Attention calling device, attention calling system and program

Abstract

The present invention makes it possible to accurately identify a section in which conversation is performed at a level equivalent to noise among a plurality of sections into which a prescribed space is divided. An ordering tablet terminal (3) is disposed on each table in a restaurant hall. A shop management device (10) comprises: a conversation extraction unit (12) which analyzes sound data collected from each table terminal (3) and extracts conversation performed in the hall; a sound volume information generation unit (13) which generates, from each of the tablet terminals (3), sound volume information including a sound volume of each conversation included in each of the sound data; a table position identification unit (14) which identifies, on the basis of the sound volume information, a table at which a loud conversation is taking place; and a notification unit (15) which sends an alert notification to the identified table. The table position identification unit (14) excludes a table from becoming a target recipient of an alert when a conversation at the table is not greater than a prescribed threshold value, even when the tablet terminal (3) collects sound data of a sound volume that is greater than the threshold value.

Description

Technical Field

[0001] The present invention relates to an alert device, an alert system, and a program, and particularly to alerting a person in a section where noise is being generated in a predetermined space.

Background Art

[0002] In infectious diseases where infection is derived from droplets, it is considered effective as an infection prevention measure not to scatter droplets.

[0003] In stores such as restaurants, it is not uncommon for conversations between close friends to flourish, and when the conversation gets lively, people tend to talk loudly without realizing it. As a result, it may lead to scattering of droplets. Even if a store clerk notices such a situation where people are talking loudly, it is difficult to give a warning due to their position.

[0004] Therefore, conventionally, a sectional noise meter has been installed for each table seat installed in a store to correspond to a partition without a partition. When a noise level equal to or higher than a certain level is detected in any section, a technique has been proposed to notify the notification device provided in that section that noise is being generated (for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0006] However, it is conceivable that the compartment noise meter may pick up voices from other compartments. That is, in a certain compartment, even if a conversation is taking place at a level that does not reach a predetermined noise level, the compartment noise meter corresponding to that compartment may pick up a loud voice emitted in a neighboring compartment, resulting in the possibility of erroneously designating that compartment as a target for noise generation notification.

[0007] An object of the present invention is to correctly identify a compartment in which a conversation is being held at a level corresponding to noise from among compartments formed by dividing a predetermined space into a plurality of parts.

Means for Solving the Problem

[0008] The attention - calling device according to the present invention includes: a collecting means for collecting sound data from sound - collecting means installed in association with each of the compartments formed by dividing a predetermined space into a plurality of parts; a conversation extraction means for extracting a conversation being held in the space by analyzing the sound data collected by the collecting means; a compartment identification means for identifying a compartment in which a conversation is being held at a volume exceeding a predetermined threshold from among the compartments in which the sound - collecting means serving as the source of the sound data exceeding the predetermined threshold is installed; and a notification means for notifying the occurrence of noise to a notification means installed in the compartment identified by the compartment identification means.

[0009] Further, the conversation extraction means is characterized in that it extracts a conversation being held in the space by referring to at least one of the start and end timings of the speech of a person in the space or the words and phrases included in the speech.

[0010] Further, the compartment identification means is characterized in that it determines that the conversation is being held in the compartment corresponding to the sound - collecting means that has collected the sound data with the maximum volume among the conversations extracted by the conversation extraction means.

[0011] Further, the predetermined space is a store, the partition is formed corresponding to each table seat arranged in the store, and the sound collection means and the notification means are terminal devices used when providing services to users of the partition at the table seat. This is the gist of the invention.

[0012] The alert system according to the present invention is a terminal device installed in association with each of a plurality of partitions formed by dividing a predetermined space, and is a terminal device having a sound collection means and a notification means, and an alert device communicably connected to each of the terminal devices. The alert device includes a collection means for collecting sound data collected by the sound collection means from each of the terminal devices, a conversation extraction means for extracting conversations being conducted in the space by analyzing the sound data collected by the collection means, a section identification means for identifying a section in which a conversation is being conducted at a volume exceeding a predetermined threshold from among the sections in which the terminal device serving as the source of the sound data having a volume exceeding the predetermined threshold is installed, and a notification means for notifying the terminal device installed in the section identified by the section identification means of the occurrence of noise.

