Information providing method, information providing device, information providing program, and information providing system

Abstract

To efficiently answer a user's question.SOLUTION: The terminal 10 detects a question of the user about the content being reproduced by voice based on the question from the user, generates an answer to the question based on the question and the content, and outputs the generated answer.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an information providing method, an information providing device, an information providing program, and an information providing system. [Background technology]

[0002] Recently, systems that read aloud (audio playback) literature such as stories are becoming more common. For example, a system is known that receives a request from a user and reads a book using a reading pattern that corresponds to the user's state (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-67685 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the conventional technology does not address any questions that the user may have about the content of the reading, and there is a problem in that the user may not be able to listen to the reading efficiently.

[0005] For example, when listening to a story being read aloud, a user may lose track of the characters and background of the story because the user forgets or becomes confused while listening to the audio. This can be due to, for example, the complexity of the story, the large number of names, or the very short explanation of the background.

[0006] Furthermore, if playback of a reading is interrupted and time passes before it is resumed, the user may forget the details of the reading. In such cases, the user may have to listen to the audio again, but simply rewinding the playback will not allow them to quickly find the relevant part. The user may become frustrated by repeated operations or become confused about the content up to the point they have already listened to, and may lose interest in continuing to listen. This phenomenon can occur not only in book readings, but also in e-learning, telework meetings, and other situations.

[0007] The present invention has been made in view of the above, and has as its object to enable a user to listen to a reading or the like efficiently and comfortably. [Means for solving the problem]

[0008] The information providing method according to the present invention detects a user's question about content being played back as audio, and generates an answer to the question based on the question and the content of the content. [Effects of the Invention]

[0009] According to the present invention, when a user has a question about content, an answer to the question is generated based on the content of the content, so that the user's question can be answered efficiently. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram illustrating an example of the configuration of an information providing system according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating an example of a hardware configuration of a terminal according to the embodiment. [Figure 3] FIG. 3 is a diagram illustrating the operation principle of the embodiment. [Figure 4] FIG. 4 is a diagram showing an overview of the first embodiment. [Figure 5] FIG. 5 is a diagram illustrating an example of a software configuration of the terminal according to the first embodiment. [Figure 6]FIG. 6 is a flowchart showing a processing flow of the terminal according to the first embodiment. [Figure 7] FIG. 7 is a diagram showing an outline of the second embodiment. [Figure 8] FIG. 8 is a diagram illustrating an example of a software configuration of a terminal according to the second embodiment. [Figure 9] FIG. 9 is a diagram illustrating an example of the response database. [Figure 10] FIG. 10 is a flowchart showing a processing flow of the terminal according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of an information providing method, an information providing device, an information providing program, and an information providing system disclosed in the present application will be described in detail with reference to the accompanying drawings. Note that the present invention is not limited to the embodiments described below.

[0012] [First embodiment] The configuration of the information providing system according to the first embodiment will be described with reference to Fig. 1. Fig. 1 is a diagram showing an example of the configuration of the information providing system according to the first embodiment.

[0013] As shown in Fig. 1, the information providing system 1 includes a terminal 10, a playback device 20, and an audio output device 30. The terminal 10 is a smartphone, a PC, or the like. The terminal 10 and the playback device 20 are connected via the Internet, for example. The audio output device 30 is a speaker, an earphone, or the like. The audio output device 30 may be integrated with the terminal 10.

[0014] The terminal 10 receives audio data from the playback device 20. Based on the received audio data, the terminal 10 outputs audio via the audio output device 30. The user listens to the audio output from the audio output device 30. The terminal 10 may output audio based on an audio file downloaded in advance from the playback device 20, or may output audio by so-called streaming playback.

[0015] The playback device 20 transmits, for example, audio data of a book being read aloud. The audio data transmitted by the playback device 20 may be data of a recorded human voice or data of a synthesized voice. Furthermore, the audio data transmitted by the playback device 20 is not limited to data of audio of a book being read aloud, but may also be data of audio of a text other than a book being read aloud, audio uttered in a radio program, or the like.

[0016] Furthermore, the terminal 10 provides an answer to a question of a user listening to the voice. For example, the terminal 10 provides an answer by voice via the voice output device 30. The process by which the terminal 10 provides an answer will be described in detail later. The terminal 10 is an example of an information providing device.

