Household appliance and voice recognition method thereof

Abstract

A household appliance is disclosed. The household appliance includes: a communication device configured to communicate with another household appliance; a microphone configured to receive a voice from a user; and a processor configured to perform signal processing on first voice data obtained from the microphone and perform voice recognition using the signal-processed first voice data, wherein the processor generates noise data using second voice data received from the other household appliance and performs signal processing on the first voice data using the generated noise data.

Description

Technical Field

[0001] Apparatuses and methods consistent with the present disclosure relate to a home appliance and a speech recognition method thereof, and more particularly, to a home appliance and a speech recognition method thereof that estimate noise using information about noise occurring in another home appliance and reduce the estimated noise to improve a speech recognition rate. Background Art

[0002] Home appliances can perform various functions according to control commands from users. Recently, home appliances adopt a voice recognition function to receive control commands through user voices, as well as receive control commands via input devices such as keyboards, remote controllers, etc.

[0003] In particular, home appliances that employ a method of activating a voice recognition function based on a specific start command (eg, Bixby) when a user speaks the specific start command have been increasingly adopted.

[0004] Recently, it is common for households to have and use a plurality of home appliances. Here, if a large amount of noise is generated in a home appliance, a voice recognition function of another home appliance may not be able to operate smoothly due to the noise. Summary of the invention

[0005] Technical issues

[0006] Embodiments of the present disclosure overcome the above disadvantages and other disadvantages not described above.In addition, the present disclosure is not required to overcome the disadvantages described above, and embodiments of the present disclosure may not overcome any of the problems described above.

[0007] The present disclosure provides a home appliance which estimates noise using information on noise generated in another home appliance and reduces the estimated noise to improve a voice recognition rate, and a voice recognition method thereof.

[0008] Solution to the problem

[0009] According to an embodiment of the present disclosure, a household appliance includes: a communication device configured to communicate with another household appliance; a microphone configured to receive voice from a user; and a processor configured to perform signal processing on first voice data obtained from the microphone and perform voice recognition using the signal-processed first voice data, wherein the processor uses second voice data received from the other household appliance to generate noise data and uses the generated noise data to perform signal processing on the first voice data.

[0010] According to another embodiment of the present disclosure, a method for voice recognition of a household appliance includes: obtaining first voice data from a microphone; performing signal processing on the obtained first voice data; and performing voice recognition using the signal-processed first voice data, wherein performing signal processing includes: receiving second voice data from another household appliance; generating noise data using the received second voice data; and performing signal processing on the first voice data using the generated noise data.

[0011] According to another embodiment of the present disclosure, a recording medium stores a program for executing a method for speech recognition, wherein the method includes: requesting speech data from a first home appliance and a second home appliance; receiving first speech data and second speech data obtained from the first home appliance and the second home appliance respectively based on the request; generating noise data using the received second speech data; performing signal processing on the first speech data using the generated noise data; and sending the signal-processed first speech data to the first home appliance.

[0012] Additional aspects and / or other aspects and advantages of the present disclosure will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above aspects and / or other aspects of the present disclosure will become more apparent by describing certain embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0014] Figure 1 is a block diagram of a speech recognition system according to an embodiment of the present disclosure;

[0015] Figure 2 is a block diagram illustrating a simple configuration of a first home appliance according to an embodiment of the present disclosure;

[0016] Figure 3 is a block diagram illustrating a specific configuration of a first home appliance according to an embodiment of the present disclosure;

[0017] Figure 4 is a block diagram illustrating a simple configuration of a second home appliance according to an embodiment of the present disclosure;

[0018] Figure 5 is a block diagram illustrating a simple configuration of a server according to an embodiment of the present disclosure;

[0019] Figure 6 is a diagram illustrating a noise removing method according to a first embodiment;

[0020] Figure 7 is a diagram illustrating a noise removing method according to a first embodiment;

[0021] Figure 8 is a diagram illustrating a noise removing method according to a second embodiment;

[0022] Fig. 9 is a diagram illustrating a noise removing method according to a second embodiment;

[0023] Fig.10 is a diagram illustrating a noise removing method according to a second embodiment;

[0024] Fig.11 is a diagram illustrating a noise removing method according to a third embodiment;

[0025] Fig.12 is a diagram illustrating a noise removing method according to a third embodiment;

[0026] Fig.13 is a diagram illustrating a noise removing method according to a third embodiment;

[0027] Fig.14 is a sequence diagram illustrating a speech recognition method according to the first embodiment;

[0028] Fig.15 is a sequence diagram illustrating a speech recognition method according to the first embodiment;

[0029] Fig.16 is a sequence diagram illustrating a speech recognition method according to the first embodiment;

[0030] Fig.17 is a flowchart illustrating a speech recognition method according to a second embodiment;

[0031] Fig.18 is a flowchart illustrating a speech recognition method according to a second embodiment;

[0032] Fig.19 is a flowchart illustrating a voice recognition method of a first home appliance according to an embodiment of the present disclosure;

[0033] Fig. 20 is a flowchart illustrating a voice recognition method of a second home appliance according to an embodiment of the present disclosure; and

[0034] Fig.21 is a flowchart illustrating a speech recognition method of a server according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0035] Before proceeding to the following detailed description, it may be advantageous to set forth definitions of certain words and phrases used throughout this patent document: the terms "include" and "comprise" and their derivatives are meant to include, but are not limited to; the term "or" is inclusive, meaning and / or; the phrases "associated with" and "associated with" and their derivatives may mean include, be included within, interconnected with, contain, be contained within, connect to or connect with, couple to or couple with, be communicative with, cooperate with, interleave, juxtapose, be proximate to, be bound to or bound with, have, have the characteristics of, etc.; and the term "controller" means any device, system, or portion thereof that controls at least one operation, such device may be implemented using hardware, firmware, or software, or some combination of at least two of them. It should be noted that the functionality associated with any particular controller may be centralized or distributed, whether locally or remotely.

[0036] Moreover, the various functions described below can be implemented or supported by one or more computer programs, each of which is formed by a computer-readable program code and embodied in a computer-readable medium. The terms "application" and "program" refer to one or more computer programs, software components, instruction sets, processes, functions, objects, categories, instances, related data or a part thereof that are suitable for implementation with a suitable computer-readable program code. The phrase "computer-readable program code" includes any type of computer code, including source code, object code and executable code. The phrase "computer-readable medium" includes any type of medium that can be accessed by a computer, such as a read-only memory (ROM), a random access memory (RAM), a hard drive, a compact disc (CD), a digital video disc (DVD) or any other type of memory. "Non-transitory" computer-readable medium excludes wired, wireless, optical or other communication links that transmit temporary electrical signals or other signals. Non-transitory computer-readable media include media that can permanently store data, and media that can store and overwrite data later, such as rewritable optical discs or erasable memory devices.

[0037] Definitions for certain words and phrases are provided throughout this patent document, those of ordinary skill in the art should understand that in many, if not most instances, such definitions apply to prior, as well as future uses of such defined words and phrases.

[0038] Discussed below Figures 1 to 21The various embodiments used to describe the principles of the present disclosure in this patent document are for illustration only and should not be interpreted in any way as limiting the scope of the present disclosure. Those skilled in the art will appreciate that the principles of the present disclosure can be implemented in any suitably arranged system or device.

[0039] Terms used in the description of various embodiments of the present disclosure are briefly described, and then various example embodiments of the present disclosure will be described in more detail.

[0040] The terms used in the example embodiments of the present disclosure are common terms that are now widely used and selected in consideration of the functions of the present disclosure. However, the terms may change depending on the intention of those skilled in the art, precedents, or the emergence of new technologies. In addition, in specified cases, the terms may be selected arbitrarily. In this case, the meaning of the terms will be explained in the corresponding description. Therefore, the terms used in the present disclosure may be defined based on the meaning of the terms and content described in the present disclosure, rather than simply defined based on the names of the terms.

[0041] Hereinafter, various embodiments of the present disclosure are described with reference to the accompanying drawings. However, it should be understood that the present disclosure is not limited to specific embodiments, and all modifications, equivalents and / or alternatives of the present disclosure also belong to the scope of the present disclosure. The description of the well-known functions and structures incorporated herein may be omitted to avoid confusing the subject matter of the present disclosure.

[0042] In the present disclosure, terms including ordinal numbers such as "first", "second", etc. may be used to describe various components, but the components should not be construed as being limited to these terms. The terms are only used to distinguish one component from other components.

[0043] In the description, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. It will be understood that the terms "comprising", "including", "having" and variations thereof specify the presence of the stated features, numbers, steps, operations, elements, components and / or groups thereof, but do not exclude the presence or addition of one or more other features, numbers, steps, operations, elements, components and / or groups thereof.

[0044] Hereinafter, the embodiments will be described in detail with reference to the accompanying tables and drawings so that those skilled in the art can easily practice these embodiments. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the present disclosure. In the accompanying drawings, for the sake of clarity, parts not related to the description of the present disclosure will be omitted. Similar reference numerals refer to similar elements throughout.

[0045] Hereinafter, the present disclosure will be described in detail with reference to the accompanying drawings.

[0046] Figure 1 is a diagram illustrating a speech recognition system according to an embodiment of the present disclosure.

[0047] refer to Figure 1 , the voice recognition system 1000 includes a first home appliance 100 , a second home appliance 200 , and a server 300 .

[0048] The first and second home appliances 100 and 200 are electrical machine tools for home use, and include a cleaning robot, a vacuum cleaner, an electric range, a gas range, an electric wave oven, a range hood, a washing machine, a dryer, a refrigerator, a dishwasher, an air conditioner, and the like.

[0049] The first home appliance 100 and the second home appliance 200 may perform a voice recognition function and perform functions according to the voice recognition result. Here, voice recognition refers to a technology that converts an acoustic signal of an input speech into a word or sentence.

[0050] In addition, the first home appliance 100 and the second home appliance 200 can detect the voice spoken by the user and perform voice recognition on the detected voice. Specifically, when a wake-up word (WUW) as a trigger voice command for activating the voice recognition function is detected, the first home appliance 100 and the second home appliance 200 activate the voice recognition function and perform voice recognition using the voice data of the voice input thereafter.

