Analysis device

Abstract

The challenge is to contribute to resolving various problems related to noise in apartment buildings. [Solution] The analysis device 10 includes an audio input means that receives audio signals from one or more sensors 12 installed or positioned in a dwelling unit 2 of an apartment building 1, which convert everyday sounds into electrical audio signals; an acoustic analysis means that analyzes the audio signals input by the audio input means; and a utilization means that utilizes the analysis results of the acoustic analysis means. The analysis device 10 also includes a video input means that receives video signals from a camera 14 installed or positioned in the dwelling unit 2, which films the inside of the dwelling unit 2; and a video analysis means that analyzes the video signals input by the video input means, and the utilization means utilizes the analysis results of the video analysis means in addition to the analysis results of the acoustic analysis means.

Description

【Technical Field】 【0001】 The technology disclosed in this specification relates to analysis devices Place . 【Background Art】 【0002】 In apartment houses, it is difficult to block the living sounds generated in the dwelling units, so the living sounds may be annoying to the residents of the upper floor, lower floor or adjacent dwelling units. Patent Document 1 discloses a noise management system devised to solve such problems. The noise management system includes a plurality of noise detection devices and a management server. The plurality of noise detection devices are respectively arranged in a plurality of dwelling units. The noise detection device detects the noise level of the dwelling unit where it is installed and generates noise data indicating the noise level. The noise detection device periodically transmits the noise data to the management server. The management server stores the noise data for each dwelling unit in a noise database in chronological order. The management server periodically compares the noise level indicated by the noise data stored in the noise database with a threshold value for each dwelling unit. As a result of the comparison, if the noise level is below the threshold value, the management server registers the resident of the dwelling unit corresponding to the noise level in the white list. As a result of the comparison, if the noise level exceeds the threshold value, the management server registers the resident of the dwelling corresponding to the noise level in the black list. The management server notifies the administrator of the registration results of the black list and the white list. The management server notifies the communication terminal of the resident registered in the white list that it will provide a useful service to encourage the resident to continuously keep the noise level below the threshold value in the future. The management server notifies the communication terminal of the resident registered in the black list of a warning to encourage the resident to continuously keep the noise level below the threshold value in the future. 【0003】 Residents of apartment buildings cannot objectively judge whether their actions and the resulting noises are bothering their neighbors. When residents are engrossed in something, they may unknowingly generate loud noises. Residents cannot accurately identify the source or cause of loud noises from others. Even if a resident can identify the source or cause of loud noises from others, that resident may not be able to take steps to resolve the noise problem. When a resident identifies the source of loud noises from others and files a complaint, it can lead to new problems. Loud and bothersome noises in apartment buildings are a source of trouble not only for residents but also for building managers. As described above, loud and bothersome noises can be the cause of various problems. [Prior art documents] [Patent Documents] 【0004】 [Patent Document 1] Japanese Patent Publication No. 2006-29859 [Overview of the project] [Problems that the invention aims to solve] 【0005】 The problem that the technology disclosed herein aims to solve is to contribute to resolving various problems related to noise in apartment buildings. [Means for solving the problem] 【0006】 To solve the above problems, an audio input means for receiving audio signals from one or more sensors installed or arranged in the dwelling units of an apartment building and which convert everyday sounds into electrical audio signals, and an acoustic analysis means for analyzing the audio signals input by the audio input means, A video input means that receives a video signal from a camera installed or positioned in the dwelling unit and that films the inside of the dwelling unit, and a video analysis means that analyzes the video signal received by the video input means, Analysis results of the acoustic analysis means and the analysis results of the video analysis means A means of using and The video analysis means includes a prediction means for predicting the type of future gesture that a person or figure shown in the video signal input by the video input means will subsequently take, and the utilization means includes a identification means for identifying the cause of everyday noises that are likely to be transmitted to the neighbors of the dwelling based on the analysis results of the acoustic analysis means and the analysis results of the video analysis means, and a second prediction means for predicting the possibility of everyday noises being transmitted to the neighbors of the dwelling based on the type predicted by the prediction means and the cause identified by the identification means. An analysis device will be provided. [Effects of the Invention] 【0007】 The analysis and management equipment will contribute to resolving various problems related to noise in apartment buildings. [Brief explanation of the drawing] 【0008】 [Figure 1] Figure 1 is a longitudinal cross-section of an apartment building. [Figure 2] Figure 2 is a diagram showing the configuration of the management system. [Figure 3] Figure 3 is a schematic diagram of the model represented by skeletal posture data. [Modes for carrying out the invention] 【0009】 [Summary of Disclosure] This specification discloses the following analysis and control devices. The following reference numerals in parentheses are referenced in Figures 1 to 3. The reference numerals in parentheses are included for the purpose of facilitating understanding of the analysis and control devices, and the scope of the present invention is not limited to the examples shown in the drawings. 【0010】 [1] The analysis device (10) comprises: an audio input means for inputting audio signals from one or more sensors (12) installed or positioned in a dwelling unit (2) of an apartment building (1) that convert everyday sounds into electrical audio signals; an acoustic analysis means for analyzing the audio signals input by the audio input means; and a utilization means for utilizing the analysis results of the acoustic analysis means. 【0011】 According to [1], since the sensor (12) is installed or positioned in the dwelling unit (2), the sounds of daily life in the dwelling unit (2) are converted into audio signals. The audio input means of the analysis device (10) receives these audio signals, and the acoustic analysis means of the analysis device (10) analyzes these audio signals, thereby allowing the analysis device (10) to acquire various information about the sounds of daily life. Since the user means of the analysis device (10) utilizes the analysis results from the acoustic analysis means, the analysis device (10) contributes to resolving various problems related to the sounds of daily life. 