instructions executed by the processor 120.
Including,
The processor 120,
When the instruction is executed, determine the category of the utterance 195;
Based on the utterance data 112 corresponding to the category, the pattern of the utterance 195 is compared with the pattern stored in the utterance data 112;
If the pattern of the utterance 195 is different from the pattern stored in the utterance data 112, determine whether the utterance 195 is a complete sentence, an incomplete sentence, or an ambiguous sentence;
Recognition of the utterance 195 based on at least one of a comparison result between the utterance 195 and the pattern stored in the utterance data 112 and whether the utterance 195 is a complete sentence, an incomplete sentence, or an ambiguous sentence. determine an end point detection time for termination;
If no additional user utterance 195 is input within the endpoint detection time, utterance 195 recognition is terminated,
The speech data 112 is,
Sentence data 113 containing the pattern regarding the sentence type of the utterance 195 and word data 114 containing the words of the utterance 195,
Electronic device (101). According to clause 10,
The processor 120,
Using the utterance 195 and the utterance data 112, determine whether the utterance 195 substantially matches the pattern stored in the utterance data 112;
determining the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern substantially match;
Electronic device (101). According to any one of claims 10 and 11,
The processor 120,
Using the utterance 195 and the utterance data 112, determine whether the utterance 195 is similar to the pattern stored in the utterance data 112;
Determining the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern are similar,
Electronic device (101). According to any one of claims 10 to 13,
The processor 120,
Based on the frequency of the pattern, determining the endpoint detection time,
Electronic device (101). An operation to recognize an utterance (195);
An operation of determining an end point detection time for ending recognition of the utterance 195 based on the utterance data 112 stored in the memory 130; and
If no additional user utterance 195 is input within the endpoint detection time, an operation to end recognition of the utterance 195
Including,
The speech data 112 is,
Containing sentence data 113 containing a pattern related to the sentence type of the utterance 195 and word data 114 containing the words of the utterance 195,
How to determine endpoint detection time. According to clause 14,
Further comprising the operation of determining the category of the utterance,
The operation of determining the endpoint detection time is,
Determining the endpoint detection time using the sentence data 113 and the word data 114 corresponding to the category,
How to determine endpoint detection time. According to any one of claims 14 and 15,
The operation of determining the endpoint detection time is,
Using the utterance 195 and the utterance data 112, determining whether the utterance 195 substantially matches the pattern stored in the utterance data 112; and
An operation of determining the endpoint detection time depending on whether the utterance 195 and the utterance 195 pattern substantially match.
Including,
How to determine endpoint detection time. According to any one of claims 14 to 16,
The operation of determining the endpoint detection time is,
Using the utterance 195 and the utterance data 112, determining whether the utterance 195 is similar to the pattern stored in the utterance data 112; and
An operation of determining the end point detection time according to whether the utterance 195 and the utterance 195 pattern are similar.
Including,
How to determine endpoint detection time. According to any one of claims 14 to 17,
The operation of determining the endpoint detection time is,
Using the utterance 195 and the utterance data 112, determining whether the utterance 195 is a complete sentence; and
An operation of determining the endpoint detection time depending on whether the utterance 195 is a complete sentence.
Including,
How to determine endpoint detection time. According to any one of claims 14 to 18,
The operation of determining the endpoint detection time is,
Using the utterance 195 and the utterance data 112, determining whether the utterance 195 is an incomplete sentence; and
An operation of determining the endpoint detection time depending on whether the utterance 195 is an incomplete sentence.
Including,
How to determine endpoint detection time.

Images