Information processor and device system using the same

Abstract

To provide an information processor which generates an appropriate command for a prescribed device based on voice uttered by a user by using a large scale language model even in a noise environment in which voice uttered by the user is difficult to be recognized, and to provide a device system using the information processor.SOLUTION: An information processor 100 includes: prompt generation means 130 for generating a prompt by combining input data generated based on voice of a user with command data 50 constituted by a plurality of commands which can control a prescribed device 10; prompt transmission means 140 for transmitting the prompt to a large scale language model 40; response reception means 150 for receiving a response from the large scale language model 40; and command generation means 160 for generating the command to control the prescribed device 10 based on the response.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to an information processing apparatus and a device system using the same.

Background Art

[0002] Generally, a system in which a user inputs a command to a device by voice and operates the device has become widespread. In such a system, it is important to appropriately recognize the voice spoken by the user.

[0003] Also, in a factory where devices such as manufacturing equipment and machines are operating, a system for controlling the operation of the devices by recognizing the voice spoken by the user is known. In a factory, motor sounds and machine sounds are generated in the devices of the manufacturing facilities, and furthermore, there is also noise such as collision sounds that suddenly occur when members and equipment collide. Therefore, it may be difficult to appropriately recognize the voice spoken by the user in such a noisy environment.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] By the way, in recent years, AI (Artificial Intelligence) technology utilizing large language models, which involves inputting a question (prompt) and obtaining an answer thereto, has been rapidly developing. For example, Patent Document 1 discloses a technique for generating a prompt that adds an effective sentence as reference information to the input question sentence within the range of a character limit in order to easily obtain an appropriate answer sentence.

[0006] Even in a noisy environment as described above, it is conceivable to use such a large language model for the purpose of appropriately operating a device based on the voice spoken by the user.

[0007] Therefore, an object of the present invention is to provide an information processing apparatus that generates an appropriate command for a predetermined device based on the voice spoken by the user using a large language model even in a noisy environment where it is difficult to recognize the voice spoken by the user, and a device system using the same.

Means for Solving the Problems

[0008] An information processing apparatus according to an aspect of the present invention is an information processing apparatus that processes a command for controlling a predetermined device, and includes a prompt generation means for generating a prompt by combining command data composed of a plurality of commands capable of controlling the predetermined device with input data generated based on the voice of the user, a prompt transmission means for transmitting the prompt to a large language model, a response reception means for receiving a response from the large language model, and a command generation means for generating a command for controlling the predetermined device based on the response.

[0009] According to this aspect, the prompt generation means generates a prompt by combining the input data generated based on the voice of the user with the command data composed of a plurality of commands capable of controlling the predetermined device. The prompt transmission means transmits the prompt to the large language model, and the response reception means receives a response from the large language model. Then, the command generation means generates a command for controlling the predetermined device based on the response. Thereby, even in a noisy environment where it is difficult to recognize the voice spoken by the user, an appropriate command for the predetermined device can be generated based on the voice spoken by the user using the large language model. As a result, the predetermined device can operate appropriately according to the appropriate command.

[0010] In the above aspect, the command data may be preset according to the predetermined device.

[0011] According to this aspect, since the command data corresponding to the predetermined device is preset, the prompt generation means can appropriately generate a prompt configured to include a command capable of controlling the predetermined device.

[0012] In the above aspect, the prompt may include content for selecting one command corresponding to the input data from a plurality of commands capable of controlling the predetermined device.

[0013] According to this aspect, since the prompt generation means generates a prompt for selecting one command from a plurality of commands capable of controlling the predetermined device, the answer receiving means can receive, as an answer from the large language model, a command capable of controlling the predetermined device.

[0014] In the above aspect, a determination means for determining whether detailed information indicating the degree is required for the command generated based on the answer is further provided. When it is determined that detailed information is required, the prompt generation means generates a second prompt including content from which detailed information can be obtained from the input data, the prompt transmission means transmits the second prompt to the large language model, the answer receiving means receives the second answer from the large language model, and the command generation means may generate a command including detailed information for controlling the predetermined device based on the answer and the second answer.