【0012】 [2] The analysis device (10) in [1] includes a video input means that receives a video signal from a camera (14) installed or positioned in the dwelling unit (2) and taking pictures inside the dwelling unit (2), and a video analysis means that analyzes the video signal received by the video input means, and the utilization means utilizes the analysis results of the video analysis means in addition to the analysis results of the acoustic analysis means. 【0013】 According to [2], since the camera (14) is installed or positioned in the dwelling unit (2), the inside of the dwelling unit (2) is photographed by the camera (14). The video input means of the analysis device (10) receives the video signal from the camera (14), and the video analysis means of the analysis device (10) analyzes the video signal, so the analysis device (10) acquires various information regarding secondary elements of everyday sounds. Since the user means of the analysis device (10) utilizes the analysis results from the video analysis means in addition to the analysis results from the acoustic analysis means, the analysis device (10) contributes to resolving various problems related to everyday sounds. 【0014】 [3] In the analysis device (10) of [1], the acoustic analysis means includes an extraction means for extracting a second audio signal for each type of ambient noise from the audio signal, a volume calculation means for calculating the volume at the respective location of the sensor (12) for each second audio signal extracted by the extraction means, and a sound source location calculation means for calculating the location of a sound source based on the same type of second audio signal extracted from the audio signal of the sensor (12), and the utilization means includes a determination means for determining whether the location of the sound source calculated by the sound source location calculation means is inside the dwelling unit (2). 【0015】 According to [3], by employing the extraction means, the volume at the location of the sensor (12) is calculated for each type of everyday noise by the volume calculation means, the location of the sound source is calculated for each type of everyday noise by the sound source location calculation means, and the determination means determines whether or not the location of the sound source for each type of everyday noise is inside the dwelling unit (2). The determination by the determination means contributes to resolving various problems related to everyday noise. 【0016】 〔4〕In the analysis device (10) of 〔3〕, when the utilization means determines that the determination means makes an affirmative determination, the utilization means has output means for outputting voice or display for presenting that the position of the sound source is within the dwelling unit (2). 【0017】 According to 〔4〕, the output means outputting voice or display gives notice to the residents of the dwelling unit (2). That is, the residents of the dwelling unit (2) notice the possibility of disturbing neighbors or the like by the occurrence of living sounds in their own dwelling unit (2). This contributes to the resolution of various problems related to living sounds. 【0018】 〔5〕In the analysis device (10) of 〔1〕, the acoustic analysis means has extraction means for extracting a second voice signal for each type of living sound from the voice signal, and for each of the second voice signals extracted by the extraction means, volume calculation means for calculating the volume at the position of each of the sensors (12) based on the second voice signal extracted by the extraction means. The utilization means has determination means for determining whether the volume calculated by the volume calculation means exceeds a threshold value for each of the second voice signals extracted by the extraction means. 【0019】 According to 〔5〕, by adopting the extraction means, the volume at the position of the sensor (12) is calculated for each type of living sound by the volume calculation means, and it is determined by the determination means whether the volume at the position of the sensor (12) exceeds the threshold value for each type of living sound. The determination by the determination means contributes to the resolution of various problems related to living sounds. 【0020】 〔6〕In the analysis device (10) of 〔5〕, when the utilization means determines that the determination means makes an affirmative determination, the utilization means has output means for outputting a warning by voice or display. 【0021】 According to [6], the output means outputting a warning alerts the residents of the household (2). That is, the residents of the household (2) become aware that the occurrence of loud living sounds in their own household (2) may cause trouble to neighbors or the like. This contributes to the resolution of various problems related to living sounds. 【0022】 〔7〕In the analysis device (10) of 〔1〕, the acoustic analysis means includes extraction means for extracting a second voice signal for each type of living sound from the voice signal, and volume calculation means for calculating the volume at the position of the sound source based on the same type of second voice signal for each of the same type of second voice signals extracted from the voice signal of the sensor (12). The utilization means has determination means for determining whether or not the volume calculated by the volume calculation means exceeds a threshold value for each of the second voice signals extracted by the extraction means. 【0023】 According to [7], by adopting the extraction means, the volume at the position of the sound source is calculated for each type of living sound by the volume calculation means, and it is determined by the determination means whether or not the volume at the position of the sound source exceeds the threshold value for each type of living sound. The determination by the determination means contributes to the resolution of various problems related to living sounds. 【0024】 〔8〕In the analysis device (10) of 〔7〕, the utilization means has output means for outputting a warning by voice or display when the determination means makes an affirmative determination. 【0025】 According to [8], the output means outputting a warning alerts the residents of the household (2). That is, the residents of the household (2) become aware that the occurrence of loud living sounds in their own household (2) may cause trouble to neighbors or the like. This contributes to the resolution of various problems related to living sounds. 【0026】 [9] In the analysis device (10) of [1], the acoustic analysis means includes an extraction means for extracting a second audio signal for each type of ambient sound from the audio signal, and a volume calculation means for calculating the volume at each of the sensor (12) locations based on the second audio signal extracted by the extraction means, and the utilization means includes an output means for outputting an audio or display for presenting the volume calculated by the volume calculation means for each of the second audio signals extracted by the extraction means. 【0027】 According to [9], by employing the extraction means, the volume at the location of the sensor (12) is calculated by the volume calculation means for each type of everyday noise, and an audio or display for indicating the volume is output by the output means. The output of audio or a display by the output means raises awareness among the residents of the dwelling unit (2). In other words, the residents of the dwelling unit (2) become aware of whether the volume of everyday noises generated in their dwelling unit (2) is causing a nuisance to their neighbors, etc. This contributes to resolving various problems related to everyday noise. 【0028】