[0013] The program according to the present invention causes a computer to function as a collection means for collecting sound data from a sound collection means installed in association with each of a plurality of partitions formed by dividing a predetermined space, a conversation extraction means for extracting conversations being conducted in the space by analyzing the sound data collected by the collection means, a section identification means for identifying a section in which a conversation is being conducted at a volume exceeding a predetermined threshold from among the sections in which the sound collection means serving as the source of the sound data having a volume exceeding the predetermined threshold is installed, and a notification means for notifying a notification means installed in the section identified by the section identification means of the occurrence of noise.

Advantages of the Invention

[0014] According to the present invention, it is possible to correctly identify a section in which a conversation is being conducted at a level corresponding to noise from among a plurality of sections formed by dividing a predetermined space.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Modes for Carrying Out the Invention

[0016] Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

[0017] FIG. 1 is a schematic configuration diagram of a store to which an embodiment of a warning system according to the present invention is applied. In the present embodiment, in order to describe a restaurant as an example of a store, FIG. 1 shows a schematic diagram of a hall 1 of a restaurant with some parts omitted. The hall 1 forms a predetermined space. In the restaurant, services are provided to customers in the hall 1. In the hall 1, table seats 2a to 2d where customers sit are arranged neatly. Note that when it is not necessary to distinguish between the table seats 2a to 2d, they are collectively referred to as "table seat 2" by omitting the subscript. In the following description, the same applies to other components with subscripts in the reference numerals.

[0018] A plurality of table seats 2 are arranged in the hall 1, and partitions are formed by being divided for each table seat 2. In FIG. 1, four table seats 2a to 2d are illustrated, and partitions are formed corresponding to each of the table seats 2a to 2d. A partition plate may be provided between the partitions, but in the present embodiment, since no partition plate is provided, the hall 1 is in an environment where conversations of adjacent seats can be easily heard.

[0019] Tablets 3 are respectively arranged on the table seats 2. The tablet 3 is a terminal device used when providing services to the users of the table seats 2. The tablet 3 can be realized with a general-purpose hardware configuration that has existed conventionally. That is, the tablet 3 has a CPU, a ROM, a RAM, a storage as a storage means, a network interface as a communication means, and a touch panel as a user interface. Further, the tablet 3 incorporates a microphone as a sound collection means and a speaker which is a form of a notification means together with the touch panel. An application specialized for providing services in a restaurant is installed on the tablet 3. In the case of a restaurant, the tablet 3 is generally used for ordering and the like.

[0020] As described above, in the present embodiment, the partition, the table seat 2, and the tablet terminal 3 are associated with each other in a one-to-one-to-one manner. Therefore, for example, specifying the tablet terminal 3 is synonymous with specifying the table seat 2 and the partition. Therefore, in the following description, it is also possible to read "partition", "table seat", and "tablet terminal" interchangeably.

[0021] In the hall 1, an access point (AP) 4 that is wirelessly connected to each tablet terminal 3 is installed. In the store, further, a store management device 10 is installed at a predetermined location, generally in a room such as a staff room where customers are prohibited from entering. The store management device 10 is an information processing device equipped with functions such as sales management and staff schedule management in a restaurant. In the present embodiment, it is a computer used as a warning device.

[0022] The store management device 10 in the present embodiment can be realized with a general-purpose hardware configuration that has existed conventionally, such as a personal computer (PC). That is, the store management device is configured by connecting a CPU, a ROM, a RAM, a hard disk drive (HDD) as a storage means, a network interface as a communication means, an input means such as a mouse and a keyboard, and a display means such as a display to an internal bus.