[0017] The hardware configuration of the terminal 10 will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the hardware configuration of the terminal according to the first embodiment. As shown in Fig. 2, the terminal 10 includes a communication unit 11, an input unit 12, an output unit 13, a memory 14, and a controller 15.

[0018] The communication unit 11 is an interface for communicating data between the terminal 10 and other devices. For example, the communication unit 11 is a communication module such as a network interface card (NIC). The input unit 12 is an interface that receives signal input from an input device. For example, the input device is a microphone, a keyboard, a mouse, a touch panel, etc. The output unit 13 is an interface that outputs signals to an output device including an audio output device 30. For example, the output device is a speaker, an earphone, a display, etc.

[0019] The memory 14 is a storage medium such as an eMMC (embedded multi media card), etc. The memory 14 functions as a ROM (read only memory) or a RAM (random access memory).

[0020] The controller 15 reads and executes a program stored in the memory 14. The controller 15 may be a central processing unit (CPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a graphics processing unit (GPU), a system on a chip (SoC), or the like. The controller 15 may be a single processor. The controller 15 may have a multiprocessor configuration. The controller 15 may also have a multicore configuration having multiple cores in a single chip connected via a single socket.

[0021] Here, the operating principle common to each embodiment will be explained using Fig. 3. Fig. 3 is a diagram for explaining the operating principle of the embodiment.

[0022] As shown in Fig. 3, the playback device 20 provides an audio reading service. That is, the playback device 20 provides audio based on the audio information of a work. The user listens to the audio provided via the audio output device 30. The terminal 10 detects a user question via an HMI (Human Machine Interface). Then, upon detecting the user question, the terminal 10 performs a playback stop process on the playback device 20.

[0023] The terminal 10 also stores the audio provided by the playback device 20 as playback information, which is converted into text by speech recognition. The playback information may be the original data of the played portion (e.g., the text of the entire audio) or processed data (e.g., a summary).

[0024] Furthermore, the terminal 10 creates an answer to the question based on the question information and playback information obtained by analyzing the user's question. The terminal 10 outputs the created answer as question-answering audio information. The question-answering audio information is output as audio by the audio output device 30. This allows the user to obtain the answer to the question as audio. Then, after the terminal 10 answers the user's question, it performs playback resume processing on the playback device 20.

[0025] The operation principle of Fig. 3 is common to the multiple embodiments described below, but the specific methods of generating answers differ from one another in the embodiments.

[0026] [First embodiment] In the first embodiment, the terminal 10 generates a response using LLM (Large Language Models). An overview of the first embodiment will be described with reference to Fig. 4. Fig. 4 is a diagram illustrating an overview of the first embodiment.

[0027] 4, the terminal 10 generates a prompt based on the question information obtained by analyzing the user's question and the reproduced text data (an example of reproduced information). The terminal 10 inputs the generated prompt into the LLM to obtain an answer.

[0028] Fig. 5 is a diagram showing an example of the hardware and software configuration of a terminal according to the first embodiment. The functions of the components shown in Fig. 5 are realized by the controller 15 executing programs stored in the memory 14, etc. As shown in Fig. 5, the memory 14 has a reproduced text storage unit 141, a question data storage unit 142, and an LLM command template table 143.

[0029] The reproduced text storage unit 141 stores text data created by performing speech recognition on the speech provided from the playback device 20. For example, the text data is a transcript of the content that has been read aloud. The question data storage unit 142 stores user questions. The questions may not be the data of the user's speech itself, but may be data that has been analyzed and converted to suit subsequent processing. The LLM command template table 143 is a table that stores template sentences for creating prompts.

[0030] The controller 15 also includes an acquisition unit 151 , a voice recognition unit 152 , a question analysis unit 153 , a prompt generation unit 154 , an LLM collaboration unit 155 , an answer output unit 156 , and a playback control unit 157 .

[0031] The acquisition unit 151 acquires data input from the outside or data stored in the memory 14. For example, the acquisition unit 151 acquires audio data of content, user question data, etc. The voice recognition unit 152 performs voice recognition on the voice. The voice recognition unit 152 obtains text data transcribed from the voice through the voice recognition. The text data is stored in the reproduced text storage unit 141. For example, the voice recognition unit 152 performs voice recognition using a machine learning model.