[10] In the analysis device (10) of [1], the acoustic analysis means includes an extraction means for extracting a second audio signal for each type of ambient sound from the audio signal, and a volume calculation means for calculating the volume at the location of the sound source based on the same type of second audio signal extracted from the audio signal of the sensor (12), and the utilization means includes an output means for outputting an audio or display for presenting the volume calculated by the volume calculation means for each second audio signal extracted by the extraction means. 【0029】 According to

[10] , by employing the extraction means, the volume at the location of the sound source is calculated by the volume calculation means for each type of everyday noise, and an audio or display for indicating the volume is output by the output means. The output of the audio or display by the output means raises awareness among the residents of dwelling unit (2). In other words, the residents of dwelling unit (2) become aware of whether the volume of everyday noises generated in their dwelling unit (2) is causing a nuisance to their neighbors, etc. This contributes to resolving various problems related to everyday noise. 【0030】

[11] In the analysis device (10) of [1], the utilization means has a means for identifying the cause of everyday noises that are likely to be transmitted to the vicinity of the dwelling unit (2) based on the analysis results of the acoustic analysis means. 【0031】 According to

[11] , identifying the sources of everyday noises that may be transmitted to neighbors contributes to resolving various problems related to everyday noise. 【0032】

[12] In the analysis device (10) of [2], the utilization means has a determination means that identifies the cause of everyday noises that are likely to be transmitted to the vicinity of the dwelling unit (2) based on the analysis results of the acoustic analysis means and the analysis results of the video analysis means. 【0033】 According to

[12] , the identification of the source of everyday noises that may be transmitted to neighbors is based not only on the analysis results of the acoustic analysis means but also on the analysis results of the video analysis means. Therefore, the source of everyday noises can be identified more accurately. Identifying the source of everyday noises contributes to resolving various problems related to everyday noises. 【0034】

[13] In the analysis device (10) of

[12] , the video analysis means has a prediction means that predicts the type of future gesture that a person or individual shown in the video of the video signal input by the video input means will subsequently take, and the utilization means has a second prediction means that predicts the possibility of the occurrence of everyday noises that may be transmitted to the vicinity of the dwelling unit (2) based on the type predicted by the prediction means and the cause identified by the identification means. 【0035】 According to

[13] , the possibility of everyday noises that may be transmitted to neighbors can be predicted based on the prediction of the actions of people or individuals within the dwelling unit (2). This leads to the prevention of everyday noises that may be transmitted to neighbors. Problems related to everyday noises are resolved. 【0036】

[14] The analysis device (10) of

[13] includes a determination means for determining whether the utilization means is highly likely to be predicted by the second prediction means, and an output means for outputting a warning by voice or display when the determination means makes a positive determination. 【0037】 According to

[14] , if there is a high probability of generating everyday noises that may be heard by neighbors, the output device will issue a warning. The resident of dwelling unit (2) will receive the warning and refrain from making gestures that may cause everyday noises that may be heard by neighbors. This will help prevent everyday noises that may be heard by neighbors from occurring. The problem of everyday noises will be resolved. 【0038】

[15] In the analysis device (10) of [2],

[12] ,

[13] , or

[14] , the acoustic analysis means, the video analysis means, or the utilization means is realized by artificial intelligence. 【0039】 According to

[15] , if acoustic analysis means are implemented using artificial intelligence, the accuracy of analysis by acoustic analysis means will increase. If video analysis means are implemented using artificial intelligence, the accuracy of analysis by video analysis means will increase. If utilization means are implemented using artificial intelligence, the utilization efficiency of analysis by acoustic analysis means or video analysis means will improve. 【0040】

[16] The management device (70) communicates with any of the multiple analysis devices (10) from [1] to

[14] which are arranged or installed in each of the multiple dwelling units (2) of the apartment building (1). The management device (70) includes: a management means for managing collected data including the analysis results of the acoustic analysis means by collecting the analysis results of the acoustic analysis means from the multiple analysis devices (10); a receiving means for receiving complaint data transmitted by the analysis device (10) when a complainant inputs a complaint to any of the multiple analysis devices (10); an estimation means for estimating the source of nuisance noise from among the multiple dwelling units (2) based on the complaint data by referring to the collected data managed by the management means; and a transmitting means for transmitting a warning signal to the analysis device (10) arranged or installed in the dwelling unit (2) of the source estimated by the estimation means. 【0041】 According to

[16] , even if the complainant does not directly file a complaint with the resident of the dwelling unit (2) that is the source of the nuisance noise, the resident will receive a warning and will therefore endeavor to reduce the amount of noise they make. This also prevents disputes between the complainant and the resident. 【0042】

[17] In the management device (70) of

[16] , the estimation means is realized by artificial intelligence. 【0043】 According to

[17] , if the estimation means is implemented by artificial intelligence, the accuracy of the estimation by the estimation means will increase. 【0044】

[18] The management device (70) communicates with a plurality of analysis devices (10) described in any one of [1] to