[0051] Hereinafter, for the sake of convenience, it is assumed that the first home appliance 100 performs a voice recognition function. Here, the second home appliance 200 may be a home appliance that generates a large amount of noise. Specifically, the second home appliance 200 may generate a large amount of noise due to a motor, a fan, etc. included therein.

[0052] In addition, the voice recognition function of the first home appliance 100 may not operate normally due to the large noise generated in the second home appliance 200. Therefore, the first home appliance 100 can estimate the noise using information about the large noise generated in the second home appliance 200 and reduce the estimated noise to improve the voice recognition rate.

[0053] To this end, the first home appliance 100 may transmit a recording request signal to the second home appliance 200. In addition, the first home appliance 100 may receive second voice data obtained based on the recording request signal from the second home appliance 200. Here, the second voice data is voice data about noise generated in the second home appliance 200.

[0054] In addition, the first home appliance 100 may obtain the first voice data based on the recording request signal.

[0055] Therefore, the first home appliance 100 can obtain the first voice data and the second voice data of the simultaneously detected voice.

[0056] In addition, the first home appliance 100 can generate noise data using the second voice data. Here, the noise data refers to sound data corresponding to the pure noise of the noise source. Specifically, the first home appliance 100 can generate noise data by applying the second voice data to the noise path estimation filter. Figures 6 to 10 Describe the specific operations for generating noisy data.

[0057] In addition, the first home appliance 100 can use the generated noise data to perform signal processing on the first voice data. Specifically, the first home appliance 100 can use the generated noise data to remove the noise included in the first voice data.

[0058] In addition, the first home appliance 100 may perform voice recognition using the first signal-processed voice data.

[0059] The second home appliance 200 may be connected to the first home appliance 100 through wired or wireless communication. The second home appliance 200 may receive a recording request signal from the first home appliance 100.

[0060] In addition, the second home appliance 200 may obtain second voice data based on the received recording request signal, and the obtained second voice data may be transmitted to the first home appliance 100.

[0061] At the same time, the recording request signal sending operation and the signal processing operation for the first voice data of the first home appliance 100 may also be performed by the server 300 .

[0062] Specifically, the server 300 may be connected to the first home appliance 100 and the second home appliance 200 through wired or wireless communication. The server 300 may transmit a recording request signal to the first home appliance 100 and the second home appliance 200.

[0063] In addition, the server 300 may receive first voice data obtained based on the recording request signal from the first home appliance 100. The server 300 may receive second voice data obtained based on the recording request signal from the second home appliance 200.

[0064] In addition, the server 300 may generate noise data using the second voice data. Specifically, the server 300 may generate noise data by applying the second voice data to a noise path estimation filter.

[0065] The server 300 may perform signal processing on the first voice data using the generated noise data. The server 300 may transmit the signal-processed first voice data to the first home appliance 100.

[0066] At the same time, Figure 1 In the embodiment, the first home appliance 100 is illustrated and described as being connected to one second home appliance 200, but it may also be implemented such that the first home appliance 100 is connected to a plurality of home appliances and signal processing is performed using information on noise generated in the plurality of home appliances.

[0067] In addition, Figure 1 In the embodiment, it is illustrated and described that the server 300 is connected to two home appliances, but it may also be implemented such that the server 300 is connected to three or more home appliances.

[0068] In addition, Figure 1 , it is illustrated and described that the server 300 is included in the voice recognition system 1000 , but only the first home appliance 100 and the second home appliance 200 may be implemented without the server 300 .

[0069] Figure 2 is a block diagram illustrating a simple configuration of a first home appliance according to an embodiment of the present disclosure.

[0070] refer to Figure 2 , the first home appliance 100 may include a microphone 110 , a communication device 120 , and a processor 130 .

[0071] The microphone 110 , a device that converts sound into a sound signal, can output the user's spoken voice and surrounding sounds as voice data. The microphone 110 can deliver the output voice data to the processor 130 .

[0072] The microphone 110 may be disposed on the surface of the housing. Specifically, the microphone 110 may be disposed on the surface of the housing to generate first voice data corresponding to the sound near the first home appliance 100 so as to collect the user's spoken voice. In addition, the arrangement position of the microphone 110 is not limited to the example described above.

[0073] In addition, in the present disclosure, it is illustrated and described that the first home appliance 100 includes one microphone, but in implementation, the first home appliance 100 may include two or more microphones.

[0074] The communication device 120 can be connected to an external device and can receive various data from the external device. Specifically, the communication device 120 can be connected to the external device through a local area network (LAN) and the Internet, and through a universal serial bus (USB) port or wireless communication (e.g., Wi-Fi 802.11a / b / g / n, NFC, Bluetooth). Here, the external device can be a PC, a notebook, a smart phone, a server, etc.

[0075] In addition, the communication device 120 can be connected to the second home appliance 200 or the server 300 in the voice recognition system 1000. For example, the communication device 120 can communicate with the second home appliance 200 in the voice recognition system 1000 through Bluetooth communication, and communicate with the server 300 in the voice recognition system 1000 through Wi-Fi communication. Meanwhile, the method for the first home appliance 100 to communicate with the second home appliance 200 or the server 300 is not limited to the example described above.

[0076] The processor 130 controls the first home appliance 100. Specifically, the processor 130 may control each component of the first home appliance 100 according to a control command of a user. For example, when the first home appliance 100 is a refrigerator, the processor 130 may control the operation of the compressor to provide refrigerated air to a refrigeration chamber containing food based on receiving a refrigeration command.

[0077] In addition, the processor 130 may perform voice recognition using the first voice data obtained through the microphone 110. Meanwhile, the first voice data may include noise as well as the user's voice, and a voice recognition rate may be reduced due to the included noise.

[0078] Therefore, the processor 130 may perform preprocessing on the first voice data. Here, preprocessing is a series of signal processing performed before voice recognition and may remove noise included in the voice data. In addition, the processor 130 may perform voice recognition using the preprocessed first voice data.

[0079] Meanwhile, the noise included in the first voice data may be caused by the second home appliance 200 that is disposed near the first home appliance 100 and generates a large amount of noise. Therefore, the processor 130 obtains information about the noise generated in the second home appliance 200 from the second home appliance 200 and pre-processes the first voice data using the obtained information about the noise.

[0080] Hereinafter, an operation of performing pre-processing on the first voice data using information on the noise generated in the second home appliance 200 will be described in detail.

[0081] First, the processor 130 may control the communication device 120 to detect another connectable home appliance. Specifically, the processor 130 may control the communication device 120 to detect a second home appliance connectable via short-range wireless communication or a second home appliance 200 connected to an access point (AP) to which the first home appliance 100 is connected.

[0082] Here, the second home appliance 200 available for short-range wireless communication with the first home appliance 100 or connected to the AP to which the first home appliance 100 is connected is expected to be set near the first home appliance 100. Therefore, it can be inferred that the noise generated in the second home appliance 200 can affect the voice recognition function of the first home appliance 100.

[0083] In addition, the processor 130 may control the communication device 120 to request the second home appliance 200 to send voice data. Specifically, the processor 130 may control the communication device 120 to send a recording request signal for requesting the generation and transmission of second voice data to the second home appliance 200, the second voice data being voice data of noise generated in the second home appliance 200.

[0084] Here, the recording request signal may include information about the scheduled time. The second home appliance 200 may detect noise using a microphone at the scheduled time based on the received recording request signal and generate the second voice data. For example, when the information about the scheduled time included in the recording request signal is 3:00:30 p.m., the second home appliance 200 may detect noise using a microphone at 3:00:30 p.m. and generate the second voice data.

[0085] Meanwhile, the information about the predetermined time may be implemented to include absolute time information as in the example described above, and may be implemented to include relative time information such as “x seconds after receiving the recording request signal”.

[0086] In addition, the recording request signal may include information about a predetermined time length of the second voice data. The second home appliance 200 may generate second voice data having a predetermined time length based on the received recording request signal. For example, when the information about the predetermined time length included in the recording request signal is 3 seconds, the second home appliance 200 may generate second voice data having a length of 3 seconds.

[0087] Meanwhile, the information that may be included in the recording request signal is not limited to the examples described above.

[0088] In addition, the processor 130 may receive second voice data generated based on the request sent to the second home appliance 200 from the second home appliance 200. Specifically, the processor 130 may receive second voice data generated based on the predetermined time information and the predetermined time length information included in the recording request signal from the second home appliance 200.

[0089] For example, when the scheduled time information included in the recording request signal is 3:00:30 pm and the scheduled time length information is 2 seconds, the processor 130 may receive second voice data regarding the voice detected by the microphone of the second home appliance 200 from 3:00:30 pm to 3:00:32 pm.

[0090] In addition, the processor 130 may obtain the first voice data based on the request sent to the second home appliance 200. Specifically, the processor 130 may obtain the first voice data based on the predetermined time information and the predetermined time length information included in the recording request signal sent to the second home appliance 200.

[0091] For example, when the predetermined time information included in the recording request signal is 3:00:30 pm and the predetermined time length information is 2 seconds, the processor 130 may obtain first voice data regarding a voice detected through the microphone 110 from 3:00:30 pm to 3:00:32 pm.

[0092] Here, the obtained first voice data and second voice data are generated based on the recording request signal, and thus correspond to voice data simultaneously detected by the first home appliance 100 and the second home appliance 200. The second voice data is voice data of noise directly detected in the second home appliance 200, and thus may include noise having an amplitude greater than that included in the first voice data.

[0093] Therefore, when the second voice data is used, noise included in the first voice data can be extracted more accurately than when only the first voice data is used.

[0094] In addition, the processor 130 may generate noise data, which is sound data corresponding to pure noise of a noise source in the second home appliance 200, using the second voice data.

[0095] The processor 130 may generate noise data using a noise path estimation filter. Here, the noise path estimation filter refers to a filter for filtering components other than the noise of the noise source included in the second voice data. Here, the noise path estimation filter may be referred to as a filtering algorithm.

[0096] For example, when the second voice data is input to the noise path estimation filter, the speech component of the user included in the second voice data may be filtered to output noise data. Figures 6 to 10 The specific operation of using a noise path estimation filter to generate noise data is described.