[0032] The question analysis unit 153 analyzes the user's operation and determines whether the user has asked a question. The question analysis unit 153 also extracts the content of the user's question. For example, the question analysis unit 153 extracts a question from the text data output by the voice recognition unit 152 and stores the extracted question text in the question data storage unit 142. The user's operation may be voice input via a microphone or text input via a keyboard or the like.

[0033] The prompt generation unit 154 generates a prompt by inserting a question and played-back text into an LLM command template. The question text is obtained from the question data storage unit 142. The played-back text is obtained from the played-back text storage unit 141. The LLM command template is obtained from the LLM command template table 143. For example, an LLM command template is "Please answer question A's question regarding statement B's statement." The prompt generation unit 154 inserts the question text into A. Furthermore, the prompt generation unit 154 inserts the played-back text into B.

[0034] The LLM cooperation unit 155 inputs a prompt to the LLM and acquires an answer output from the LLM. The prompt generation unit 154 and the LLM cooperation unit 155 function as an answer generation unit that generates an answer to a question. The answer output unit 156 outputs the answer acquired by the LLM cooperation unit 155. For example, the answer output unit 156 outputs the answer to the user by voice. The playback control unit 157 controls playback and stopping of the audio of the content. For example, the playback control unit 157 stops (pauses) playback of the audio from the time the user asks a question until the answer is output.

[0035] It should be noted that the machine learning model such as LLM in the above description may be provided inside the terminal 10, or may be provided in an external server connected to the terminal 10.

[0036] 6 is a flowchart showing the flow of processing of the terminal according to the second embodiment. This flowchart is executed when content playback (reading) is started by a user operation or the like. As shown in FIG. 6, the playback control unit 157 (controller 15) plays back content (step S101). For example, the playback control unit 157 starts or resumes the reading audio.

[0037] Next, if the content playback has not ended (step S102; No), the controller 15 proceeds to step S103. If the content playback has ended (step S102; Yes), the controller 15 ends the process. In step S103, the voice recognition unit 152 (controller 15) performs voice recognition on the voice of the content, and stores text data obtained by the voice recognition in the played text storage unit 141. That is, the controller 15 accumulates information on the played content in a format that can be used later to generate an answer.

[0038] Next, if there is no user operation (step S104; No), the controller 15 returns to step S102. If there is a user operation (step S104; Yes), the question analysis unit 153 (controller 15) analyzes the user operation content (step S105). During the analysis, the user operation content is converted into text data. For example, if it is a voice operation, it is converted into text data by voice recognition, and if it is input by keyboard operation or the like, it is converted into text data using, for example, a conversion table between operation content and text data.

[0039] Next, the question analysis unit 153 determines whether the user's operation was a question (step S106). If the user's operation was not a question (step S106; No), the controller 15 returns to step S102. If the user's operation was a question (step S106; Yes), the prompt generation unit 154 (controller 15) generates a prompt (step S107). Specifically, the prompt generation unit 154 generates the prompt by inserting the question and the reproduced text of the content stored in the reproduced text storage unit 141 into an LLM command template. At this time, the controller 15 also stops the reproduction of the content.

[0040] When generating a prompt, the question analysis unit 153 converts the user's operation (voice) into text data using language analysis processing, voice recognition, or the like, and then converts the text data into text data in an expression format suitable for the prompt to be input to the LLM so that the user's question is clearer, that is, in a format suitable for the prompt to be input to the LLM. For example, suppose the user's question is, "Who is Urashima Taro?" The question analysis unit 153 converts this question to, for example, "Please tell me about Urashima Taro's profile" so that the intent of the question is clearer. The question analysis unit 153 stores the converted question in the question data storage unit 142.

[0041] The LLM collaboration unit 155 (controller 15) sends a prompt to the LLM (step S108) and receives an answer from the LLM (step S109). The answer output unit 156 (controller 15) outputs the answer received from the LLM by audio playback (step S110). For example, the answer output unit 156 outputs the answer received from the LLM, "Urashima Taro is a fisherman. However, the answer to the question changes as playback progresses," as an answer to the user. The controller 15 then waits until the answer output is complete (step S111; No), and when the answer output is complete (step S111; Yes), returns to step S101.

[0042] In this way, the controller 15 detects a user's doubt about the content being played back based on the question from the user, generates an answer to the question based on the question and the content of the content, and outputs the generated answer. As a result, when a user has a question about the content, the controller 15 generates an answer to the question based on the content of the content, so that the user's question can be answered efficiently.