[14] , which are arranged or installed in each of the plurality of dwelling units (2) of the apartment building (1). The management device (70) includes: a management means for managing collected data including the analysis results of the acoustic analysis means by collecting the analysis results of the acoustic analysis means from the plurality of analysis devices (10); a receiving means for receiving complaint data transmitted by the first analysis device (10) when a complainant inputs a complaint to one of the plurality of analysis devices (10); an estimation means for estimating the source of the nuisance noise from among the plurality of dwelling units (2) based on the complaint data by referring to the collected data managed by the management means; and a transmitting means for transmitting to the first analysis device (10) that the estimation means has estimated the source of the nuisance noise from among the plurality of dwelling units (2). 【0045】 According to

[18] , when a complainant inputs a complaint into the first analysis device (10), and the source of the nuisance noise is estimated, a message to that effect is transmitted from the management device (70) to the first analysis device (10). As a result, the complainant recognizes that the source of the nuisance noise has been identified and gains a sense of relief. 【0046】

[19] In the management device (70) of

[18] , the estimation means is implemented by artificial intelligence. 【0047】 According to

[19] , if the estimation means is implemented by artificial intelligence, the accuracy of the estimation by the estimation means will increase. 【0048】 The features, purpose, and advantages of the analysis device (10) and control device (70) described above will become clear from the drawings illustrating preferred embodiments and the explanation using the drawings. 【0049】 [Embodiment] Embodiments will be described below with reference to the drawings. The features and technical effects of the embodiments will be understood from the following detailed description and drawings. However, the scope of the present invention is not limited to the embodiments disclosed below. The scope of the present invention is not limited to the examples in the drawings, as the drawings are provided for illustrative purposes only. 【0050】 <1. Apartment Buildings and Management Systems> Figure 1 is a cross-sectional view of apartment building 1. Figure 2 is a block diagram of the management system to be adopted in this apartment building 1. 【0051】 Apartment building 1 is a so-called condominium or apartment. Apartment building 1 is made of wood, reinforced concrete, steel frame, steel-reinforced concrete, or a hybrid structure. A hybrid structure means a combination of two or more of the following: wood, reinforced concrete, steel frame, and steel-reinforced concrete. Apartment building 1 has multiple dwelling units 2. Dwelling unit 2 is separated from dwelling unit 2 on the floor below by the floor of dwelling unit 2 and the ceiling of dwelling unit 2 on the floor below. Dwelling unit 2 is separated from dwelling unit 2 on the floor above by the ceiling of dwelling unit 2 and the floor of dwelling unit 2 on the floor above. The floor of dwelling unit 2 may be a direct floor or a double floor. The ceiling of dwelling unit 2 may be a direct ceiling or a suspended ceiling. 【0052】 The management system comprises multiple analysis devices 10, multiple storage devices 30, a management device 70, and a storage device 80. The analysis devices 10 are installed or located in each dwelling unit 2. The storage devices 30 are provided for each dwelling unit 2. The management device 70 is installed in the data center. The storage devices 80 are connected to the management device 70. 【0053】 <2. Each dwelling unit> Each dwelling unit 2 is assigned an identification number, such as a room number. The identification number assigned to each dwelling unit 2 is a unique value among all the dwelling units 2. 【0054】 One or more sensors 12 are installed in each dwelling unit 2. The sensors 12 are acoustic sensors or vibration sensors, and for example, microphones or piezoelectric elements may be applied to the sensors 12. The sensors 12 are electroacoustic transducers that convert acoustic vibrations into analog electrical signals. The sensors 12 may also have A / D converters that convert continuous analog signals into discrete digital signals. The acoustic vibrations are everyday sounds that occur in the dwelling unit 2. Everyday sounds include, for example, footsteps, impact sounds, voices, sounds of electrical appliances, or animal (pet) noises. 【0055】 The sensor 12 may be installed on the floor, ceiling, partition wall, interior wall, or exterior wall of dwelling unit 2. If the sensor 12 is installed on the floor of dwelling unit 2, it may be placed on the surface of the floor, embedded in the floor, or installed in the underfloor space of a double floor. If the sensor 12 is installed on the ceiling of dwelling unit 2, it may be placed on the surface of the ceiling, embedded in the ceiling, or installed in the space above the suspended ceiling. If the sensor 12 is installed on the partition wall, interior wall, or exterior wall of dwelling unit 2, it may be placed on the surface of the partition wall, interior wall, or exterior wall, or embedded in the partition wall, interior wall, or exterior wall. 【0056】 If sensor 12 is installed on the floor of dwelling unit 2, it converts everyday sounds generated in dwelling unit 2 and transmitted to dwelling unit 2 on the floor below into electrical signals. If sensor 12 is installed on the ceiling, it converts everyday sounds generated in dwelling unit 2 and transmitted to dwelling unit 2 on the floor above into electrical signals. Sensor 12 installed on the floor of dwelling unit 2 may be shared as a sensor installed on the ceiling of dwelling unit 2 on the floor below. Sensor 12 installed on the ceiling of dwelling unit 2 may be shared as a sensor installed on the floor of dwelling unit 2 on the floor above. If sensor 12 is installed on the partition wall of dwelling unit 2, it converts everyday sounds generated in dwelling unit 2 and transmitted to the adjacent dwelling unit 2 into electrical signals. Sensor 12 installed on the partition wall may be shared by adjacent dwelling units 2 separated by that partition wall. 【0057】 Furthermore, there is a risk that sensor 12 may collect not only everyday noises generated in the dwelling unit 2 where sensor 12 is installed, but also sounds generated in other locations. 