[0015] According to this aspect, when the determination means determines that detailed information indicating the degree is required for the command generated based on the answer, the prompt generation means further generates a second prompt including the content from which the detailed information can be obtained from the input data. The prompt transmission means transmits the second prompt to the large language model, and the answer reception means receives a second answer from the large language model. Then, the command generation means generates a command for controlling a predetermined device based on the answer and the second answer. Thereby, even in a noisy environment where the voice spoken by the user is difficult to recognize, an appropriate command for the predetermined device can be generated using the large language model based on the voice spoken by the user, and furthermore, a command including detailed information can also be appropriately generated. As a result, the predetermined device can operate appropriately according to the appropriate command.

[0016] In the above aspect, the detailed information may include numerical information.

[0017] According to this aspect, the answer reception means can receive numerical information as the second answer from the large language model, and the command generation means can generate an appropriate command including the numerical information.

[0018] The device system according to an aspect of the present invention includes the information processing device described above and a predetermined device that operates by receiving a command generated by the information processing device.

[0019] According to this aspect, the above-described information processing device can generate an appropriate command for a predetermined device based on the voice spoken by the user using the large language model even in a noisy environment where the voice spoken by the user is difficult to recognize, and the predetermined device can operate appropriately according to the appropriate command.

Effect of the Invention

[0020] According to the present invention, even in a noisy environment where it is difficult to recognize the voice spoken by the user, an information processing apparatus that generates an appropriate command for a predetermined device based on the voice spoken by the user using a large language model, and a device system using the same can be provided.

Brief Description of the Drawings

[0021]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Embodiments for Carrying Out the Invention

[0022] Hereinafter, each embodiment of the present invention will be specifically described with reference to the drawings. Note that each embodiment described below is merely a specific example for implementing the present invention, and does not limit the interpretation of the present invention. Also, for ease of understanding the description, the same reference numerals are given to the same components in each drawing as much as possible, and redundant descriptions may be omitted.

[0023] <First Embodiment> [Configuration of Information Processing Apparatus] FIG. 1 is a functional block diagram showing each function of an information processing apparatus 100 according to the first embodiment of the present invention. As shown in FIG. 1, the information processing apparatus 100 includes a voice processing unit 110, a storage unit 120, a prompt generation unit 130, a prompt transmission unit 140, a response reception unit 150, and a command generation unit 160, and generates a command for controlling the welding machine 10 to control the operation of the welding machine 10.

[0024] The welding machine 10 receives the command generated by the information processing apparatus 100 and operates according to the command.

[0025] The microphone 20, the speech recognition system 30, and the large language model 40 are each communicably connected to the information processing apparatus 100.

[0026] For example, the microphone 20 is disposed near the welding machine 10 and is connected to the information processing apparatus 100 by wire or wirelessly (typically, Wi-Fi or Bluetooth (registered trademark)).

[0027] The speech recognition system 30 and the large language model 40 are, for example, cloud systems, and the information processing apparatus 100 is configured to be able to use these systems via a communication network such as the Internet.

[0028] The voice processing means 110 receives the voice data spoken by the user and converts it into character data. For example, the voice spoken by the user operating the welding machine 10 is input into the voice input unit 21 in the microphone 20 and transmitted to the information processing device 100 by the voice transmission unit 22.

[0029] Then, the voice processing means 110 converts the received voice data into character data by performing speech recognition using the speech recognition system 30. The speech recognition system 30 may be a so-called speech recognition AI such as Whisper (OpenAI) and Speech-to-Text (Google), for example.

[0030] Here, the microphone 20 is arranged near the welding machine 10 as a separate device from the information processing device 100, but it is not limited to this. For example, if the voice data spoken by the user can be directly acquired by arranging the information processing device 100 near the welding machine 10, the information processing device 100 may be configured to have the function of the microphone 20.

[0031] Also, here, the speech recognition system 30 is a cloud system and is configured to be used from the information processing device 100 via the Internet, but it is not limited to this. For example, the information processing device 100 may be equipped with an application or the like having a speech recognition function capable of converting voice data into character data.