[0043] The controller 15 also generates a prompt for inputting the LLM, which includes a question, the content (played portion), and a command statement for answering the question about the content. The controller 15 also inputs the prompt into the LLM, and uses the LLM answer to the prompt obtained from the LLM as the answer. By using the LLM, the controller 15 can easily obtain an answer in a natural language format that is easy for the user to understand.

[0044] Furthermore, questions (prompts) posed to the LLM are based on the text data of the already played portion of the content. Therefore, the LLM's answers are based on the content of the already played content, and the answers to the user are also similar. This prevents the user from being informed of the content of the unplayed portion, i.e., from being spoiled.

[0045] Furthermore, if the controller 15 detects that the user has a question about the content (for example, if it is determined in step S106 that the user is asking a question), it stops the audio playback of the content and resumes the audio playback of the content after the question is answered. This prevents the user from continuing to listen to the content while still having a question, and also allows the user to listen to the content again without any special operation after the question is resolved.

[0046] [Second embodiment] In the second embodiment, a database (DB) is used to generate a response. An outline of the second embodiment will be explained using Fig. 7. Fig. 7 is a diagram showing an outline of the second embodiment. In the explanation of the second embodiment, explanations of parts common to the first embodiment will be omitted as appropriate.

[0047] The first and second embodiments differ in the process of generating an answer. As shown in Fig. 7, the terminal 10 generates a prompt for creating an answer database based on the reproduced text data and the answer table generation LLM command statement table. The terminal 10 extracts an answer based on the user's question from the created answer database.

[0048] 8 is a diagram illustrating an example of the hardware and software configuration of a terminal according to the second embodiment. As shown in FIG. 8, the memory 14 stores an answer table generation LLM command statement table 145 instead of the LLM command fixed phrase table 143 of the first embodiment. The memory 14 also stores an answer database 146. The answer table generation LLM command statement table 145 is a table that stores fixed phrases for creating prompts. However, the prompts are prompts for generating the answer database 146.

[0049] FIG. 9 is a diagram showing an example of the answer database. As shown in FIG. 9, the answer database 146 is information in which questions and answers are associated with each other. For example, in the answer database 146, a question about "Urashima Taro" is associated with the answer "fisherman." The answer database 146 changes (is updated) as the playback of the content progresses. Each record in the answer database 146 may also be stored in association with the playback time of the content. In this case, it becomes possible to read out the question and answer from the answer database 146 by specifying the playback time. In other words, the playback time is data indicating that the answer is based on the content that has been played up to that playback time.

[0050] While the prompt generation unit 154 in the first embodiment creates prompts for generating answers, the prompt generation unit 154 in the second embodiment creates prompts for generating (updating) the answer database 146. The prompt generation unit 154 generates a prompt by inserting the played-back text and question items stored in the played-back text storage unit 141 into an LLM command template. The played-back text is acquired from the played-back text storage unit 141. The LLM command template is acquired from the answer table generation LLM command statement table 145.

[0051] For example, an LLM command template is "Please generate a database of answers to questions for each item C in response to sentence B." The prompt generation unit 154 inserts the reproduced text into B. The prompt generation unit 154 also inserts a predetermined item into C. For example, the predetermined items are expected questions about the reproduced content, such as "profiles of characters," "profiles of featured objects (animals and facilities)," and "summary," and are appropriately set by, for example, a developer.

[0052] The LLM collaboration unit 155 inputs a prompt to the LLM and obtains the answer output from the LLM in response to the prompt. The answer here is not an answer to the user's question, but an answer for the database corresponding to the LLM command template. The LLM collaboration unit 155 reflects the answer from the LLM in the answer database 146.

[0053] Furthermore, the controller 15 has an answer extraction unit 158 ​​in addition to the configuration of the first embodiment. The answer extraction unit 158 ​​searches the answer database 146 with the content of the user's question and obtains an answer to the question. The answer output unit 156 outputs the answer extracted by the answer extraction unit 158. For example, the answer output unit 156 outputs the extracted answer to the user by voice.

[0054] Fig. 10 is a flowchart showing the flow of processing of the terminal according to the second embodiment. This flowchart is executed when content playback (reading) is started by a user operation or the like. As shown in Fig. 10, the playback control unit 157 (controller 15) plays back content (step S201). For example, the playback control unit 157 starts or resumes the reading audio.