[0097] In addition, the processor 130 may use a pre-stored noise path estimation filter or receive a noise path estimation filter from an external device, and generate noise data using the received noise path estimation filter. Figure 11 to Figure 13 A specific operation of receiving a noise path estimation filter from the outside and generating noise data is described.

[0098] The processor 130 may preprocess the first voice data using the noise data. Specifically, the processor 130 may preprocess the first voice data by removing a component corresponding to noise of a noise source from the first voice data using the noise data generated via the noise path estimation filter.

[0099] The operation of the processor 130 may be repeatedly performed according to a predetermined cycle. Specifically, the processor 130 may perform the following sequence of operations according to a predetermined cycle: requesting voice data from the second home appliance 200; receiving second voice data obtained based on the request from the second home appliance 200; obtaining first voice data corresponding to the request; generating noise data using the received second voice data; and performing preprocessing on the first voice data using the generated noise data.

[0100] Meanwhile, the processor 130 may change a pre-processing method for the first voice data according to whether noise is currently generated in the second home appliance 200 .

[0101] For example, it is assumed that the sound generated in the second home appliance 200 is generated by a motor included in the second home appliance 200. In this case, the processor 130 may change the pre-processing method for the first voice data according to whether the motor of the second home appliance 200 is driven.

[0102] Specifically, the processor 130 may control the communication device 120 to determine whether the motor of the second home appliance 200 is driven. If the motor of the second home appliance 200 is driven, the amplitude of the noise generated in the second home appliance 200 is large, and therefore, the first voice data may be pre-processed using the second voice data according to the method described above.

[0103] At the same time, when the motor of the second home appliance 200 is not driven, there is no noise generated in the second home appliance 200 itself, so the first voice data can be used instead of the second voice data, and voice recognition can be performed immediately, or the first voice data can be pre-processed according to the relevant technical method.

[0104] Meanwhile, even when the noise source of the second home appliance 200 is a fan or the like instead of a motor, the processor 130 may change the pre-processing method for the first voice data according to whether the noise source is driven.

[0105] The processor 130 may perform voice recognition using the preprocessed first voice data.

[0106] At the same time, Figure 2 In the figure, it is illustrated and described that the first home appliance performs the operation of sending a recording request signal to the second home appliance 200, receiving the second voice data from the second home appliance 200, and performing preprocessing on the first voice data, but it can be implemented so that the server 300, instead of the first home appliance 100, performs the above-described operations, and the first home appliance 100 only receives the preprocessed first voice data from the server 300 and performs voice recognition. Figure 5 Describe its specific operation.

[0107] In addition, although only simple components configuring the first household appliance are illustrated and described above, various components may be further provided during implementation. Figure 3 Describe this.

[0108] Figure 3 is a block diagram illustrating a specific configuration of a first home appliance according to an embodiment of the present disclosure.

[0109] refer to Figure 3 , the first home appliance 100 according to an embodiment of the present disclosure may include a microphone 110 , a communication device 120 , a processor 130 , an input device 140 , a memory 150 , and a display 160 .

[0110] The microphone 110 and the communication device 120 perform the Figure 2 The microphone and communication device have the same functions, so the detailed description thereof will be omitted. Figure 2 The processor 130 has been described, and therefore, the description thereof will not be repeated. Figure 2 , and the following will only describe the Figure 3 Component-related content.

[0111] The input device 140 may include a plurality of function keys for the user to set or select various functions supported by the first home appliance 100. In this way, the user can input various control commands for the first home appliance 100. For example, when the first home appliance 100 is a washing machine, the user can input a spin-dry command of the washing machine through the input device 140.

[0112] The memory 150 stores various data for the overall operation of the first home appliance 100, such as a program for processing or controlling the processor 130. Specifically, the memory 150 may store a plurality of application programs running in the first home appliance 100 and data and instructions for operating the first home appliance 100.

[0113] The memory 150 can be accessed by the processor 130, and data reading / writing / modification / deletion / update can be performed by the processor 130. The memory 150 can be implemented not only as a storage medium in the first home appliance 100, but also as an external storage medium, a removable disk including a USB memory, a network server via a network, etc.

[0114] The memory 150 may store a noise path estimation filter required to generate noise data.

[0115] The display 160 may display various types of information provided by the first home appliance 100. Specifically, the display 160 may display an operating state of the first home appliance 100 or may display a user interface window for selecting functions and options selected by a user.

[0116] For example, when the first household appliance 100 is a washing machine, the display 160 may display that the washing machine is performing a spin-drying operation or display an interface window that allows the user to select how many minutes the spin-drying function will operate. Alternatively, the display 160 may display the result of performing the voice recognition function, or may display an interface window so that the user can change the setting of the voice recognition function.

[0117] In the related art, for preprocessing, a beamforming technique is used to remove noise by obtaining multiple speech signals including speech and noise through multiple microphones and separating the speech and noise using the characteristics that the speech and noise are transmitted in different directions and their spectra are different.

[0118] However, if the amplitude of the noise is greater than the amplitude of the speech in the actual environment, it is difficult to identify the expected directional difference or spectral difference, resulting in the problem that it is difficult to distinguish between speech and noise in the related art method. For example, when a large amplitude of noise is generated from other household appliances arranged nearby, it is difficult to fully remove the noise for speech recognition only by the related art.

[0119] At the same time, as described above, according to the present embodiment, the first household appliance obtains second voice data about the corresponding noise from another household appliance that generates noise of large amplitude, and uses the obtained second voice data to perform preprocessing, thereby accurately removing the noise included in the voice data even if the amplitude of the noise generated from the outside is large.

[0120] Figure 4 is a block diagram illustrating a simple configuration of a second home appliance according to an embodiment of the present disclosure.

[0121] refer to Figure 4 According to an embodiment of the present disclosure, the second home appliance 200 may include a microphone 210, a communication device 220, a processor 240, a motor 230, an accelerometer 250, an input device 260, and a memory 270.

[0122] The microphone 210 is a device that converts sound into a sound signal and can output the user's spoken voice and surrounding sounds as voice data. The microphone 210 can deliver the output voice data to the processor 240.

[0123] The microphone 210 may be disposed on the surface of the housing. Specifically, in order to collect the user's spoken voice, the microphone 210 may be disposed on the surface of the housing to generate second voice data corresponding to the sound around the second home appliance 200.

[0124] Alternatively, the microphone 210 may be disposed in the housing. Specifically, in order to collect the noise sound generated by the second home appliance 200 itself, the microphone 210 may be disposed inside the housing (specifically, near a noise source (such as a motor) generating the noise) to generate the second voice data corresponding to the sound generated by the second home appliance 200.

[0125] Meanwhile, the arrangement position of the microphone 210 is not limited to the example described above. In addition, in the present disclosure, although it is illustrated and described that the second home appliance 200 includes one microphone, when implemented, the second home appliance 200 may include two or more microphones.

[0126] The communication device 220 can be connected to an external device and receive various data from the external device. Specifically, the communication device 220 can be connected to an external device through a local area network (LAN) and the Internet, and through a universal serial bus (USB) port or a wireless communication (Wi-Fi 802.11a / b / g / n, NFC, Bluetooth) port. The external device can be a PC, a notebook, a smart phone, a server, etc.

[0127] In addition, the communication device 220 can be connected to the first home appliance 100 or the server 300 in the voice recognition system 1000. For example, the communication device 220 can communicate with the first home appliance 100 in the voice recognition system 1000 through Bluetooth communication, and communicate with the server 300 in the voice recognition system 1000 through Wi-Fi communication. Meanwhile, the method for the second home appliance 200 to communicate with the first home appliance 100 or the server 300 is not limited to the example described above.

[0128] The motor 230 is provided in the second home appliance 200 to drive components related to performing functions of the second home appliance 200. For example, when the second home appliance 200 is a washing machine, the motor 230 can rotate the drum containing clothes at a high speed to spin the clothes. Here, vibration and noise may occur when the motor 230 starts the drum.

[0129] As another example, when the second home appliance 200 is a refrigerator, the motor 230 may start a refrigerant compressor that generates refrigerant. Here, vibration and noise may be generated when the motor 230 starts the refrigerant compressor.

[0130] As described above, when the second home appliance 200 itself generates a large amount of noise, the motor 230 may be a noise source. Therefore, the microphone 210 may be disposed near the motor 230 and detect the sound generated by the motor 230 and generate second voice data corresponding to the sound.

[0131] The processor 240 controls the second home appliance 200. Specifically, the processor 240 may control each component of the second home appliance 200 according to a control command of the user. For example, when the second home appliance 200 is a washing machine, the processor 240 may control the operation of the motor to provide a rotational force to the drum containing the clothes when a spin-drying command is received.

[0132] The processor 240 may receive a request to send the second voice data. Specifically, the processor 240 may receive a recording request signal requesting to generate and send the second voice data from the first home appliance 100 or the server 300 connected to the second home appliance 200, the second voice data being voice data about the noise generated in the second home appliance 200.

[0133] In addition, the processor 240 may obtain the second voice data based on the received request. Specifically, the processor 240 may detect noise using the microphone 210 based on the received recording request signal and obtain the second voice data. More specifically, the processor 240 may obtain the second voice data based on the information about the predetermined time and the predetermined time length information included in the received recording request signal.

[0134] In addition, the processor 240 may control the communication device 220 to transmit the obtained second voice data. Specifically, the processor 240 may transmit the obtained second voice data to the first home appliance 100 connected to the second home appliance 200 or the server 300.

[0135] As such, the second voice data transmitted to the first home appliance 100 or the server 300 is voice data about noise generated in the second home appliance 200 , and therefore, can be used to perform preprocessing to remove the noise included in the first voice data obtained in the first home appliance 100 .

[0136] For example, when the noise source of the second home appliance 200 is the motor 230, the first home appliance 100 or the server 300 can generate noise data (the noise data is pure motor noise data about the motor of the second home appliance 200) and use the generated motor noise data to perform preprocessing to remove the motor noise from the first voice data.