[0023] Figure 2 is a block configuration diagram of the store system in the present embodiment. The store system has the tablet terminal 3 and the store management device 10 described above. Since the plurality of tablet terminals 3 each have the same functions, only one is shown in Figure 2. The tablet terminal 3 and the store management device 10 can communicate with each other via the access point 4.

[0024] The tablet terminal 3 has a sound collection unit 31, a display unit 32, and a sound output unit 33. Note that components not used in the description of this embodiment are omitted from the figures. The sound collection unit 31 is realized by the microphone described above and collects ambient sound. The display unit 32 is realized by the touch panel described above and performs a display for calling attention to conversations at a loud volume. The sound output unit 33 is realized by the speaker described above and outputs a sound for calling attention to conversations at a loud volume.

[0025] The store management device 10 has a sound data collection unit 11, a conversation extraction unit 12, a volume information generation unit 13, a table position identification unit 14, a notification unit 15, a table information storage unit 16, a sound data information storage unit 17, a conversation information storage unit 18, and a volume information storage unit 19. Note that components not used in the description of this embodiment are omitted from the figures.

[0026] The sound data collection unit 11 collects the sound data collected by each tablet terminal 3 and stores it in the sound data information storage unit 17. The conversation extraction unit 12 analyzes the sound data collected by the sound data collection unit 11 to extract the conversations taking place in Hall 1. The conversation extraction unit 12 also generates conversation information by associating the sound data including the extracted conversation for each extracted conversation and stores it in the conversation information storage unit 18. The volume information generation unit 13 generates volume information indicating the volume of each conversation included in the sound data and stores it in the volume information storage unit 19. The table position identification unit 14 functions as a section identification means and identifies the section in which conversations are taking place at a volume exceeding a predetermined threshold from among the sections where the tablet terminal 3 serving as the collection source of the sound data having a volume exceeding the predetermined threshold is installed. The notification unit 15 notifies the tablet terminal 3 installed in the section identified by the table position identification unit 14 of the occurrence of noise.

[0027] FIG. 3 is a diagram showing an example of the data configuration of the table information stored in the table information storage unit 16 in the present embodiment. The table information is information regarding the table seat 2 arranged in Hall 1, and it is necessary to set it in advance before performing the attention-calling process described later. The table information includes a table number, the number of people, position information, a terminal number, and address information. The table number is information for identifying the table seat 2. The number of people is the number of customers who can use the table seat 2. The position information is information for specifying the position of the table seat 2 in Hall 1. When represented by two-dimensional coordinates with a point in Hall 1, for example, a corner of the room as the origin, the position information is represented by coordinate data indicating the position of a point of the table seat 2, for example, the center point of the table. The terminal number is identification information of the tablet terminal 3 installed on the table seat 2. The address information is information necessary for communicating with the tablet terminal 3 installed on the table seat 2. For example, an IP address is used as the address information.

[0028] In the present embodiment, the case where the table seat 2 and the tablet terminal 3 are in a one-to-one relationship will be described as an example. However, by setting the information of a plurality of tablet terminals 3 in the terminal number and address information of the table information, it is also possible to associate a plurality of tablet terminals 3 with one table seat 2.

[0029] The various information stored in the other storage units 17 to 19 will be described together with the description of the operation.

[0030] Each component 11 to 15 in the store management device 10 is realized by the cooperative operation of a computer forming the store management device 10 and a program operating on a CPU mounted on the computer. Further, each storage unit 16 to 19 is realized by an HDD mounted on the store management device 10. Alternatively, a RAM or an external storage means may be used via a network.

[0031] In addition, the program used in this embodiment can be provided not only by communication means but also by being stored in a computer-readable recording medium such as a CD-ROM or a USB memory and provided. The program provided from the communication means or the recording medium is installed in a computer, and various processes are realized by the CPU of the computer sequentially executing the program.

[0032] Next, the operation in this embodiment will be described.