[0055] Next, if the content playback has not ended (step S202; No), the controller 15 proceeds to step S203. If the content playback has ended (step S202; Yes), the controller 15 ends the process. In step S203, the voice recognition unit 152 (controller 15) performs voice recognition on the voice of the content, and stores text data obtained by the voice recognition in the played text storage unit 141. That is, the controller 15 accumulates information on the played content in a format that can be used later in generating an answer database.

[0056] Next, if it is not time to update the answer database 146 (step S204; No), the controller 15 returns to step S202. For example, the time to update the answer database 146 is when a predetermined time set in advance by a developer or the like has elapsed since the last update. If it is time to update the answer database 146 (step S204; Yes), the prompt generation unit 154 (controller 15) generates a prompt for generating (updating) the answer database 146 as described above (step S205).

[0057] The LLM collaboration unit 155 (controller 15) sends the generated prompt to the LLM (step S206) and receives an answer from the LLM (step S207). Then, the LLM collaboration unit 155 updates the answer database 146 with the answer from the LLM (step S208). That is, the LLM collaboration unit 155 adds questions and answers to the answer database 146, and changes or deletes questions and answers in the answer database 146.

[0058] Next, if there is no user operation (step S209; No), the controller 15 returns to step S202. If there is a user operation (step S209; Yes), the answer extraction unit 158 ​​(controller 15) searches the answer database 146 for an answer based on the question content. Note that searching for an answer to the question content is an example of analyzing a user operation. At this time, the controller 15 stops the playback of the content. For example, suppose the user's question is "Who is Urashima Taro?" The answer extraction unit 158 ​​regards "Urashima Taro" as a keyword for the question content (the object to be explained), and searches the answer database 146 (question data in the answer database 146) for the keyword (step S212).

[0059] The answer output unit 156 (controller 15) outputs the answer obtained by the search of the answer extraction unit 158 ​​by audio playback (step S213). Next, the controller 15 waits until the answer output is completed (step S214; No), and when the answer output is completed (step S214; Yes), the process returns to step S201.

[0060] In this way, the controller 15 generates an answer database 146 that associates questions with answers to those questions based on the content of the played content, and searches the answer database 146 using the detected user question to generate an answer to the question. This allows the controller 15 to provide an answer quickly. That is, the controller 15 generates a prompt for inputting an LLM that includes a predetermined question item that may be a question, the content of the content, and an instruction statement to ask a question item related to the content content. The controller 15 inputs the prompt into the LLM, and stores the LLM answer to the prompt obtained from the LLM and the question item in the answer database 146. In this way, by generating the answer database 146 in advance using the LLM, the controller 15 can easily and quickly obtain an answer when a user asks a question.

[0061] [Other embodiments] The content is not limited to book readings. For example, the controller 15 may accumulate summaries of speech in a meeting or conversation. In this case, the controller 15 outputs answers to questions about the meeting or conversation. In this case, the controller 15 stores the speaker's utterances in the reproduced text storage unit 141. For example, in response to the question, "What did Mr. A say?", the controller 15 inputs a prompt to the LLM, "Please extract Mr. A's previous utterances." This allows the controller 15 to obtain an answer to the person's utterance content.

[0062] In a conference or dialogue, participants do not always listen carefully to everything being said. According to the embodiment, even in such a case, participants can easily check the contents of the conference. In this case, each function of the controller 15 may be implemented as part of an application of the audio Web conference system.

[0063] Furthermore, instead of acquiring a user's question, the controller 15 may generate an answer triggered by a change in the emotions of the conference participants. When the controller 15 detects that a conference participant is feeling anxiety or impatience, the controller 15 can output a voice suggesting the creation of an answer. For example, the controller 15 outputs a voice saying, "Did you miss something? Shall I provide a summary of what has been said so far?" In response to this, if an operation indicating affirmative action is performed (such as pressing a button or inputting "yes"), the controller 15 creates and outputs a summary of what has been said in the conference so far as an answer. The controller 15 can estimate the emotions of the participants using a known method (see, for example, Japanese Patent Application Laid-Open No. 2022-134929).