[0137] At the same time, as described above, in addition to the method in which the processor 240 obtains the second voice data and sends the obtained second voice data each time a recording request signal is received, the processor 240 can also obtain the second voice data in each predetermined period when a recording request signal is received, and send the obtained second voice data as described above.

[0138] Meanwhile, in addition to the second voice data generated by detecting the noise generated in the second home appliance 200, the processor 240 may also provide other types of reference data about the noise to the first home appliance 100 or the server 300.

[0139] Here, the reference data may be data including information of a noise source. The information of the noise source may include the amplitude and phase of vibration generated from the noise source, the amplitude and phase of noise generated from the noise source, main frequency information, and the like.

[0140] For example, the processor 240 may generate noise data by obtaining reference data through the accelerometer 250 or by obtaining reference data through a control command input through the input device 260. The specific operation thereof will be described below.

[0141] The accelerometer 250 is a device for measuring the acceleration of an object. In addition, the accelerometer 250 may be disposed near the motor 230 to measure the acceleration of the motor 230 and generate information about the measured acceleration.

[0142] The processor 240 may extract the operating frequency of the motor 230 from the obtained acceleration information and generate reference data using the extracted operating frequency.

[0143] For example, when 50 Hz is extracted as the operating frequency of the motor 230 from the acceleration information obtained from the accelerometer 250 , the processor 240 may generate reference data represented by a trigonometric function having a specific magnitude and phase using 50 Hz.

[0144] The input device 260 may include a plurality of function keys that allow the user to set or select various functions supported in the second home appliance 200. In this way, the user can input various control commands for the second home appliance 200.

[0145] Meanwhile, the control command input through the input device 260 may be related to the driving of the motor 230. In this case, the operating frequency of the motor 230 corresponding to the control command input through the input device 260 may be checked.

[0146] For example, when the second home appliance 200 is a washing machine and a spin-drying command is input, the motor 230 may rotate the drum of the washing machine to perform a spin-drying function. In this case, it may be checked that the operating frequency of the motor 230 corresponding to the spin-drying command is 50 Hz.

[0147] The processor 240 may generate reference data using the operating frequency of the motor 230 corresponding to the control command.

[0148] This may apply in the same manner to control commands generated by the processor 240 itself as determined by the situation.

[0149] In this way, the processor 240 can transmit at least one of the second voice data obtained by the microphone 210 and the reference data generated using the driving frequency identified from the control command or the acceleration information of the accelerometer 250 to the first home appliance 100 or the server 300. In addition, the first home appliance 100 or the server 300 can generate noise data using the received reference data, which is sound data corresponding to the pure noise of the noise source in the second home appliance 200.

[0150] Meanwhile, the processor 240 may not transmit the second voice data and the reference data to the first home appliance 100 or the server 300, and instead, the processor 240 may generate noise data using the second voice data and the reference data and transmit the generated noise data to the first home appliance 100 or the server 300. To this end, the noise path estimation filter may be pre-stored in the second home appliance 200.

[0151] In this case, the first home appliance 100 or the server 300 may receive the noise data and perform preprocessing using the received noise data to remove the noise included in the first voice data.

[0152] The memory 270 stores various data for the overall operation of the second home appliance 200, such as a program for processing or controlling the processor 240. Specifically, the memory 270 may store a plurality of application programs driven in the second home appliance 200 and data and instructions for the operation of the second home appliance 200.

[0153] For example, the memory 270 may store driving frequency information of the motor 230 corresponding to a control command input through the input device 260. The processor 240 may recognize a driving frequency corresponding to the input control command and generate reference data using the recognized driving frequency.

[0154] At the same time, despite Figure 4 The noise source is illustrated and described as the motor 230 in FIG. 2 , but in implementation, the noise source may be a fan or another component instead of the motor 230 .

[0155] In addition, Figure 4 In the present invention, a single motor is illustrated and described, but in implementation, a plurality of motors may be provided, and a plurality of motors may be provided and reference data of each of the plurality of motors may be used to estimate noise.

[0156] In addition, Figure 4 In the embodiment, the second household appliance includes both a microphone and an accelerometer, but only an accelerometer may be provided in the implementation, and the reference data may be obtained through the accelerometer. Alternatively, only a microphone may be provided, and the second voice data may be obtained through the microphone. Alternatively, the second microphone and the accelerometer may not be provided, and the reference data may be obtained through the input device.

[0157] In addition, Figure 4 In the embodiment, the operation of generating reference data using the operating frequency of the motor is illustrated and described as being performed by a processor, but in implementation, a sine wave signal generator that generates a sine wave signal when receiving operating frequency information may be provided. In this case, the signal generated by the sine wave signal generator may be used as reference data.

[0158] In addition, Figure 4 In the embodiment, the second home appliance 200 is illustrated and described as a home appliance having a configuration different from that of the first home appliance 100 , but in implementation, the second home appliance 200 may be a home appliance having the same configuration as the first home appliance 100 .

[0159] As described above, the second home appliance generates reference data about noise of an internal noise source and provides the generated reference data to the first home appliance or the server, so that the first home appliance or the server can perform preprocessing by using the reference data to remove the noise included in the voice data (which includes the user's speech).

[0160] Figure 5 is a block diagram illustrating a simple configuration of a server according to an embodiment of the present disclosure.

[0161] refer to Figure 5 , the server 300 may include a communication device 310 , a memory 320 , and a processor 330 .

[0162] The server 300 refers to a computer or a program that provides information or services to clients via a network.

[0163] The communication device 310 can be connected to an external device and can receive various data from the external device. Specifically, the communication device 310 can be connected to the external device through a local area network (LAN) and the Internet, and through a universal serial bus (USB) port or a wireless communication (Wi-Fi 802.11a / b / g / n, NFC, Bluetooth) port. The external device can be a PC, a notebook, a smart phone, a server, etc.

[0164] In addition, the communication device 310 can be connected to the first home appliance 100 or the second home appliance 200 in the voice recognition system 1000. For example, the communication device 310 can perform communication with the first home appliance 100 or the second home appliance 200 in the voice recognition system 1000 through Wi-Fi communication. Meanwhile, the method for the server 300 to communicate with the first home appliance 100 or the second home appliance 200 is not limited to the example described above.

[0165] The memory 320 stores various data for the overall operation of the server 300, such as a program for processing or controlling the processor 330. Specifically, the memory 320 may store a plurality of application programs running in the server 300 and data and instructions for the operation of the server 300.

[0166] The memory 320 can be accessed by the processor 330, and data reading / writing / modification / deletion / update can be performed by the processor 330. The memory 320 can be implemented not only as a storage medium in the server 300, but also as an external storage medium, a removable disk including a USB memory, a network server via a network, etc.

[0167] The memory 320 may store device information of a plurality of home appliances. Specifically, the memory 320 may store device information of a plurality of home appliances connected to the server 300 or a plurality of home appliances having a connection history to the server. The device information of the home appliance may include serial number or universal unique identifier (UUID) information, and the included information is not limited to the examples described above.

[0168] The memory 320 may store a noise path estimation filter required to generate noise data. Alternatively, the memory 320 may store an artificial intelligence model for generating a noise path estimation filter. Fig.11 Describes the specific operations of using an artificial intelligence model to generate a noisy path estimation filter.

[0169] The processor 330 controls the server 300. Specifically, the processor 330 may control each component of the server 300 according to a control command of a user.

[0170] In addition, instead of the first home appliance 100, the processor 330 may pre-process the voice data obtained by the first home appliance 100. Specifically, the processor 330 may receive first voice data corresponding to a sound near the first home appliance 100 from the first home appliance 100. In addition, the received first voice data may be pre-processed.

[0171] Meanwhile, when the second home appliance 200 is disposed near the first home appliance 100 and generates a large amount of noise, the processor 330 may perform pre-processing on the received first voice data using information about the noise generated in the second home appliance 200 .

[0172] Hereinafter, a specific operation of performing pre-processing on the first voice data using information on the noise generated in the second home appliance 200 will be described.

[0173] First, the processor 330 may receive information about a second home appliance 200 connected to the first home appliance 100 from the first home appliance 100. Specifically, the processor 330 may receive information about a second home appliance 200 that may be connected to the first home appliance 100 through short-range wireless communication or a second home appliance 200 connected to an access point (AP) to which the first home appliance 100 is connected from the first home appliance 100.

[0174] The processor 330 may be connected to the second home appliance 200. Specifically, the processor 330 may be connected to the second home appliance 200 using information of the second home appliance 200 received from the first home appliance 100.

[0175] The processor 330 may control the communication device 310 to request voice data from the first home appliance 100 and the second home appliance 200. Specifically, the processor 330 may control the communication device 310 to send a recording request signal for requesting generation and transmission of voice data to the first home appliance 100 and the second home appliance 200.

[0176] The recording request signal may include information about the predetermined time and information about the predetermined time length. The information included in the recording request signal is not limited to the examples described above.

[0177] The processor 330 may receive voice data generated based on the request. Specifically, the processor 330 may receive first voice data obtained based on a recording request signal from the first home appliance 100. The processor 330 may receive second voice data obtained based on a recording request signal from the second home appliance 200.

[0178] The processor 330 may generate noise data, which is sound data corresponding to pure noise of a noise source in the second home appliance 200, using the received second voice data.

[0179] In addition, in addition to the second voice data generated by detecting the noise generated in the second home appliance 200, the processor 330 may also receive other types of reference data about the noise.

[0180] For example, the processor 330 may receive reference data obtained by an accelerometer of the second home appliance 200 or by a control command input to the second home appliance 200. Here, the reference data may be generated based on noise source information including the amplitude and phase of vibration generated from the noise source, the amplitude and phase of noise generated from the noise source, and main frequency information.

[0181] The processor 330 may generate noise data, which is sound data corresponding to pure noise of a noise source in the second home appliance 200, using the received reference data.

[0182] The processor 330 may use a noise path estimation filter to generate noise data. Figures 6 to 10 The specific operation of using a noise path estimation filter to generate noise data is described.

[0183] The processor 330 may also be implemented to receive its own noise data generated in the second home appliance 200 without receiving the second voice data or the reference data from the second home appliance 200 .