[0033] In this embodiment, a customer who visits a restaurant has a meal at any of the table seats 2. As time passes, among close friends, the conversation may become lively and they may start talking loudly. In order to prevent the spread of infectious diseases derived from droplets such as COVID-19, the store wants to minimize loud conversations, but it is difficult for the store staff to pay attention due to their position. Therefore, in this embodiment, the attention is called mechanically instead of the store staff doing it. Hereinafter, the attention-calling process in this embodiment will be described using the flowchart shown in FIG. 4. Here, it is described assuming that the customers are using the table seats 2a to 2d.

[0034] When it is detected that the customer has arrived at the table seat 2 and started eating or drinking by operating the tablet terminal 3, the sound collection unit 31 in the tablet terminal 3 starts sound collection.

[0035] The sound data collection unit 11 in the store management device 10 collects sound data from the tablet terminal 3 of the seated table seat 2 and stores it in the sound data information storage unit 17 (step 110). The collected sound data includes the voices of the customers talking.

[0036] FIG. 5 is a diagram showing an example of the data configuration of the sound data information stored in the sound data information storage unit 17 in the present embodiment. As shown in FIG. 5, the sound data information is stored in pairs of the terminal number of the tablet terminal 3, which is the source of the sound data collection, and the identification information of the sound data. Note that the identification information of the sound data may be the sound data itself, the storage location information of the sound data, the file name of the sound data, or the like. The sound data will be constantly collected by the sound collection unit 31 while the customer is at the table seat 2. However, the sound data collection unit 11 may collect the sound data from the sound collection unit 31, for example, every time a predetermined time period elapses, for example, every five minutes. In this case, the start period and the end period of the collection period may be added to the sound data and recorded.

[0037] Subsequently, the conversation extraction unit 12 extracts the human voice by taking out the sound data from each tablet terminal 3 with a predetermined time length from the sound data information storage unit 17 and analyzing it, and extracts the conversation being held in the hall 1 by grouping (step 120). Then, the conversation extraction unit 12 generates conversation information and stores it in the conversation information storage unit 18.

[0038] FIG. 6 is a diagram showing an example of the data configuration of the conversation information stored in the conversation information storage unit 18 in the present embodiment. The conversation information is generated by associating the conversation number that identifies each conversation held in the hall 1 with the identification information of the sound data including the conversation.

[0039] Incidentally, although it may depend on the sensitivity and directivity of the microphone, the volume of the customer's voice, etc., the microphone of the tablet terminal 3 may pick up not only the table seat 2 where the tablet terminal 3 is installed, but also the voices of conversations taking place at the surrounding table seats 2. That is, a conversation carried out by a group of customers may be picked up not only by the tablet terminal 3 installed at the table seat 2 used by that group of customers, but also by, for example, the tablet terminal 3 at the adjacent table seat 2. FIG. 6 shows that the conversation of a group of customers is included in a plurality of sound data. That is, it shows that the conversation of a group of customers is picked up by the microphones of a plurality of tablet terminals 3. Incidentally, the fact that a conversation is picked up by a plurality of microphones also means that the conversation is likely to be a relatively loud voice, that is, a loud voice.

[0040] Here, the classification of the customers having a conversation, in other words, the association between the conversation and the customers having the conversation, will be described with reference to the diagram shown in FIG. 7. FIG. 7 shows the voice of each customer obtained by analyzing the sound data. In FIG. 7, the voices of six customers A to F are extracted. In FIG. 7, the vertical direction represents time, and the rectangle 21 shown corresponding to the customer indicates the time when the customer is speaking.