【0058】 Sensor 12 is connected to analysis device 10. Sensor 12 outputs a signal of ambient noise to analysis device 10, and analysis device 10 receives the signal of ambient noise. If sensor 12 does not have an A / D converter, analysis device 10 has an A / D converter, and this A / D converter converts the analog signal input from sensor 12 into a digital signal. Hereinafter, the digital signal input to analysis device 10 from sensor 12, or the digital signal obtained by A / D conversion of the analog signal input to analysis device 10 from sensor 12, will be referred to as the first audio signal. 【0059】 One or more surveillance cameras 14 are installed or positioned in each dwelling unit 2. The surveillance camera 14 has an image sensor, lens, and image processing circuit, etc. If there are multiple surveillance cameras 14, they are positioned in various locations and orientations within the dwelling unit 2. The surveillance cameras 14 may be fixed in place. The surveillance cameras 14 may be disguised as, for example, furniture, a clock, a power outlet, daily necessities, stationery, office supplies, miscellaneous goods, or works of art. 【0060】 One or more surveillance cameras 14 are connected to the analysis device 10 via an interface such as USB (registered trademark) or a local area network. The surveillance cameras 14 capture images of the interior of the dwelling unit 2 that they are pointed at, and generate video signals from the images. The surveillance cameras 14 immediately output the video signals to the analysis device 10, and the analysis device 10 inputs these video signals. The video signals may be encoded by the surveillance cameras 14. 【0061】 The analysis device 10 is installed or positioned in each dwelling unit 2. The analysis device 10 may be installed on the wall of dwelling unit 2. Each analysis device 10 is assigned an identification number. The identification number assigned to each analysis device 10 is a unique value among these analysis devices 10. Each analysis device 10 may be assigned the same identification number as the dwelling unit 2 in which it is installed. 【0062】 A storage device 30 is provided in each dwelling unit 2. The storage device 30 is connected to the analysis device 10. The storage device 30 may also be connected to the analysis device 10 by an interface. The storage device 30 may also be connected to the analysis device 10 via a LAN, the Internet 4, or both. The storage device 30 may be built into the analysis device 10. The storage device 30 may be externally attached to the analysis device 10. The storage device 30 may be installed at a distance from the analysis device 10. The storage device 30 is a semiconductor memory device, a magnetic memory device, a NAS (Network Attached Storage), a data server, a file server, or a cloud computing system. 【0063】 An input device 16 is installed or positioned in each dwelling unit 2. The input device 16 is integrated with the analysis device 10 or is attached externally to the analysis device 10. The input device 16 may be installed on the wall of the dwelling unit 2. The input device 16 may be portable together with the analysis device 10. 【0064】 The input device 16 is a keyboard, mouse, touch panel, touchpad, stylus, pointing device, key, or push button. When the input device 16 is operated by the user, the input device 16 outputs a signal to the analysis device 10 corresponding to the operation. The analysis device 10 receives signals from the input device 16, thereby acquiring commands or various inputs from the user. 【0065】 An output device 18 is installed or positioned in each dwelling unit 2. The output device 18 is integrated with the analysis device 10 or is attached externally to the analysis device 10. The output device 18 may be integrated with the input device 16, or the output device 18 may be installed on the wall of the dwelling unit 2. The output device 18 may be portable together with the analysis device 10. 【0066】 The output device 18 is a dot matrix display, a segment display, a single-color illuminator, a multi-color illuminator, a multi-color flashing device, a single-color flashing device, or an audio output device. The dot matrix display displays images according to the video signal input from the analysis device 10. The dot matrix display may be, for example, a liquid crystal display device or an organic EL display device. The segment display displays text such as numbers, English letters, kana characters, kanji characters, or symbols. The single-color illuminator, multi-color illuminator, multi-color flashing device, or single-color flashing device outputs signals using light. The audio output device outputs sound according to the audio signal input from the analysis device 10. The audio output device may be, for example, a speaker. 【0067】 The analysis device 10 is a device on which a general-purpose OS (Operating System) or an embedded OS is installed. The analysis device 10 may be, for example, a mobile phone, smartphone, tablet computer, laptop computer, or desktop computer. The general-purpose OS may be, for example, Windows®, Android®, iOS®, macOS®, Linux®, or Unix®. The embedded OS may be, for example, TRON OS. The analysis device 10 can communicate with the management device 70 via the Internet 4. The analysis device 10 may also access the management device 70 via the Internet 4 using a secure communication protocol such as a VPN. 【0068】 The analysis device 10 includes a main board, one or more hardware processors, a GPU (Graphics Processing Unit), RAM (Random Access Memory), a memory device, and a communication device. The main board includes a bus, a bus controller, and interface circuits, and transmits information between the hardware processor, GPU, RAM, memory device, input device 16, output device 18, and communication device. The hardware processor may be, for example, a CPU (Central Processing Unit). The hardware processor performs various arithmetic operations. The RAM provides the hardware processor with a storage area or work area when performing arithmetic operations. The GPU performs processing that can be done faster than the hardware processor (for example, image processing and matrix operations) under the command of the hardware processor. The communication device may be, for example, a network card or a WiFi® client device. 【0069】 The analysis device 10 stores the program 20 in its memory device. The analysis device 10 executes various processes according to the program 20. The processes executed by the analysis device 10 are described in detail below. 