[0184] The processor 330 may perform preprocessing on the first voice data using the noise data. Specifically, the processor 330 may perform preprocessing on the first voice data by removing a component corresponding to noise of a noise source from the first voice data using the noise data generated via the noise path estimation filter.

[0185] In addition, the processor 330 may transmit the preprocessed first voice data to the first home appliance 100. In addition, the first home appliance 100 may perform voice recognition using the preprocessed first voice data received from the server 300.

[0186] In this way, the server 300 receives first voice data from a first home appliance that will perform a voice recognition function, obtains second voice data or reference data about noise from a second home appliance that generates noise of a large amplitude, and performs preprocessing on the first voice data using the obtained second voice data or reference data, thereby accurately removing the noise included in the voice data even if the amplitude of the noise is large.

[0187] Figure 6 and Figure 7 is a diagram illustrating a noise removing method according to the first embodiment.

[0188] Figure 6 is a simple block diagram illustrating a noise removing method according to the first embodiment.

[0189] refer to Figure 6 , noise data y is generated using reference data r, speech data d, and a noise estimation algorithm, and noise of the speech data d is removed using the generated noise data y.

[0190] Here, the reference data r may be at least one of second voice data obtained through the microphone 210 of the second home appliance 200, acceleration information of the accelerometer 250 of the second home appliance 200, or reference data generated using a driving frequency checked from a control command.

[0191] The voice data d may correspond to the first voice data obtained by the microphone 110 of the first home appliance 100 .

[0192] The processor 130 of the first home appliance 100 or the processor 330 of the server 300 can generate noise data y using the noise estimation algorithm and the reference data r. Specifically, the processors 130 and 330 can use the noise path estimation filter information included in the noise estimation algorithm to extract noise data y from the reference data r, which is sound data corresponding to the noise of the source noise.

[0193] The noise path estimation filter may be implemented as a finite impulse response (FIR) filter or an infinite impulse response (IIR) filter in the time domain. Alternatively, the noise path estimation filter may be implemented in the form of a transfer function predetermined for each frequency band in the frequency domain.

[0194] The noise path estimation filter may have a linear structure as in the examples described above, but is not limited thereto, and may have a nonlinear structure.

[0195] The noise path estimation filter may be fixed and used as one noise path estimation filter, and in the case where a plurality of noise path estimation filter information is stored in advance, one of the noise path estimation filters may be selected and used according to circumstances to generate the noise data y.

[0196] The processors 130 and 330 may perform pre-processing on the voice data d by removing a component corresponding to noise of a noise source included in the voice data d using the generated noise data y.

[0197] The processor 330 of the server 300 may perform preprocessing and transmit the voice data e from which noise is removed to the first home appliance 100. In addition, the processor 130 of the first home appliance 100 may perform voice recognition using the voice data e from which noise is removed by performing preprocessing.

[0198] Meanwhile, the processor 130 or 330 may update a method of generating noise data y using the voice data d so that accurate noise data y may be generated even when noise of a noise source changes or a surrounding environment changes.

[0199] Specifically, after performing preprocessing, the processors 130 and 330 may use the speech data d including noise to update the noise path estimation filter. Figure 7 Describe its details.

[0200] Figure 7 is a block diagram illustrating a method of removing noise in the frequency domain according to the first embodiment.

[0201] refer to Figure 7 , it can be identified that a noise estimation algorithm using a noise path estimation filter H in the frequency domain and performing an update using speech data d including noise.

[0202] The processors 130 and 330 may use a fast Fourier transform (FFT) to convert the reference data r into the frequency domain. The processors 130 and 330 may obtain the noise data Y (Y=R·H) converted into the frequency domain by applying the converted reference data R to the noise path estimation filter H. In addition, the processors 130 and 330 may use an inverse fast Fourier transform (IFFT) to convert the noise data Y into the time domain. The processors 130 and 330 may use the converted noise data y to remove the noise of the voice data d.

[0203] Meanwhile, after performing the preprocessing, the processors 130 and 330 may use the speech data d mixed with noise to update the noise path estimation filter H. Specifically, the processor 130 or 330 may use the correlation between the speech data D converted into the frequency domain from the speech data d including noise by FFT and the converted reference data R to update the noise path estimation filter H.

[0204] For example, we can assume that the k-1th noise path estimation filter is H (k-1) , including the subsequent k-th noise speech data is d (k) , the kth reference data is r (k) , the speech data converted to the kth frequency domain is D (k) , and the reference data converted to the kth frequency domain is R (k) In this case, the k-th converted reference data R (k) The autocorrelation value of G RR(k) The inverse matrix of the k-th reference data R (k) With voice data D (k) The cross-correlation value of G DR(k) Multiply to generate the new noise path estimation filter H (k) (H (k) =G RR(k) ·G DR(k) (-1) ).

[0205] Here, G RR(k) yes (R (k) H is the reference data (R (k) ) and

[0206] In addition, as the λ value, a constant determined according to the system may be used, or for the stability of the algorithm, the λ value may be used variably. For example, when used in a variable manner, the kth λ value may be λ (k) =α·(e f(k) / (1+e f(k) ).also, M(k)=(1-γ)·M(k-1)+γ·D (k) , (α, β and γ are constants).

[0207] When the k+1th reference data (r (k+1) ) and voice data (d (k+1) ), the processors 130 and 330 use the updated new noise path estimation filter H (k) Generate noise data y from the next reference data (k+1) , and use the generated noise data y (k+1) From the voice data (k+1) Remove noise.

[0208] Meanwhile, the method of updating the noise path estimation filter H is not limited to the example described above. For example, in the example described above, the update is performed whenever the speech data and reference data including noise are obtained, but it can also be implemented so that the update is performed when the speech data and reference data including a predetermined amount of noise are obtained. In this case, the update can be performed using the speech data and reference data including a predetermined amount of noise together.

[0209] exist Figure 6 and Figure 7 , it is illustrated and described that the processors 130 and 330 update the noise path estimation filter after performing preprocessing on the speech data d including noise, but it can also be implemented so that the noise path estimation filter can be updated first, and then the preprocessing can be performed on the speech data d including noise.

[0210] Figures 8 to 10 is a diagram illustrating a noise removing method according to the second embodiment.

[0211] Figure 8 is a simple block diagram illustrating a noise removing method according to the second embodiment.

[0212] refer to Figure 8 , it can be seen that the noise data y is generated using the reference data r, the noise-removed speech data e, and the noise estimation algorithm, and the generated noise data y is used to remove the noise of the speech data d.

[0213] Here, the reference data r may be at least one of the second voice data obtained by the microphone 210 of the second home appliance 200 and the reference data generated using the driving frequency recognized from the control command, or the acceleration information of the accelerometer 250 of the second home appliance 200 .

[0214] The voice data d may correspond to the first voice data obtained from the microphone 110 of the first home appliance 100 .

[0215] The processors 130 and 330 can extract noise data y, which is sound data corresponding to the noise of the noise source, from the reference data r using the noise path estimation filter information included in the noise estimation algorithm. Here, the configuration of the noise estimation filter is similar to Figure 7 The configuration is the same as that of , so its detailed description will be omitted.

[0216] The processors 130 and 330 may perform pre-processing on the voice data d by removing a component corresponding to noise of a noise source included in the voice data d using the generated noise data y.

[0217] The processor 330 of the server 300 may perform preprocessing to transmit the noise-removed voice data e to the first home appliance 100. In addition, the processor 130 of the first home appliance 100 may perform voice recognition using the noise-removed voice data e by performing preprocessing.

[0218] Figure 8 The second embodiment of Figure 6 The difference between the first embodiment and the second embodiment is the method of updating the noise path estimation filter. Specifically, in the second embodiment, the processors 130 and 330 may update the method of generating noise data using the speech data without noise instead of the speech data including noise. Fig. 9 and Fig.10 Describe its details.

[0219] Fig. 9 is a block diagram illustrating a method of removing noise in the time domain according to the second embodiment.

[0220] refer to Fig. 9 , it can be identified that a noise estimation algorithm using a noise path estimation filter in the time domain (which is a FIR filter) and performing an update using the de-noised speech data.

[0221] The processors 130 and 330 may obtain noise data y (y=h*r) by applying the reference data r to the noise path estimation filter h in the time domain. In addition, the processors 130 and 330 may obtain de-noised speech data e by de-noising the speech data d using the noise data y.

[0222] Meanwhile, the processors 130 and 330 may update the noise path estimation filter h using the de-noised speech data e. Specifically, after performing pre-processing, the processors 130 and 330 may update the noise path estimation filter h using the correlation between the de-noised speech data e and the reference data r.

[0223] For example, we can assume that the k-1th noise path estimation filter is h (k-1), the kth denoised speech data is e (k) , and the kth reference data is r (k) In this case, the kth reference data r (k) Store to buffer r (k) , whose length is similar to the noise path estimation filter h (k-1) The length of the buffer is the same as that of (k) and the kth denoised speech data e (k) To generate a new noise path estimation filter h (k) (h (k) =h (k-1) +μ·e (k) ·r (k) ).

[0224] Here, as μ, a constant determined according to the system may be used, or μ may be used variably for the stability of the algorithm. For example, when used in a variably manner, μ may be μ (k) =α / P(k). In addition, P(k)=(1-γ)·P(k-1)+γ·r(k) T ·r(k) (α and γ are constants).

[0225] When the processors 130 and 330 obtain the next reference data r' and speech data d', the processors 130 and 330 generate noise data y' from the next reference data using the updated new noise path estimation filter h' and remove noise from the speech data d' using the noise data y', thereby obtaining noise-removed speech data e'.

[0226] Meanwhile, the method of updating the noise path estimation filter h in the time domain is not limited to the example described above. For example, in the example described above, the update is performed whenever the speech data and reference data from which the noise is removed are obtained, but it can also be implemented so that the update is performed when the speech data and reference data from which a predetermined amount of noise is removed are obtained. In this case, the update can be performed together using the speech data and reference data from which a predetermined amount of noise is removed.

[0227] Meanwhile, the noise path estimation filter h may be a filter implemented in the form of a predetermined transfer function for each frequency band in the frequency domain, rather than an FIR filter in the time domain, and the noise path estimation filter h may be updated using the de-noised speech signal e. Fig.10 Describe its details.