[0041] Generally, conversation is a situation where multiple people (referred to as "customers" here) talk to each other. Therefore, while one customer in a group of customers is speaking, the other customers in the same group usually listen. When the conversation pauses, it is normal for any one of the other customers in the same group to start speaking. In Figure 7, an example where Customer B's speech 21b starts at the timing when Customer A's speech 21a ends corresponds to this. Of course, there may be cases where there is a gap, such as between Customer A's speech 21c and Customer B's speech 21d. Conversely, there may also be cases where one person starts speaking before another person's speech ends, such as between Customer C's speech 21e and Customer B's speech 21f. There may be various other patterns, but basically, since conversation is an interaction of speaking, it is considered that some relationship occurs at each timing of the start or end of the speech of customers in the same group. Therefore, according to the example of the timing of the customers' speeches shown in Figure 7, it can be inferred that Customers A to D are included in the same group of customers. Similarly, since the start and end timings of the speeches of Customers E and F can be recognized as an interaction of words, it can be inferred that Customers E and F are included in the same group of customers.

[0042] On the other hand, for example, since the speech timing of Customer E overlaps with that of Customers A, B, and C, it can be inferred that Customer E is having a different conversation, that is, Customer E is included in a group of customers different from Customers A, B, and C.

[0043] In this embodiment, if the time difference from the end of a speech to the start of another person's speech is within a predetermined threshold, it is determined that the customers are participating in the same conversation. In this way, the conversation extraction unit 12 extracts the conversation.

[0044] The above is the case of classifying and extracting the conversations taking place in Hall 1 by grouping each customer based on the timing of each customer's speech. In addition, by performing speech recognition on each customer's speech to convert the spoken words into a character string, and further identifying the genre or topic of the conversation, etc., the customers may be grouped according to the identified genre, etc., and the conversation may be extracted. The conversation extraction unit 12 may adopt at least one of these extraction methods.

[0045] According to this embodiment, as described above, it is possible to extract the conversation that the customer is having. When there are many customers, it may not be easy to group the customers in Hall 1 and extract the conversation with only a few exchanges of words. However, usually, since the customers sit at Table 2 for a relatively long time and have conversations, it is considered possible to extract the conversation from the voices of the customers.

[0046] As described above, when the conversation is extracted from the audio data, subsequently, the volume information generation unit 13 acquires the volume of each conversation included in each audio data to generate volume information, and stores it in the volume information storage unit 19 (step 130). Subsequently, the table position identification unit 14 refers to the volume information to identify Table 2 where a conversation is being held loudly (step 140).

[0047] FIG. 8 is a diagram showing an example of the data configuration of the volume information stored in the volume information storage unit 19 in this embodiment. In the volume information, the volume of each conversation included in each audio data is set for each audio data. Note that in FIG. 8, the larger the volume, the larger the numerical value is set. The numerical values shown in FIG. 8 are values indicating the relative magnitude of the volume, and are not numerical values indicating a specific volume such as decibels. During a predetermined time period, there are differences in the volume of voices, and there may also be silent periods. The numerical values indicated by the volume information are obtained by taking the average value of these voice situations. In FIG. 6, each audio data including the conversation is associated with the conversation, but in FIG. 8, the volume of the conversations included in each classified audio data is shown.

[0048] Conversations are basically carried out when customers are seated at any of the table seats 2. When there are a large number of customers, they may be divided among multiple table seats 2. A tablet terminal 3 is arranged corresponding to each table seat 2. However, the tablet terminal 3 arranged at the table seat 2 where the conversation is taking place should collect the conversation at the highest volume. This is because for a single conversation, the tablet terminal 3 that picks up the highest volume is the closest to the customers having the conversation. Therefore, in a single conversation, it can be determined that the conversation is taking place at the table seat 2 from which the sound data with the highest volume is collected.

[0049] According to the setting example shown in FIG. 8, for example, for the conversation with conversation number "G2" (hereinafter simply referred to as "conversation G2"; the same applies to other conversations), among the sound data, the sound data v03 is collected at the highest volume of "5". Therefore, by referring to the sound data information and the table information, it can be determined that conversation G2 is taking place at table seat 2c with table number "TA03" associated with the sound data v03. Similarly, for conversation G3, among the sound data, the sound data v04 is collected at the highest volume of "4". Therefore, by referring to the sound data information and the table information, it can be determined that conversation G3 is taking place at table seat 2d with table number "TA04" associated with the sound data v04.