[0032] The storage means 120 is, for example, a memory or the like, and stores in advance the command data corresponding to the welding machine 10. The command data is composed of a plurality of commands capable of controlling the welding machine 10, and specifically, it may correspond to a menu operated by the user when using the welding machine 10.

[0033] FIG. 2 is a diagram showing an example of command data 50 for controlling the welding machine 10. As shown in FIG. 2, commands for controlling the welding machine 10 include "turn on the power", "turn off the power", "change to DC mode", "change to pulse mode", "increase voltage", "decrease voltage", "increase current", "decrease current", etc. That is, these commands may be those normally provided as a menu for the user to operate the welding machine 10.

[0034] Here, although the command data 50 corresponding to the welding machine 10 is stored in advance in the storage means 120 in the information processing apparatus 100, it is not limited thereto. For example, the command data 50 may be stored in the memory of an external device separate from the information processing apparatus 100, and the information processing apparatus 100 may refer to the command data 50 by accessing the memory of the external device via a network.

[0035] The prompt generation means 130 generates a prompt by combining the character data converted by the voice processing means 110 with the command data 50 corresponding to the welding machine 10. For example, the prompt is of such content as to select one command from a plurality of commands that can control the welding machine 10.

[0036] Specifically, the prompt may be of such content as to select one command from the plurality of commands (such as "turn on the power", "turn off the power", "change to DC mode", "change to pulse mode", "increase voltage", "decrease voltage", "increase current", "decrease current", etc.) that make up the command data 50 illustrated in FIG. 2. Details of the prompt will be described later.

[0037] The prompt transmission means 140 transmits the prompt generated by the prompt generation means 130 to the large language model 40. Here, the large language model 40 for the prompt is, for example, GTP3.5 or the like.

[0038] The response receiving means 150 receives a response to the prompt from the large language model 40 to which the prompt is transmitted by the prompt transmitting means 140.

[0039] Based on the response from the large language model 40 received by the response receiving means 150, the command generation means 160 generates a command for controlling the welding machine 10.

[0040] The information processing apparatus 100 transmits the thus generated command to the welding machine 10, and the welding machine 10 operates according to the command.

[0041] [Regarding the prompt in the large language model] Hereinafter, the prompt transmitted to the large language model 40 will be described in detail. FIG. 3 is a diagram showing the state until a command for controlling the welding machine 10 is generated from the voice spoken by the user.

[0042] First, the user who operates the welding machine 10 speaks "Change to pulse mode" toward the microphone 20. The voice processing means 110 in the information processing apparatus 100 converts the voice data "Change to pulse mode" into character data using the speech recognition system 30.

[0043] Here, assuming that in a noisy environment in the factory, "Change to pulse mode" spoken by the user is speech-recognized (converted) into character data "Change to ars mode".

[0044] The prompt generation means 130 generates a prompt "Which of "Change to ars mode", "Turn on the power", "Turn off the power", "Change to DC mode", "Change to pulse mode", "Increase the voltage", "Decrease the voltage", "Increase the current", "Decrease the current",... does it correspond to? Answer only with the answer."

[0045] Then, the prompt is sent by the prompt sending means 140 to the large language model 40, and as an answer from the large language model 40, the answer "Change to pulse mode" is received by the answer receiving means 150.

[0046] As the configuration of the prompt, the character data "Change to pulse mode" is combined with a plurality of commands (such as "Turn on the power", "Turn off the power", "Change to DC mode", "Change to pulse mode", "Increase the voltage", "Decrease the voltage", "Increase the current", "Decrease the current", etc.) that make up the command data 50 shown in FIG. 2, and the content is such that one command is selected from among the plurality of commands.

[0047] As described above, all of the plurality of commands that make up the command data 50 are commands that can control the welding machine 10 and are usually provided as a menu for the user to operate the welding machine 10. That is, even if the user's spoken "Change to pulse mode" cannot be properly transcribed in a noisy environment in the factory, the prompt sent to the large language model 40 is such that one command is selected from among the plurality of commands that can control the welding machine 10. Therefore, no matter which command is selected, the welding machine 10 can be properly operated according to the command.