[0064] Furthermore, the configuration of the information providing system 1 is not limited to those shown in Figs. 5, 8, etc. For example, the functions of each processing unit of the controller 15 may be provided in a server different from the terminal 10. For example, a controller capable of executing the functions of the voice recognition unit 152, question analysis unit 153, prompt generation unit 154, and LLM collaboration unit 155 may be provided in an answer creation server capable of communicating with the terminal 10. Furthermore, the answer creation server stores the played text storage unit 141, etc. In this case, the terminal 10 transmits audio data of the content and the user's question to the answer creation server. The terminal 10 can obtain an answer to the question from the answer creation server and output the audio of the answer. This reduces the processing load on the terminal 10.

[0065] Furthermore, the terminal 10 may have the same functions as the playback device 20. In this case, instead of receiving audio data from the playback device 20, the terminal 10 can output the audio of the content based on audio data stored in itself.

[0066] Further advantages and modifications will readily occur to those skilled in the art. Therefore, the invention in its broader aspects is not limited to the specific details and representative embodiments shown and described above. Accordingly, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents. [Explanation of symbols]

[0067] 1 Information provision system 10 devices 11 Communications Department 12 Input section 13 Output section 14 Memory 15 Controller 20 Playback device 30 Audio output device 141 Played text storage unit 142 Question data storage unit 143 LLM Directive Template Table 151 Acquisition Department 152 Voice Recognition Unit 153 Question Analysis Department 154 Prompt Generation Unit 155 LLM Collaboration Department 156 Answer output section 157 Playback control unit 158 Answer extraction part

Claims

1. Detecting a user's doubt about the content being played back; Generate an answer to the question based on the question and the content of the content; outputting the generated answer; The information delivery method performed by the controller.

2. The controller Generate an answer to the question based on the content that has been played back. The information providing method according to claim 1 .

3. The controller If a user's doubt about the content is detected, audio playback of the content is stopped.

3. The information providing method according to claim 1 or 2.

4. The controller generating a first prompt for LLM input including the question, the content, and an instruction to answer the question regarding the content; Entering the first prompt into the LLM; The answer is the LLM response to the first prompt obtained from the LLM.

3. The information providing method according to claim 1 or 2.

5. The controller generating a database that associates questions with answers to the questions based on the content; Searching the database based on the detected user's question to generate an answer to the question.

3. The information providing method according to claim 1 or 2.

6. The controller generating a second prompt for LLM input including a question item that may be a predetermined question, the content of the content, and an instruction sentence to ask the question item regarding the content of the content; Entering said second prompt into the LLM; storing the LLM answer to the second prompt and the question item obtained from the LLM in the database; The information providing method according to claim 5.

7. An information providing device that provides information related to audio playback content, Detecting a user's doubt about the content being played back; Generate an answer to the question based on the question and the content of the content; outputting the generated answer; Information provision device.

8. An information providing device that provides information related to audio playback content, Detecting a user's doubt about the content being played back; generating a third prompt for inputting the LLM, the third prompt including the question, the content, and an instruction to answer the question regarding the content; sending the generated third prompt to an LLM server; obtaining an answer to the third prompt from the LLM server; outputting the obtained answer; Information provision device.

9. Detecting a user's doubt about the content being played back; Generate an answer to the question based on the question and the content of the content; outputting the generated answer; An information providing program executed by the controller.

10. Detecting a user's doubt about the content being played back; generating a third prompt for inputting the LLM, the third prompt including the question, the content, and an instruction to answer the question regarding the content; sending the generated third prompt to an LLM server; obtaining an answer to the third prompt from the LLM server; outputting the obtained answer; An information providing program executed by the controller.

11. An information providing system including a terminal that provides a user with a playback sound of content, and a server device that is communicatively connected to the terminal, The terminal Detecting a user's doubt about the content during audio playback; transmitting the detected question to the server device; Obtaining an answer to the question from the server device; Outputting the obtained answer; The server device Acquire the content of the content; acquiring the question from the terminal; Generate an answer to the question based on the question and the content of the content; Transmitting the generated response to the terminal; Information provision system.

12. An information providing system including a terminal that provides a user with a playback sound of content, and an LLM server device that is connected to the terminal for communication, The terminal generating a first prompt for LLM input including a user's question about the content being played back, the content itself, and an instruction to answer the question about the content itself; transmitting the generated first prompt to the LLM server device; obtaining an answer to the first prompt from the LLM server device; Outputting the obtained answer; The LLM server device obtaining the first prompt from the terminal; generating the answer to the first prompt; Transmitting the generated response to the terminal; Information provision system.

Images