[0050] Also, for conversation G1, among the sound data, the sound data v01 is collected at the highest volume of "10". By referring to the sound data information and the table information, it can be determined that conversation G1 is taking place at table seat 2a with table number "TA01" associated with the sound data v01.

[0051] By the way, if we only focus on the maximum volume of the conversation as described above, the audio data v02 may not be selected. However, when we focus on the audio data v02, among the conversations G1 and G2 included in the audio data v02, conversation G1 has the maximum volume. Therefore, it can be determined that conversation G1 is taking place at the table seat 2b associated with the audio data v02. That is, it can be determined that conversation G1 is taking place across the two table seats 2b and 2c. In the above manner, conversations G1 to G3 can be associated with table seats 2a to 2d. In FIG. 1, the groups of customers having conversations G1 to G3 are associated with table seats 2a to 2d by dashed lines.

[0052] Here, for example, assume that the volume of conversation G1 included in the audio data v03 is not "4" but "6". Then, since conversation G1 has the maximum volume among the conversations included in the audio data v03, it might be considered that conversation G1 is also taking place at the table seat 2c associated with the audio data v03.

[0053] However, since conversation G2 is collected at the maximum volume in the audio data v03, the association between the conversation and the table seat 2 will be prioritized. Incidentally, if the volume of conversation G1 included in the audio data v03 is "6", which is greater than the volume "4" of conversation G2, it means that conversation G2 taking place at the table seat 2c is quieter than conversation G1 taking place nearby.

[0054] Here, the point to note is the total shown in the rightmost column of the table of volume information shown in FIG. 8. The total in the volume information is the total volume of the sound collected by the tablet terminal 3 associated with the audio data. Suppose a predetermined threshold for determining whether a person is speaking loudly is "8". In this case, the audio data v1, v2, and v3, whose numerical values exceed "8", that is, the table seats 2a, 2b, and 2c, are considered to be candidates for objects of concern.

[0055] However, the tablet terminal 3c at the table seat 2c picked up not only the conversation G2 at volume "5" in the associated table seat 2c, but also the conversation G1 at volume "4", so it picked up sound at a volume exceeding the threshold "8" which is the total criterion for a loud conversation.

[0056] Therefore, although the sound data v01, v02, v03 at volumes exceeding a predetermined threshold are picked up at the table seats 2a, 2b, 2c respectively, the table position specifying unit 14 in the present embodiment excludes the table seat 2c where the conversation G2 is not being held at a volume exceeding the predetermined threshold, and specifies that the table seats 2a, 2b where the conversation is being held at a volume exceeding the predetermined threshold are the ones where the conversation G1 is taking place.

[0057] In the present embodiment, it is possible to correctly specify that the conversations being held as the objects of attention calls are taking place at the table seats 2a, 2b.

[0058] As described above, when the table seats 2a, 2b where loud conversations are taking place are specified, the notification unit 15 refers to the table information, acquires the address information of the tablet terminals 3a, 3b installed at the table seats 2a, 2b which are the objects of attention calls, and notifies the tablet terminals 3a, 3b that noise is being generated, that is, that a loud conversation exceeding a predetermined threshold is taking place. Thereby, a notification for attention call is made to the table seats 2a, 2b which are the objects of attention calls (step 150).

[0059] When the tablet terminals 3a, 3b receive the notification that noise is being generated, they give an attention call to the group customers using the table seats 2a, 2b. Various methods of giving an attention call can be considered.

[0060] For example, a message asking to suppress loud conversations or an alert is displayed on the touch panels of the tablet terminals 3a, 3b. Or, the touch panel may be made to blink to notify that a loud conversation is taking place. Or, an audio message may be output from the speakers of the tablet terminals 3a, 3b.