[0048] [Command Generation Method] Next, a method for generating a command for controlling the welding machine 10 from the voice spoken by the user will be specifically and detailedly described.

[0049] FIG. 4 is a flowchart showing the processing flow of the command generation method M100 executed by the information processing apparatus 100 according to the first embodiment of the present invention. As shown in FIG. 4, the command generation method M100 includes steps S110 to S160, and each step is executed by a processor included in the information processing apparatus 100.

[0050] In step S110, the voice processing means 110 acquires the voice uttered by the user as voice data via, for example, the microphone 20.

[0051] In step S120, the voice processing means 110 converts the voice data acquired in step S110 into character data by performing speech recognition using, for example, the speech recognition system 30.

[0052] In step S130, the prompt generation means 130 generates a prompt by combining the character data obtained by speech recognition in step S120 with the command data 50 corresponding to the welding machine 10 as shown in FIG. 2 pre-stored in the storage means 120. Specifically, the prompt generation means 130 generates a prompt as shown in FIG. 3, which contains content for selecting one command from a plurality of commands that can control the welding machine 10.

[0053] In step S140, the prompt transmission means 140 transmits the prompt generated in step S130 to the large language model 40.

[0054] In step S150, the response receiving means 150 receives from the large language model 40 a response to the prompt transmitted in step S140. Here, the received response is one selected command from a plurality of commands that can control the welding machine 10.

[0055] In step S160, the command generation means 160 generates a command for controlling the welding machine 10 based on the response from the large language model 40 received in step S150.

[0056] As described above, according to the information processing apparatus 100 and the command generation method M100 according to the first embodiment of the present invention, the prompt generation means 130 combines the input data transcribed based on the user's voice with the command data 50 composed of a plurality of commands that can control the welding machine 10 stored in advance to generate a prompt. The prompt transmission means 140 transmits the prompt to the large language model 40, and the answer receiving means 150 receives the answer from the large language model 40. Then, the command generation means 160 generates a command for controlling the welding machine 10 based on the answer. Thereby, even in a noisy environment where it is difficult to recognize the voice spoken by the user, an appropriate command for the welding machine 10 can be generated based on the voice spoken by the user using the large language model 40. As a result, the welding machine 10 can operate appropriately according to the appropriate command.

[0057] That is, even in a noisy environment where the voice spoken by the user cannot be properly transcribed, as the content for selecting one of the plurality of commands that can control the welding machine 10 as a prompt in the large language model 40, no matter which command is selected, the welding machine 10 can be appropriately operated according to the command.

[0058] Also, when no correct transcription can be made from the voice spoken by the user in a noisy environment, the answer receiving means 150 may receive an answer such as "There are no options." or "I don't understand with the current information." from the large language model 40 and notify the user of it (by voice and / or on the screen).

[0059] In the present embodiment, the operation of the welding machine 10 based on the voice spoken by the user in the factory has been described. However, the present invention is not limited thereto. For example, as the device that operates based on the voice spoken by the user, other devices such as a plasma cutting machine, a press working machine, each manufacturing device in a food factory, and construction equipment at a construction site may be used. Further, not only industrial equipment but also household and personal electric appliances, and various devices mounted on an automobile may be used.

[0060] <Second Embodiment> Next, the second embodiment will be described. Here, mainly, the configuration different from that of the first embodiment of the present invention will be described in detail, and for the configuration identical to that of the first embodiment, the same reference numerals will be given, and the detailed description will be omitted or simplified.

[0061] FIG. 5 is a functional block diagram showing each function of the information processing apparatus 200 according to the second embodiment of the present invention. As shown in FIG. 5, the information processing apparatus 200 includes a voice processing means 110, a storage means 120, a prompt generation means 230, a prompt transmission means 240, a response reception means 250, a command generation means 260, and a determination means 270, and processes a command for controlling the welding machine 10.

[0062] The prompt generation means 230 generates a prompt by combining the character data converted by the voice processing means 110 with the command data 50 corresponding to the welding machine 10. For example, as described in the first embodiment of the present invention, the prompt is such that one command is selected from a plurality of commands capable of controlling the welding machine 10.