[0228] Fig.10 is a block diagram illustrating a method of removing noise in the frequency domain according to the second embodiment.

[0229] refer to Fig.10 , a noise estimation algorithm using a noise path estimation filter in the frequency domain and performing an update using the denoised speech data can be identified.

[0230] The processors 130 and 330 may use FFT to convert the reference data r into the frequency domain. The processors 130 and 330 may obtain the noise data Y (Y=R·H) converted into the frequency domain by applying the converted reference data R to the noise path estimation filter H. In addition, the processors 130 and 330 may use IFFT to convert the noise data Y into the time domain. The processor 130 may use the converted noise data y to remove the noise of the speech data d.

[0231] At the same time, the processors 130 and 330 can update the noise path estimation filter H in the frequency domain using the de-noised speech data e. Specifically, the processors 130 and 330 can update the noise path estimation filter H using the correlation between the speech data E converted from the speech data e to the frequency domain by FFT and the converted reference data R.

[0232] For example, we can assume that the k-1th noise path estimation filter is H (k-1) , the k-th transformed speech data with noise removed is E (k) , and the kth reference data converted to the frequency domain is R (k) Here, the following components can be added to the k-1th noise path estimation filter H (k-1) To generate a new noise path estimation filter H (k) (H (k) =H (k-1) +μ·R (k) H ·E (k) ): This component is obtained by transforming the kth reference data R into the frequency domain (k) The Hermitian matrix and the k-th de-noised speech data E (k) Multiply and obtain.

[0233] Here, as μ, a constant determined according to the system may be used, or μ may be used variably for the stability of the algorithm. For example, in the case where μ is used variably, the kth μ may be μ(k)=α·(e f(k) / (1+e f(k) ). In addition, it is possible to M(k)=(1-γ)·M(k-1)+γ·D (k) , (α, β and γ are constants).

[0234] When the next reference data r is obtained (k+1) and voice data (k+1)When the processors 130 and 330 can use the updated new noise path estimation filter H (k) Generate noise data y from the next reference data (k+1) , and the generated noise data y can be used (k+1) From the voice data (k+1) Remove noise from the .

[0235] Meanwhile, the method of updating the noise path estimation filter H is not limited to the example described above. For example, in the example described above, the update is performed whenever the speech data and reference data from which noise is removed are obtained, but the update may also be performed when the speech data and reference data from which a predetermined amount of noise is removed are obtained. In this case, the update may be performed using the speech data and reference data from which a predetermined amount of noise is removed together.

[0236] In addition, Figures 8 to 10 , it is illustrated and described that the processors 130 and 330 update the noise path estimation filter after performing preprocessing on the speech data d including noise, but it may also be implemented such that the noise path estimation filter is first updated and preprocessing is performed on the speech data d including noise.

[0237] Figure 11 to Figure 13 is a diagram illustrating a noise removing method according to the third embodiment.

[0238] Information about the noise path estimation filter may be pre-stored in the first home appliance 100 or the server 300, and the noise data as described above may be generated using the information. However, the noise path estimation filter may not be pre-stored at the manufacturing stage of the first home appliance 100, but may be generated by an artificial intelligence model after the first home appliance 100 is installed in the home. Alternatively, the server 300 may generate a noise path estimation filter suitable for a home appliance connected by an artificial intelligence model.

[0239] In addition, the first home appliance 100 or the server 300 may perform noise removal using the generated noise path estimation filter.

[0240] In the following description, it is assumed that the first home appliance 100 obtains a noise path estimation filter from the external device 400 including an artificial intelligence model.

[0241] Fig.11 is a sequence diagram illustrating a method of obtaining a noise path estimation filter by the first home appliance 100.

[0242] refer to Fig.11, first, the first home appliance 100 can obtain voice data and reference data at a time when there is no user's speech (S1110). Specifically, in order to generate a noise path estimation filter, voice data and reference data in which there is no user's spoken voice and only noise of a noise source is detected are required. Therefore, when voice data determined to be without the user's speech is determined as a result of performing voice recognition, reference data obtained at the same time point as the time point of the corresponding voice data can be checked.

[0243] In operation S1120, the first home appliance 100 may transmit the obtained voice data and reference data to the external device 400. Specifically, the first home appliance 100 may transmit the obtained voice data and reference data as signals in the time domain, or convert them into the frequency domain and transmit them to the external device 400.

[0244] Meanwhile, when the first home appliance 100 has only voice data at a time when there is no user's speech, the external device 400 or the first home appliance 100 may request reference data from the second home appliance 200 at a corresponding time and obtain the reference data.

[0245] In addition, the noise of the noise source of the second home appliance 200 may be different according to the change of the operating mode or environment of the second home appliance 200. Therefore, it is necessary to generate a noise path estimation filter suitable for each case. To this end, when sending information to the external device 400, the first home appliance 100 can send information about the operating mode of the second home appliance 200 or information about the surrounding environment together.

[0246] For example, when the second home appliance 200 is a washing machine, the rotation speed of the motor included in the washing machine may vary according to the operation mode. Therefore, the amplitude or characteristics of the noise of the noise source (motor) may vary according to the operation mode.

[0247] Therefore, information about each operation mode of the second home appliance 200 and voice data and reference data obtained when the second home appliance 200 operates in each operation mode can be transmitted together so that the external device 400 can generate a noise path estimation filter applicable to each operation mode.

[0248] The external device 400 may calculate a noise path estimation filter using the received voice data and reference data (S1130). Specifically, the external device 400 may obtain a noise path estimation filter using an artificial intelligence model that receives the voice data and reference data and outputs a noise path estimation filter corresponding thereto. The artificial intelligence model may be a linear regression model.

[0249] The external device 400 may transmit the calculated noise path estimation filter information to the first home appliance 100 (S1140). Meanwhile, when the second home appliance 200 is implemented by obtaining the second voice data and generating noise data using the obtained second voice data, the noise path estimation filter may be transmitted to the second home appliance 200.

[0250] The first home appliance 100 may store the received information of the noise path estimation filter (S1150).

[0251] In addition, the first home appliance 100 can generate noise data using the obtained reference data and noise path estimation filter information, and use the generated noise data to remove noise from voice data including the user's spoken voice. The first home appliance 100 can perform a voice recognition function using the voice data from which noise is removed.

[0252] At the same time, Fig.11 In the figure, the external device is illustrated and described as a device other than the speech recognition system, but it can be implemented as a server in the speech recognition system.

[0253] In addition, Fig.11 In the present invention, the voice data and reference data are sent to an external device and a noise path estimation filter generated by an artificial intelligence model is received from the external device, but in implementation, the first home appliance can use a previously stored artificial intelligence model to generate the noise path estimation filter.

[0254] Alternatively, a separate device in the home appliance that is separate from the processor may generate a noise path estimation filter using a pre-stored artificial intelligence model, and the processor may use the generated noise path estimation filter. Meanwhile, the method of generating the noise path estimation filter is not limited to the examples described above.

[0255] Hereinafter, a specific operation of removing noise using a noise path estimation filter (generated using an artificial intelligence model) will be described.

[0256] Fig.12 is a block diagram illustrating a method of removing noise in the time domain according to the third embodiment.

[0257] refer to Fig.12 , a noise removal method using the noise path estimation filter h in the time domain can be identified.

[0258] The processor 130 of the first home appliance 100 may remove noise using the noise path estimation filter h received from the external device 400. Alternatively, when the server 300 generates a noise path estimation filter using an artificial intelligence model, the processor 330 of the server 300 may perform noise removal using the generated noise path estimation filter h.

[0259] Specifically, the processors 130 and 330 may obtain noise data y (y=h*r) by applying the reference data r to the noise path estimation filter h. In addition, the processors 130 and 330 may obtain noise-removed speech data e by removing noise from the speech data d using the noise data y.

[0260] Fig.13 is a block diagram illustrating a method of removing noise in the frequency domain according to the third embodiment.

[0261] refer to Fig.13 , a noise removal method using a noise path estimation filter H in the frequency domain can be identified.

[0262] The processors 130 and 330 may convert the reference data r into the frequency domain using FFT. The processors 130 and 330 may obtain the noise data Y converted into the frequency domain by applying the converted reference data R to the noise path estimation filter H (Y=R·H).

[0263] Furthermore, the processors 130 and 330 may remove noise from the voice data D after the conversion by FFT using the noise data Y. Furthermore, the processors 130 and 330 may obtain the noise-removed voice data e by converting the noise-removed voice data E into the time domain using IFFT.

[0264] exist Fig.12 and Fig.13 In the figure, it is illustrated and described that no updating is performed on the noise path estimation filter, but when implemented, the noise path estimation filter may be updated according to the updating method described above.

[0265] Figure 14 to Figure 16 is a sequence diagram illustrating a speech recognition method performed without intervention of a server.

[0266] Fig.14 is a sequence diagram illustrating a speech recognition method performed without intervention of a server according to the first embodiment of the present disclosure.

[0267] refer to Fig.14First, the first home appliance 100 may transmit a recording request signal to the second home appliance 200 (S1410). In this case, the recording request signal may include information about the predetermined time and information about the predetermined time length.

[0268] The first home appliance 100 may obtain the first voice data corresponding to the recording request signal (S1420). Specifically, the first home appliance 100 may obtain the first voice data based on the predetermined time information and the predetermined time length information included in the recording request signal.

[0269] The second home appliance 200 may obtain the second voice data corresponding to the recording request signal (S1425). Specifically, the second home appliance 200 may obtain the second voice data based on the predetermined time information and the predetermined time length information included in the recording request signal.

[0270] The second home appliance 200 may transmit the obtained second voice data to the first home appliance 100 (S1430).

[0271] The first home appliance 100 can generate noise data using the received second voice data (S1440). Specifically, the first home appliance 100 can generate noise data by applying the received second voice data to a noise path estimation filter, which is sound data corresponding to pure noise of the noise source in the second home appliance 200.

[0272] The first home appliance 100 may perform pre-processing on the first voice data using the noise data (S1450).