【0070】 <3. Execution process of the analysis device> (1) Acoustic analysis The analysis device 10 performs acoustic analysis processing to analyze the first audio signal. The floor plan of the dwelling unit 2, the area of ​​each room, and the volume of each room are incorporated into the program 20, and the analysis device 10 may refer to the floor plan of the dwelling unit 2, the area of ​​each room, or the volume of each room when performing acoustic analysis processing. The location, number, and specifications of the sensors 12 are incorporated into the program 20, and the analysis device 10 may refer to the location, number, and specifications of the sensors 12 when performing acoustic analysis processing. The acoustic analysis process is as follows, for example: 【0071】 (1-1) Extraction of audio by type The analysis device 10 extracts second audio signals for each type of ambient sound from the first audio signal in real time, for each sensor 12, by frequency analysis, phase analysis, or both of the above. The types of sounds include, for example, footsteps, impact sounds, speech, sounds from electrical appliances, sounds from sound equipment, animal sounds, and ambient sounds. Speech may be subdivided by individual. For example, speech may be subdivided into the speech of a specific person (e.g., a resident) and the speech of an unspecified person (e.g., a visitor, a television presenter, a radio presenter). Specific people may be subdivided by individual. Ambient sounds refer to all sounds collected by the sensor 12. In other words, the analysis device 10 extracts the first audio signal as the second audio signal of the ambient sound. 【0072】 By extracting a second audio signal for each type of ambient sound from the first audio signal, the amount of information related to sound increases, improving the accuracy of acoustic analysis. 【0073】 (1-2) Calculation of sound volume at each measurement point The analysis device 10 instantly calculates the volume [dB] at the location of each sensor 12 for each type of ambient noise. Specifically, for each second audio signal extracted from the audio signal of each sensor 12, the analysis device 10 instantly calculates the volume at the location of each sensor 12 based on the second audio signal. 【0074】 (1-3) Calculation of the sound source location The analysis device 10 instantly calculates the location of the sound source for each type of ambient noise other than the overall sound. Specifically, for each identical second audio signal extracted from the audio signals of multiple sensors 12, the analysis device 10 instantly calculates the location of the sound source based on the identical second audio signal. 【0075】 (1-4) Calculation of volume at the sound source location The analysis device 10 instantly calculates the volume [dB] at the location of the sound source for each type of ambient noise other than the overall sound. Specifically, for each identical second audio signal extracted from the audio signals of multiple sensors 12, the analysis device 10 instantly calculates the volume at the location of the sound source based on the identical second audio signal. 【0076】 (1-5) Use of artificial intelligence The analysis device 10 may perform the acoustic analysis processing described above using artificial intelligence 22 for acoustic analysis. The artificial intelligence 22 is built into the analysis device 10 in software by program 20. The artificial intelligence 22 is a model composed of an artificial neural network having a large number of parameters, such as weights or biases. The artificial neural network may be, for example, a feedforward neural network (FNN), a multilayer persetron FNN, a convolutional neural network (CNN), a 3D-CNN, a recurrent neural network (RNN), a long short-term memory (LSTM) RNN, a gated recurrent unit (GRU) RNN, an autoencoder, a transformer, a generative-opposed network (GAN), a graph neural network (GNN), or a spiking neural network (SNN). The artificial neural network may be a combination of several of these. The artificial neural network may also be a neural network derived from these. The artificial intelligence 22 may be a model that has been deeply machine-learned. Machine learning may be supervised learning, unsupervised learning, reinforcement learning, or a combination of two or more of these. 【0077】 If the learning of artificial intelligence 22 is supervised learning, various audio data and various acoustic analysis results of those audio data may be used as training data for learning artificial intelligence 22. The sound sources of the audio data may be various kinds of everyday sounds. That is, the sound sources may be various people's footsteps, footsteps of various motions, impact sounds of various impact objects, impact sounds of various objects being hit, various people's voices, sounds of various electrical appliances, sounds of various sound equipment, and sounds of various animals. The sound sources may be moving. The sound sources may be stationary. The acoustic analysis results may be, for example, the results of frequency analysis, phase analysis, sound field analysis, fluctuating sound analysis, or inverse acoustic analysis. 【0078】 The analysis device 10 may perform a fine-tuning process to retrain the artificial intelligence 22. For fine-tuning the artificial intelligence 22, the audio data 31, first volume transition data 32, position transition data 33, or second volume transition data 34 described later may be used. The analysis device 10 may also perform a real-time fine-tuning process to immediately retrain the artificial intelligence 22 using the calculation results of (1-1) to (1-4) described above. 【0079】 Artificial intelligence 22 may be either an inference model or a learning model. An inference model is a completed model that employs a neural network with fixed parameters once learning is complete. A learning model is a model in the process of learning that employs a neural network with adjustable parameters. Furthermore, artificial intelligence 22 may also be an inference model that has been retrained and completed. 【0080】 The artificial intelligence 22 may be constructed to specialize in acoustic analysis. If such artificial intelligence 24 is built into the analysis device 10, the analysis device 10 will operate with low power consumption. Such a low-power analysis device 10 will contribute to promoting carbon neutrality, realizing a decarbonized society, and achieving the Sustainable Development Goals (SDGs). 【0081】 (2) Recording of analysis results The analysis device 10 converts the second audio signal extracted from the first audio signal into audio data and records the audio data 31 in the storage device 30. The audio data 31 is classified by the type of ambient sound and the sensor 12, with audio data 31 for each type of ambient sound and for each sensor 12. However, to ensure the privacy of the residents, the audio data 31 does not have to be recorded. In particular, the audio data 31 of the voices of specific individuals does not have to be recorded. The audio data 31 of the voices of unspecified individuals does not have to be recorded. 