[0061] As described above, according to the present embodiment, it is possible to mechanically alert a customer who is talking in a loud voice at a level that requires attention. As a result, by suppressing the loud conversation of the alerted customer, the opportunity to generate droplets can be reduced.

[0062] In the present embodiment, the hall 1 of a restaurant is described as an example of a predetermined space, but it is not necessary to be limited to this. For example, it is preferably applicable to facilities such as libraries where loud conversations are not preferred, or closed spaces where it is difficult for droplets to escape to the outside, and it can also be applied.

Explanation of Reference Numerals

[0063] 1 Hall, 2a~2d Table Seats, 3a~3d Tablet Terminals, 4 Access Point, 10 Store Management Device, 11 Sound Data Collection Unit, 12 Conversation Extraction Unit, 13 Volume Information Generation Unit, 14 Table Position Identification Unit, 15 Notification Unit, 16 Table Information Storage Unit, 17 Sound Data Information Storage Unit, 18 Conversation Information Storage Unit, 19 Volume Information Storage Unit, 31 Sound Collection Unit, 32 Display Unit, 33 Sound Output Unit.

Claims

1. A collecting means for collecting sound data from sound collecting means installed corresponding to each of a plurality of sections formed by dividing a predetermined space; A conversation extraction means for extracting a conversation being conducted in the space by analyzing the sound data collected by the collecting means; A section specifying means for specifying a section in which a conversation is being conducted at a volume exceeding a predetermined threshold from among the sections in which the sound collecting means serving as the collection source of the sound data having a volume exceeding the predetermined threshold is installed; A notification means for notifying the occurrence of noise to a notification means installed in the section specified by the section specifying means; having; The conversation extraction means extracts a conversation being conducted in the space by referring to at least one of the start and end timings of the speech of a person in the space or the words and phrases included in the speech. A warning device characterized by that.

2. The section specifying means determines that the conversation is being conducted in the section corresponding to the sound collecting means that has collected the sound data having the maximum volume among the conversations extracted by the conversation extraction means. The warning device according to claim 1, characterized by that.

3. The predetermined space is a store, The section is formed corresponding to each table seat arranged in the store, The sound collecting means and the notification means are terminal devices used when providing services to users of the section at the table seat. The warning device according to claim 1, characterized by that.

4. A terminal device installed corresponding to each of a plurality of sections formed by dividing a predetermined space and having a sound collecting means and a notification means, the terminal device being used when providing services to users of the section; A warning device communicably connected to each of the terminal devices; having, The warning device is Collecting means for collecting sound data collected by the sound collecting means from each of the terminal devices; Conversation extraction means for extracting a conversation being held in the space by analyzing the sound data collected by the collecting means; Section identification means for identifying a section where conversation is being held at a volume exceeding a predetermined threshold from among the sections where the terminal devices serving as the sources of sound data having a volume exceeding the predetermined threshold are installed; Notification means for notifying the occurrence of noise to the terminal devices installed in the section identified by the section identification means; which has, The conversation extraction means extracts the conversation being held in the space by referring to at least one of the start and end timings of the speech of the person in the space or the words and phrases included in the speech. A caution system characterized by this.

5. A computer, Collecting means for collecting sound data from sound collecting means installed corresponding to each of the sections formed by dividing a predetermined space into a plurality of sections, Conversation extraction means for extracting a conversation being held in the space by analyzing the sound data collected by the collecting means, Section identification means for identifying a section where conversation is being held at a volume exceeding a predetermined threshold from among the sections where the sound collecting means serving as the source of sound data having a volume exceeding the predetermined threshold is installed, Notification means for notifying the occurrence of noise to the notification means installed in the section identified by the section identification means, A program for causing it to function as, The conversation extraction means extracts the conversation being held in the space by referring to at least one of the start and end timings of the speech of the person in the space or the words and phrases included in the speech. A program characterized by this.

Images