[0273] The first home appliance 100 may update the noise path estimation filter (S1460). Specifically, the first home appliance 100 may update the noise path estimation filter using the second voice data and the first voice data, or may update the noise path estimation filter using the second voice data and the preprocessed first voice data. Meanwhile, the method for updating the noise path estimation filter has been described above, so a detailed description thereof will be omitted.

[0274] The first home appliance 100 may perform voice recognition using the preprocessed first voice data (S1470).

[0275] Fig.15 is a sequence diagram illustrating a speech recognition method performed without intervention of a server according to a second embodiment of the present disclosure.

[0276] refer to Fig.15 , first, the first home appliance 100 may send a recording request signal to the second home appliance 200 (S1510).

[0277] The first home appliance 100 may obtain first voice data corresponding to the recording request signal (S1520).

[0278] The second home appliance 200 may obtain second voice data corresponding to the recording request signal (S1525).

[0279] The second home appliance 200 may transmit the obtained second voice data to the first home appliance 100 (S1530).

[0280] The first home appliance 100 may update the noise path estimation filter before performing pre-processing on the first voice data (S1540).

[0281] For example, Figure 7 As shown in RR (which is the autocorrelation value of the reference data) and G DR (which is the cross-correlation value between the reference data and the speech data) to determine the noise path estimation filter (H (k) =G RR(k) ·G DR(k) (-1) ), we can update G RR and G DR to update the noise path estimation filter.

[0282] As another example, Fig. 9 As shown in FIG. 1 , when the noise path estimation filter (h (k) =h (k-1) +μ (k) ·e (k) ·r (k) 、e (k) =h (k-1) *d (k) , μ (k) =α / P(k), P(k)=(1-γ)·P(k-1)+γ·r(k) τ ·r(k)), the noise path estimation filter can be updated by updating the reference data r and P(k) in the buffer.

[0283] In another example, if Fig.10 As shown in FIG. 1 , when the noise path estimation filter (H (k) =H (k-1) +μ·R (k) H ·E (k) , E (k) =D (k) -R (k) ·H(k-1) ), the noise path estimation filter can be updated by updating the Hermitian matrix of the reference data.

[0284] The first home appliance 100 can generate noise data using the received second voice data (S1550). Specifically, the first home appliance 100 can generate noise data by applying the received second voice data to the updated noise path estimation filter, which is sound data corresponding to the pure noise of the noise source in the second home appliance 200.

[0285] The first home appliance 100 may perform pre-processing on the first voice data using the noise data (S1560).

[0286] The first home appliance 100 may perform voice recognition using the preprocessed first voice data (S1570).

[0287] Fig.16 is a sequence diagram illustrating a speech recognition method performed without intervention of a server according to a third embodiment of the present disclosure.

[0288] refer to Fig.16 , first, the first home appliance 100 may send a recording request signal to the second home appliance 200 (S1610).

[0289] In operation S1620, the first home appliance 100 may obtain the first voice data corresponding to the recording request signal. Specifically, the first home appliance 100 may obtain the first voice data based on the predetermined time information and the predetermined time length information included in the recording request signal.

[0290] The second home appliance 200 may obtain second voice data corresponding to the recording request signal (S1625).

[0291] The second home appliance 200 can generate noise data using the second voice data (S1630). Specifically, the second home appliance 200 can generate noise data by applying the second voice data to a noise path estimation filter, which is sound data corresponding to pure noise of a noise source in the second home appliance 200. The noise path estimation filter can be a filter pre-stored in the second home appliance 200 or generated by an artificial intelligence model of the server 300 or the external device 400 and sent to the second home appliance 200.

[0292] The second home appliance 200 may transmit the generated noise data to the first home appliance 100 (S1640).

[0293] The first home appliance 100 may perform pre-processing on the first voice data using the received noise data (S1650).

[0294] The first home appliance 100 may perform voice recognition using the preprocessed first voice data (S1660).

[0295] As described above, the first home appliance 100 may estimate noise using information about noise generated in another home appliance in various ways without intervention of a server, and reduce the estimated noise.

[0296] Fig.17 and Fig.18 is a sequence diagram illustrating a speech recognition method performed according to intervention of a server.

[0297] Fig.17 is a sequence diagram illustrating a speech recognition method performed according to intervention of a server according to the first embodiment of the present disclosure.

[0298] refer to Fig.17 First, the server 300 may send a recording request signal to the first home appliance 100 (S1710). The server 300 may send a recording request signal to the second home appliance 200 (S1715). In this case, the recording request signal may include information about the predetermined time and information about the predetermined time length.

[0299] The first home appliance 100 may obtain the first voice data corresponding to the recording request signal (S1720). Specifically, the first home appliance 100 may obtain the first voice data based on the predetermined time information and the predetermined time length information included in the recording request signal.

[0300] The second home appliance 200 may obtain the second voice data corresponding to the recording request signal (S1725). Specifically, the second home appliance 200 may obtain the second voice data based on the predetermined time information and the predetermined time length information included in the recording request signal.

[0301] The first home appliance 100 may transmit the obtained first voice data to the server 300 (S1730). The second home appliance 200 may transmit the obtained second voice data to the server 300 (S1735).

[0302] The server 300 may generate noise data using the received second voice data (S1740). Specifically, the server 300 may generate noise data, which is sound data corresponding to pure noise of the noise source in the second home appliance 200, by applying the received second voice data to a noise path estimation filter.

[0303] The server 300 may pre-process the first voice data using the noise data (S1750). The server 300 may transmit the pre-processed first voice data to the first home appliance 100 (S1760).

[0304] The server 300 may update the noise path estimation filter (S1770). Specifically, the server 300 may update the noise path estimation filter using the second voice data and the first voice data, or update the noise path estimation filter using the second voice data and the preprocessed first voice data. Meanwhile, the method for updating the noise path estimation filter has been described above, so a detailed description thereof will be omitted.

[0305] The first home appliance 100 may perform voice recognition using the preprocessed first voice data (S1780).

[0306] Fig.18 is a sequence diagram illustrating a speech recognition method performed under the intervention of a server according to a second embodiment of the present disclosure.

[0307] refer to Fig.18 First, the server 300 may transmit a recording request signal to the first home appliance 100 (S1810). The server 300 may transmit a recording request signal to the second home appliance 200 (S1815).

[0308] The first home appliance 100 may obtain first voice data corresponding to the recording request signal (S1820). The second home appliance 200 may obtain second voice data based on the recording request signal (S1825).

[0309] The first home appliance 100 may transmit the obtained first voice data to the server 300 (S1830). The second home appliance 200 may transmit the obtained second voice data to the server 300 (S1835).

[0310] The server 300 may update the noise path estimation filter before performing preprocessing on the first voice data (S1840). Fig.15 The operation of updating the noise path estimation filter before performing the preprocessing is described, and thus a redundant description thereof will be omitted.

[0311] The server 300 may generate noise data using the received second voice data (S1850). Specifically, the server 300 may generate noise data corresponding to pure noise of the noise source in the second home appliance 200 by applying the received second voice data to the updated noise path estimation filter.

[0312] The server 300 may pre-process the first voice data using the noise data (S1860). The server 300 may transmit the pre-processed first voice data to the first home appliance 100 (S1870).

[0313] The first home appliance 100 may perform voice recognition using the preprocessed first voice data (S1880).

[0314] As described above, the server may estimate the noise using the information about the noise generated in the second home appliance in various ways and reduce the estimated noise from the voice data of the first home appliance.

[0315] exist Fig.17 and Fig.18 In the embodiment, it is illustrated and described that the server 300 is connected to two home appliances, but in implementation, the server may be connected to two or more home appliances and estimate noise using information about noise generated in the two or more home appliances.

[0316] Fig.19 is a flowchart illustrating a voice recognition method of a first home appliance according to an embodiment of the present disclosure.

[0317] refer to Fig.19 First, first voice data is obtained through a microphone of a first home appliance (S1910).

[0318] In addition, other home appliances that can be connected to the first home appliance can be detected. Specifically, a second home appliance that can be connected to the first home appliance through short-range wireless communication or a second home appliance connected to the access point AP to which the first home appliance is connected can be detected.

[0319] The first home appliance may request the second home appliance to send voice data. Specifically, a recording request signal for requesting to generate and send second voice data may be sent to the second home appliance, where the second voice data is voice data about noise generated in the second home appliance.

[0320] Here, the recording request signal may include information about the predetermined time or information about the predetermined time length. Meanwhile, the information that may be included in the recording request signal is not limited to the examples described above.

[0321] The second voice data is received from the second home appliance (S1920). Specifically, the second voice data generated based on the predetermined time information and the predetermined time length information included in the recording request signal can be received from the second home appliance.

[0322] The first voice data may be obtained based on a request sent to the second home appliance. Specifically, the first voice data may be obtained through a microphone of the first home appliance based on the predetermined time information and the predetermined time length information included in the recording request signal sent to the second home appliance.

[0323] Noise data is generated using the received second voice data (S1930). Here, the noise data refers to sound data corresponding to the noise of the noise source.

[0324] The noise data may be obtained by extracting only a component corresponding to the noise of the noise source from the second voice data or filtering the remaining components except the component corresponding to the noise of the noise source.

[0325] The noise data may be generated by filtering remaining components except for components corresponding to a noise source included in the second voice data using at least one of a finite impulse response (FIR) filter or an infinite impulse response (IIR) filter in a time domain.

[0326] Alternatively, the noise data may extract a component corresponding to the noise of the noise source from the second voice data using a transfer function predetermined for each frequency band on the frequency domain.

[0327] Information about the filter or transfer function described above may be pre-stored in the home appliance at the manufacturing stage and used to generate noise data, but is not limited thereto.

[0328] For example, when communication with an external device is possible, information about a filter or a transfer function may be received through communication with the external device, and noise data may be generated using the received information about the filter or the transfer function.

[0329] In this case, the information about the filter or the transfer function received from the external device may be information obtained using an artificial intelligence model included in the external device.

[0330] Signal processing is performed on the first voice data using the generated noise data (S1940). Specifically, signal processing for removing a component corresponding to noise of a noise source included in the first voice data may be performed using the noise data.