【0082】 The analysis device 10 records the volume levels of each type of ambient noise at the location of each sensor 12 in the storage device 30 in a time-series format. The data generated in the storage device 30 by this recording is the first volume transition data 32. The first volume transition data 32 is classified by the type of ambient noise and the sensor 12, and there is a first volume transition data 32 for each type of ambient noise and each sensor 12. 【0083】 The analysis device 10 records the location of sound sources for each type of ambient noise other than the overall ambient noise in the storage device 30 in chronological order. The data generated in the storage device 30 by this recording is the location transition data 33. The location transition data 33 is classified by the type of ambient noise other than the overall ambient noise, and there is a separate set of location transition data 33 for each type of ambient noise other than the overall ambient noise. 【0084】 The analysis device 10 records the volume levels at the location of sound sources for each type of ambient noise other than the overall ambient noise in the storage device 30 in a time series. The data generated in the storage device 30 by this recording is the second volume transition data 34. The second volume transition data 34 is classified by the type of ambient noise other than the overall ambient noise, and there is a second volume transition data 34 for each type of ambient noise other than the overall ambient noise. 【0085】 (3) Video analysis The analysis device 10 performs video analysis processing to analyze the video signal transmitted from the surveillance camera 14 to the analysis device 10. The floor plan of the dwelling unit 2, the area of ​​each room, and the volume of each room may be incorporated into the program 20, and the analysis device 10 may refer to the floor plan of the dwelling unit 2, the area of ​​each room, or the volume of each room when performing video analysis processing. The location, number, and specifications of the surveillance cameras 14 may be incorporated into the program 20, and the analysis device 10 may refer to the location, number, and specifications of the surveillance cameras 14 when performing video analysis processing. The video analysis process is as follows: 【0086】 (3-1) Recognition of a person or identification of a person In the following explanation, "person" refers to a person who has been specifically identified as a particular individual. "A person" refers to a person who has not been specifically identified as a particular individual. 【0087】 The analysis device 10 recognizes the image of a person captured in the video signal. If multiple people are captured in the video, the analysis device 10 will recognize multiple images of people. 【0088】 The analysis device 10 may identify the elephants of people appearing in the video signal. Specifically, the analysis device 10 recognizes the elephants of people appearing in the video and determines whether the elephant of each recognized person is that of a predetermined person. The analysis device 10 may also calculate a likelihood that the elephant of a recognized person is likely to be that of a prototype person, and determine whether the elephant of a person is a person based on that likelihood. If multiple people appear in the video, the number of elephants of people identified by the analysis device 10 will be multiple. 【0089】 (3-2) Estimation of gesture type The analysis device 10 instantly estimates the type of gesture of each person or individual as it appears in the video signal. For example, as follows: 【0090】 The analysis device 10 extracts frames from the video at regular intervals. The regular interval is equal to an integer multiple of the reciprocal of the video's frame rate [fps]. If the regular interval is equal to 1x the frame rate, the analysis device 10 extracts a frame from the video every other frame. If the regular interval is equal to 2x the frame rate, the analysis device 10 extracts a frame from the video every other frame. If the regular interval is equal to 3x the frame rate, the analysis device 10 extracts a frame from the video every 2 frames. 【0091】 Each time a frame is extracted, the analysis device 10 estimates the skeletal posture of the person or figure depicted in the frame based on the extracted frame and generates skeletal posture data as shown in Figure 3. For example, each time a frame is extracted, the analysis device 10 estimates the skeletal posture of the person or figure based on the frame and generates skeletal posture data. The skeletal posture data may include position data of each joint 91 and rotation angle data of each joint 91. 【0092】 The analysis device 10 generates time-series data by arranging the skeletal posture data that is generated one after another in a time series. This time-series data is posture transition data that represents the changes in the posture of a person's or human's skeleton. By acquiring such posture transition data, the analysis device 10 estimates the changes in the posture of a person's or human's skeleton. 【0093】 The analysis device 10 estimates the type of gesture of a person or individual based on the posture transition data and generates gesture type data. If a person is identified in the process described in (3-1) above, the analysis device 10 associates the estimated type of gesture with the person who performed that type of gesture. 【0094】 (3-3) Predictions of future gesture types The analysis device 10 instantly predicts the type of future gesture that each person or individual will take in the video signal, and generates future gesture type data. For example, the analysis device 10 may instantly predict the type of future gesture based on the video signal. For example, the analysis device 10 may predict the type of future gesture based on the estimation results in "(3-2) Estimation of Gesture Type," that is, the gesture type data or its progression. 【0095】 (3-4) Use of artificial intelligence The analysis device 10 may perform the video analysis processing described above using artificial intelligence 24 for video analysis. The artificial intelligence 24 is built into the analysis device 10 in software by program 20. The artificial intelligence 24 is a model composed of an artificial neural network having a large number of parameters, such as weights or biases. The artificial neural network may be, for example, a feedforward neural network (FNN), a multilayer persetron FNN, a convolutional neural network (CNN), a 3D-CNN, a recurrent neural network (RNN), a long short-term memory (LSTM) RNN, a gated recurrent unit (GRU) RNN, an autoencoder, a transformer, a generative-opposed network (GAN), a graph neural network (GNN), or a spiking neural network (SNN). The artificial neural network may be a combination of several of these. The artificial neural network may also be a neural network derived from these. The trained model may be a model that has been deeply machine-learned. Machine learning may be supervised learning, unsupervised learning, reinforcement learning, or a combination of two or more of these. 