[0063] Furthermore, the prompt generation means 230 generates a second prompt including content such that detailed information (numerical information) can be obtained from the character data converted by the voice processing means 110 according to the determination result by the determination means 270 described later.

[0064] The prompt transmission means 240 transmits the prompt generated by the prompt generation means 230 to the large language model 40, and further, when a second prompt is generated by the prompt generation means 230, transmits the second prompt to the large language model 40.

[0065] The response receiving means 250 receives a response (second response) to the prompt (second prompt) from the large language model 40 to which the prompt (second prompt) has been transmitted by the prompt transmission means 240.

[0066] The command generation means 260 generates a command for controlling the welding machine 10 based on the response (second response) from the large language model 40 received by the response receiving means 250.

[0067] The information processing apparatus 200 transmits the command thus generated to the welding machine 10, and the welding machine 10 operates according to the command.

[0068] Here, the determination by the determination means 270 will be described. The determination means 270 determines whether detailed information indicating the degree is required for the command generated based on the response from the large language model 40 received by the response receiving means 250 with respect to the prompt transmitted to the large language model 40 by the prompt transmission means 240.

[0069] For example, in the case of the command "change to pulse mode" as exemplified in the first embodiment of the present invention, since the welding machine 10 can be operated according to the command, the determination means 270 determines that detailed information is not required.

[0070] Regarding commands generated based on the answers from the large language model 40, in cases such as "Increase the voltage", "Decrease the voltage", "Increase the current", or "Decrease the current", detailed information indicating the degree to which the voltage or current is to be adjusted is required. In this case, the determination means 270 determines that detailed information is necessary.

[0071] [Regarding the prompt (second prompt) in the large language model] The prompt (second prompt) transmitted to the large language model 40 will be described in detail below. FIG. 6 is a diagram showing the state from when the user speaks until the answer from the large language model 40 is received.

[0072] The user operating the welding machine 10 speaks "Increase voltage by 2V" towards the microphone 20, and here it is assumed that it is appropriately transcribed (converted) into the character data "Increase voltage by 2V".

[0073] The prompt generation means 230 generates, in the same process as the prompt generation means 130 described in the first embodiment of the present invention, the prompt "Which of the following does 'Increase voltage by 2V' correspond to: 'Turn on the power', 'Turn off the power', 'Change to DC mode', 'Change to pulse mode', 'Increase the voltage', 'Decrease the voltage', 'Increase the current', 'Decrease the current',...? Only answer with the answer."

[0074] Then, the prompt is transmitted to the large language model 40 by the prompt transmission means 240, and as an answer from the large language model 40, the answer "Increase the voltage" is received by the answer receiving means 250.

[0075] Here, the determination means 270 determines whether detailed information indicating the degree is required for the command generated based on the answer, and determines that detailed information is required regarding how much to increase the voltage.

[0076] FIG. 7 is a diagram showing a state until a command (with numerical information) for controlling the welding machine 10 is generated after receiving the second response from the large language model 40 after receiving the response from the large language model 40.

[0077] Based on the character data "Increase voltage by 2V" that has been speech-to-text (converted) by the user operating the welding machine 10 and the response "Increase the voltage" from the large language model 40 as described in FIG. 6, the prompt generation means 230 generates a second prompt "To follow the instruction 'Increase voltage by 2V', by how many volts should the voltage be increased? Answer with a number."

[0078] The composition of the second prompt is such that it only asks for the numerical information contained in the speech-to-text character data "Increase voltage by 2V".

[0079] Then, the second prompt is sent to the large language model 40 by the prompt sending means 240, and as the second response from the large language model 40, the response "2" is received by the response receiving means 250.

[0080] Based on the response "Increase the voltage" from the large language model 40 and the second response "2" received by the response receiving means 250, the command generation means 260 generates a command "Increase the voltage by 2V" for controlling the welding machine 10.

[0081] The information processing device 200 sends the command generated in this way to the welding machine 10, and the welding machine 10 operates to increase the voltage by 2V according to the command.