[0331] Meanwhile, the noise source of the second home appliance may be a motor provided in the housing of the second home appliance to perform a predetermined function of the second home appliance. In this case, the second voice data may be voice data obtained from a microphone of the second home appliance. The noise data may be motor noise data of the motor of the second home appliance, and preprocessing of the first voice data may be performed by removing the motor noise from the first voice data.

[0332] A signal processing method for the first voice data may be changed according to whether noise is generated in the second home appliance.

[0333] For example, it is assumed that the sound generated in the second home appliance is generated by a motor included in the second home appliance. In this case, the pre-processing method for the first voice data can be changed depending on whether the motor of the second home appliance is driven.

[0334] Specifically, it can be determined whether the motor of the second home appliance is driven. If the motor of the second home appliance is driven, the amplitude of the noise generated in the second home appliance is large, and therefore, the second voice data can be used to perform preprocessing of the first voice data according to the method described above.

[0335] At the same time, when the motor of the second home appliance is not driven, there is no noise generated in the second home appliance itself, so the first voice data can be used instead of the second voice data to directly perform voice recognition, or the first voice data can be pre-processed according to the relevant technical method.

[0336] Meanwhile, even when the noise source of the second home appliance is a fan or the like instead of a motor, the method of preprocessing the first voice data may be changed according to whether the noise source is driven.

[0337] Voice recognition is performed using the signal-processed first voice data (S1950).

[0338] In addition, even when the noise of the noise source changes or the surrounding environment changes, the method of generating noise data can be updated using at least one of the first voice data including noise, the signal-processed first voice data, or the second voice data to generate accurate noise data.

[0339] Specifically, the method of generating noise data may be updated using the correlation between at least one of the first voice data and the signal-processed first voice data and the second voice data.

[0340] Meanwhile, as described above, the operation of updating the method of generating noise data may be performed after performing signal processing on the first voice data, but alternatively, the operation of updating the method of generating noise data may be performed first before performing signal processing on the first voice data.

[0341] Therefore, in the voice recognition method of the home appliance of the present disclosure, voice data of noise is obtained from another home appliance that generates noise of large amplitude, and preprocessing is performed on the obtained noise using the voice data, thereby accurately removing the noise included in the voice data obtained by the microphone even if the amplitude of the noise generated from the outside is large. Fig.19 The speech recognition method shown in Figure 2 or Figure 3 The method may be performed on a home appliance having a configuration of, or may be performed on a home appliance having another configuration.

[0342] Furthermore, the speech recognition method as described above may be implemented by at least one executable program for executing the speech recognition method as described above, and such an executable program may be stored in a non-transitory readable medium.

[0343] Non-transitory readable media refers to media that semi-permanently stores data and is readable by a device, rather than media that stores data for a short period of time, such as registers, caches, memories, etc. Specifically, various applications or programs can be stored and provided in non-transitory readable media (such as CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, ROMs, etc.).

[0344] Fig. 20 is a flowchart illustrating a voice recognition method of a second home appliance according to an embodiment of the present disclosure.

[0345] refer to Fig. 20 First, a voice data request signal may be received (S2010). Specifically, a recording request signal requesting generation and transmission of second voice data may be received from a server connected to the second home appliance or the first home appliance, the second voice data being voice data of noise generated in the second home appliance.

[0346] The second voice data may be obtained through the microphone of the second home appliance (S2020). Specifically, the microphone of the second home appliance may be used to detect noise, and the second voice data may be obtained based on the received recording request signal. More specifically, the second voice data may be obtained based on information about the predetermined time and the predetermined time length information included in the received recording request signal.

[0347] The obtained second voice data may be transmitted (S2030). Specifically, the obtained second voice data may be transmitted to a server connected to the second home appliance or the first home appliance.

[0348] Meanwhile, the second voice data may not be transmitted to the server connected to the second home appliance or the first home appliance, and noise data may be generated using the second voice data and the generated noise data may be transmitted to the first home appliance or the server.

[0349] Therefore, in the voice recognition method of the home appliance according to the present disclosure, since reference data about the noise of the internal noise source is generated and provided to the home appliance or the server that is to perform the voice recognition function, the home appliance or the server can accurately remove the noise included in the voice data including the user's speech. Fig. 20 The speech recognition method shown in Figure 4 The method may be performed on a home appliance having a configuration of, or may be performed on a home appliance having another configuration.

[0350] Furthermore, the speech recognition method as described above may be implemented by at least one executable program for executing the speech recognition method as described above, and such an executable program may be stored in a non-transitory readable medium.

[0351] Fig.21 is a flowchart illustrating a speech recognition method of a server according to an embodiment of the present disclosure.

[0352] refer to Fig.21 First, voice data may be requested from the first home appliance and the second home appliance (S2110). Specifically, a recording request signal for requesting generation and transmission of voice data may be sent to the first home appliance and the second home appliance.

[0353] In addition, the recording request signal may include information about the predetermined time and information about the predetermined time length. The information included in the recording request signal is not limited to the examples described above.

[0354] The first voice data and the second voice data may be received (S2120). Specifically, the first voice data obtained based on the recording request signal may be received from the first home appliance. The second voice data obtained based on the recording request signal may be received from the second home appliance.

[0355] The noise data may be generated using the second voice data (S2130). Here, the noise data refers to sound data corresponding to pure noise of a noise source in the second home appliance.

[0356] Meanwhile, the method of generating noise data has been described above, so a detailed description thereof will be omitted.

[0357] Signal processing may be performed on the first voice data using the noise data (S2140). Specifically, signal processing for removing a component corresponding to noise of a noise source included in the first voice data may be performed using the noise data.

[0358] The signal-processed first voice data may be transmitted to the first home appliance (S2150).

[0359] Therefore, in the voice recognition method of the server of the present disclosure, first voice data is received from a first home appliance that will perform a voice recognition function, second voice data or reference data about noise is obtained from a second home appliance that generates a large amplitude of noise, and pre-processing is performed on the first voice data using the obtained second voice data or reference data, so that even if the amplitude of the noise is large, the noise included in the voice data can be accurately removed. Fig.21 The speech recognition method shown in Figure 5 The process may be executed on a server having a configuration of the embodiment of the present invention, or may be executed on a server having another configuration.

[0360] Furthermore, the speech recognition method as described above may be implemented using at least one executable program for executing the speech recognition method as described above, and such an executable program may be stored in a non-transitory readable medium.

[0361] Although the present disclosure has been described with reference to the accompanying drawings, it should be understood that the scope of the present disclosure is defined by the claims described below and should not be interpreted as being limited to the embodiments and / or figures described above. It should be clearly understood that improvements, changes and modifications that are obvious to those skilled in the art are also within the scope of the present disclosure as defined in the claims.

[0362] Although the present disclosure has been described with various embodiments, various changes and modifications may occur to those skilled in the art. It is intended that the present disclosure encompass such changes and modifications as fall within the scope of the appended claims.

Claims

1. A household appliance comprising: A communication device configured to communicate with another household appliance, wherein the other household appliance includes a motor disposed in a housing of the other household appliance, the motor being configured to perform a predetermined function of the other household appliance; a microphone configured to receive speech from a user; as well as The processor is configured as: Obtaining first voice data according to the received voice; controlling the communication device to determine whether the motor of the other household appliance is driven; generating noise data using second voice data received from the other home appliance, wherein the noise data is motor noise data about a motor of the other home appliance, and the second voice data is voice data obtained from a microphone provided in a housing of the other home appliance; performing signal processing on the first voice data to remove motor noise with respect to the first voice data using the motor noise data; Based on driving the motor of the other home appliance, voice recognition is performed using the signal-processed first voice data, or based on not driving the motor of the other home appliance, the voice recognition is performed using the first voice data.

2. The household appliance according to claim 1, wherein: The processor is further configured to: controlling the communication device to request transmission of voice data from the other home appliance; obtaining the first voice data from the microphone based on the request; and The second voice data obtained from the other home appliance based on the request is received.

3. The household appliance according to claim 1, wherein: The processor is further configured to generate the noise data using at least one of a finite impulse filter (FIR) or an infinite impulse filter (HR).

4. The household appliance according to claim 1, wherein: The processor is further configured to: converting the second speech data into a frequency domain; and The noise data is generated using the converted second voice data and a predetermined transfer function for each frequency band.

5. The household appliance according to claim 1, wherein: The processor is also configured to update a method of generating the noise data using at least one of the first voice data, the signal-processed first voice data, or the second voice data.

6. The household appliance according to claim 5, wherein: The processor is further configured to generate the noise data using a correlation between: at least one of the first speech data or the signal-processed first speech data; and The second voice data.

7. The household appliance according to claim 1, wherein: The processor is further configured to generate the noise data using filter information for generating the noise data received from an external device through the communication device; and The received filter information is obtained using an artificial intelligence model included in the external device.

8. A method for speech recognition of a household appliance, the method comprising: obtaining first voice data according to the voice received from the microphone; as well as determining whether a motor of another home appliance is driven, wherein the motor of the other home appliance is disposed in a housing of the other home appliance, and the motor is configured to perform a predetermined function of the other home appliance; receiving second voice data from the other household appliance; generating noise data using the received second voice data, wherein the noise data is motor noise data about a motor of the other home appliance, and the second voice data is voice data obtained from a microphone provided in a housing of the other home appliance; performing signal processing on the first voice data to remove motor noise with respect to the first voice data using the motor noise data; as well as Based on driving the motor of the other home appliance, voice recognition is performed using the signal-processed first voice data, or based on not driving the motor of the other home appliance, the voice recognition is performed using the first voice data.

9. The method according to claim 8, further comprising: Requesting voice data to be sent from another home appliance, Wherein, obtaining the first voice data comprises: obtaining the first voice data based on the sent request, The receiving of the second voice data includes: receiving the second voice data obtained from the other home appliance based on the sent request.

10. The method according to claim 8, wherein: Generating the noise data includes generating the noise data using at least one of a finite impulse filter (FIR) or an infinite impulse filter (HIR).

11. The method according to claim 8, wherein: Generating the noise data comprises: converting the second speech data into a frequency domain; and The noise data is generated using the converted second voice data and a predetermined transfer function for each frequency band.

Images