[0082] Here, the command generation means 260 generates a command for specifying the voltage increase value. However, it may generate a command for calculating the voltage value to be increased and specifying the set value of the voltage. For example, the information processing apparatus 200 receives the current set value of the voltage from the welding machine 10 and adds the voltage value for increasing the second answer from the large language model 40 to the set value. Specifically, when the current set value of the voltage is "20V" and the second answer from the large language model 40 is "2", the command generation means 260 generates a command "Set (change) the voltage to 22V".

[0083] Also, in this embodiment, the user operating the welding machine 10 is speaking towards the microphone 20, specifically including "Increase the voltage by 2V" along with what voltage exactly. However, it is not limited to this. For example, it may be configured to be able to respond to utterances such as "Increase the voltage by one step", "Decrease the voltage by one level", and "Increase the voltage by 10%". Specifically, how much (for example, how many volts or what percentage, etc.) is set in advance for "one step" or "one level", and when the second prompt is generated by the prompt generation means 230, or when the second answer is received from the large language model 40 by the answer receiving means 250, it may be converted into how many volts.

[0084] Furthermore, for example, it may be configured to be able to respond not only to numerical information such as "Increase the voltage a little", "Increase the voltage considerably", and "Decrease the voltage a little". For example, in the case of "a little" and "a little bit", it may be converted to 1V or 10%, etc., and in the case of "considerably", it may be converted to 5V or 20%, etc. Also, the current set value (for example, the current set voltage value "20V", etc.) is obtained from the welding machine 10, and the second prompt including the current set value and "a little", "considerably", and "a little bit" as they are is generated by the prompt generation means 230, and then the second answer (only a numerical value) is received from the large language model 40. In this way, the large language model 40 can receive, as the second answer (only a numerical value), the set value considering "a little", "considerably", and "a little bit" from the current set value.

[0085] [Command Generation Method] Next, a method for generating a command to control the welding machine 10 from the voice spoken by the user will be specifically and detailedly described.

[0086] FIG. 8 is a flowchart showing the processing flow of the command generation method M200 executed by the information processing apparatus 200 according to the second embodiment of the present invention. As shown in FIG. 8, the command generation method M200 includes steps S110 to S160 in the command generation method M100 according to the first embodiment of the present invention, and further includes steps S210 to S250. Each of these steps is executed by a processor included in the information processing apparatus 200.

[0087] Steps S110 to S160 are the same processing as steps S110 to S160 in the command generation method M100 described with reference to FIG. 4, and are respectively processed by the voice processing means 110, the prompt generation means 230, the prompt transmission means 240, the response reception means 250, and the command generation means 260.

[0088] In step S210, the determination means 270 determines whether detailed information indicating the degree is required for the command generated based on the response from the large language model 40 received in step S150.

[0089] For example, when a command "Change to pulse mode" as shown in FIG. 3 is generated, the determination means 270 determines that detailed information is not required because the welding machine 10 can be operated according to the command, and proceeds to the processing of step S160 ("No" in step S210).

[0090] On the other hand, when a command "Increase voltage" as shown in FIG. 6 is generated, the determination means 270 determines that detailed information indicating how much the voltage is to be increased (how many volts to increase the voltage, or to what voltage to set the voltage after increasing the voltage) is required, and proceeds to the processing of step S220 ("Yes" in step S210).

[0091] In step S220, the prompt generation means 230 generates a second prompt including content such that detailed information (numerical information) can be obtained from the character data transcribed in step S120. Specifically, the prompt generation means 230 generates a prompt as shown in FIG. 7, and generates content that causes only the numerical information included in the character data transcribed in step S120 to be answered.

[0092] In step S230, the prompt transmission means 240 transmits the second prompt generated in step S220 to the large language model 40.

[0093] In step S240, the answer receiving means 250 receives a second answer to the second prompt transmitted in step S230 from the large language model 40. Here, the received second answer is only numerical information.

[0094] In step S250, the command generation means 260 generates a command for controlling the welding machine 10 based on the answer from the large language model 40 received in step S150 and the second answer from the large language model 40 received in step S240.

[0095] As described above, according to the information processing apparatus 200 and the command generation method M200 according to the second embodiment of the present invention, the prompt generation means 230 combines the input data transcribed based on the user's voice with the command data 50 composed of a plurality of commands capable of controlling the welding machine 10 stored in advance to generate a prompt. The prompt transmission means 240 transmits the prompt to the large language model 40, and the response reception means 250 receives the response from the large language model 40. Then, for the command generated based on the response, when the determination means 270 determines that detailed information indicating the degree is required, the prompt generation means 230 further generates a second prompt including the content from which detailed information (numerical information) can be obtained. The prompt transmission means 240 transmits the second prompt to the large language model 40, and the response reception means 250 receives the second response (numerical information) from the large language model 40. The command generation means 260 generates a command for controlling the welding machine 10 based on the response and the second response. As a result, even in a noisy environment where the voice spoken by the user is difficult to recognize, an appropriate command for the welding machine 10 can be generated based on the voice spoken by the user using the large language model 40, and further, a command including detailed information (numerical information) can also be appropriately generated. As a result, the welding machine 10 can operate appropriately according to the appropriate command.

[0096] In this embodiment, a command for increasing the voltage of the welding machine 10 has been described as an example, but the present invention is not limited thereto. For example, a command for changing the current or changing the feeding speed of the welding wire may be used. Further, in the case of a command for a device other than the welding machine, for example, a command for turning off the power after a certain number of seconds or moving a certain number of meters may be used.

[0097] Also, when the speech recognition of detailed information (numerical information) from the voice uttered by the user in a noisy environment fails completely, the response receiving means 250 may receive a second response from the large language model 40, such as "I don't know how many volts should be increased.", and notify the user of it (by voice and / or on the screen display). Then, in order to obtain the detailed information (numerical information) from the user, for example, a question such as "How many volts should be increased?" may be asked.

[0098] Each of the embodiments described above is for facilitating the understanding of the present invention and is not for limiting and interpreting the present invention. Each element included in each embodiment and its arrangement, material, conditions, shape, size, etc. are not limited to those illustrated and can be changed as appropriate. Also, it is possible to partially replace or combine the configurations shown in different embodiments.

Explanation of Reference Numerals

[0099] 10... welding machine, 20... microphone, 21... voice input unit, 22... voice transmission unit, 30... speech recognition system, 40... large language model, 50... command data, 100, 200... information processing device, 110... voice processing means, 120... storage means, 130, 230... prompt generation means, 140, 240... prompt transmission means, 150, 250... response receiving means, 160, 260... command generation means, 270... determination means, M100, M200... command generation methods, S110~S160, S210~S250... steps of the command generation methods M100 and M200

Claims

1. An information processing apparatus that processes commands for controlling a predetermined device, comprising: prompt generation means for generating a prompt by combining command data composed of a plurality of commands capable of controlling the predetermined device with input data generated based on a user's voice; prompt transmission means for transmitting the prompt to a large language model; answer reception means for receiving an answer from the large language model; command generation means for generating a command for controlling the predetermined device based on the answer. An information processing apparatus.

2. The command data is preset according to the predetermined device. The information processing apparatus according to claim 1.

3. The prompt includes content for selecting one command corresponding to the input data from a plurality of commands capable of controlling the predetermined device. The information processing apparatus according to claim 1.

4. The apparatus further comprises determination means for determining whether detailed information indicating a degree is required for the command generated based on the answer. When it is determined that the detailed information is required, the prompt generation means generates a second prompt including content from which detailed information can be obtained from the input data; the prompt transmission means transmits the second prompt to the large language model; the answer reception means receives a second answer from the large language model; the command generation means generates a command including the detailed information for controlling the predetermined device based on the answer and the second answer. The information processing apparatus according to claim 1.

5. The detailed information includes numerical information. The information processing apparatus according to claim 4.

6. An equipment system comprising the information processing apparatus according to any one of claims 1 to 5, and a predetermined device that receives and operates based on the command generated by the information processing apparatus. An equipment system.

Images