【0096】 In the case of supervised learning, the data may include various human facial images, full-body images, partial body images, skeletal information, skeletal posture information, skeletal posture transition data, or behavioral videos, or the results of video analysis thereof. Behavioral videos may include videos of actions that generate everyday sounds. 【0097】 The analysis device 10 may perform a fine-tuning process to retrain the artificial intelligence 24. For fine-tuning the artificial intelligence 24, the video data 35, posture transition data 36, ​​gesture type transition data 37, or future gesture type transition data 38 described later may be used. The analysis device 10 may also perform a real-time fine-tuning process to immediately retrain the artificial intelligence 24 using the calculation results of (3-1) to (3-4) described above. The artificial intelligence 24 may be either an inference model or a learning model. 【0098】 The artificial intelligence 24 may be configured to specialize in video analysis. If such an artificial intelligence 24 is built into the analysis device 10, the analysis device 10 will operate with low power consumption. Such a low-power analysis device 10 will contribute to promoting carbon neutrality, realizing a decarbonized society, and achieving the Sustainable Development Goals (SDGs). 【0099】 (4) Recording of video footage and its analysis results The analysis device 10 converts the video signal transferred from the surveillance camera 14 to the analysis device 10 into video data 35 and records the video data 35 in the storage device 30. However, to ensure the privacy of the residents, the video data 35 does not necessarily have to be recorded. 【0100】 The analysis device 10 records the posture change data 36 for each person or individual in the storage device 30. 【0101】 The analysis device 10 records gesture type data in the storage device 30 in chronological order for each person or individual. The data generated in the storage device 30 by this recording is gesture type transition data 37. 【0102】 The analysis device 10 records future gesture type data for each person or individual in the storage device 30 in chronological order. The data generated in the storage device 30 by this recording is the future gesture type transition data 38. 【0103】 (5) Use of the results of acoustic analysis or video analysis The analysis device 10 performs various processes using the results of the acoustic analysis, video analysis, or both described above. All of the processes listed below may be performed, or one or more of the processes listed below may not be performed. The analysis device 10 may perform the following processes using the artificial intelligence 26 of the inference model. 【0104】 (5-1) Possibility of noises from daily life occurring within the dwelling unit The analysis device 10 determines whether the location of the sound source is inside the dwelling unit 2 for each type of ambient noise other than ambient noise. The location of the sound source for each type of ambient noise other than ambient noise is the result of the analysis described in (1-3) above. In making this determination, the analysis device 10 may refer to the floor plan of the dwelling unit 2, the area of ​​each room, or the volume of each room, which are incorporated into the program 20. 【0105】 (5-2) Excessive volume of ambient noise at the measurement point The analysis device 10 determines whether the volume at the location of each sensor 12 exceeds a threshold for each type of ambient noise. The volume at the location of each sensor 12 for each type of ambient noise is the result of the analysis described in (1-2) above. The threshold may differ for each sensor 12. The threshold may differ for each type of ambient noise. The threshold may differ for each sensor 12 and for each type of ambient noise. 【0106】 (5-3) Excessive volume of ambient noise at the location of the sound source The analysis device 10 determines whether the volume at the sound source location exceeds a threshold for each type of ambient noise other than the overall sound. The volume at the sound source location is the analysis result in (1-4) above. The threshold may differ for each sensor 12. The threshold may differ for each type of ambient noise. The threshold may differ for each sensor 12 and for each type of ambient noise. The threshold in (5-3) may differ from the threshold in (5-2). 【0107】 (5-4) Presentation of the volume of everyday noises at the measurement point The analysis device 10 immediately outputs audio or a display from the output device 18 to indicate the volume at the location of each sensor 12 for each type of ambient noise. The volume at the location of each sensor 12 is the result of the analysis described in (1-2) above. 【0108】 (5-5) Presentation of the volume of everyday sounds at the location of the sound source The analysis device 10 immediately outputs audio or a display via the output device 18 to indicate the volume at the location of the sound source for each type of ambient noise other than the overall sound. The volume at the location of the sound source is the result of the analysis described in (1-4) above. 【0109】 (5-6) Presentation of the occurrence of everyday noises within the dwelling unit. If the analysis device 10 determines that the location of the sound source is inside the dwelling unit 2 based on the determination process described in (5-1) above, the analysis device 10 outputs an audio or display message to indicate this using the output device 18. 【0110】 (5-7) Warning due to excessive noise levels at the measurement point If the analysis device 10 determines, through the determination process described in (5-2) above, that the volume threshold at the location of the sensor 12 has been exceeded, the analysis device 10 outputs a warning using the output device 18. The warning may be output as a warning sound, a warning voice, or a warning display. For example, if the volume of the impact sound at the location of the sensor 12 exceeds the threshold, the output device 18 may output a display or voice to that effect as a warning. 【0111】 (5-8) Warning due to excessive noise levels at the measurement point If the analysis device 10 determines, through the determination process described in (5-3) above, that the volume at the sound source location exceeds a threshold, the analysis device 10 outputs a warning using the output device 18. The warning may be output as a warning sound, a warning voice, or a warning display. For example, if the volume of the impact sound at the sound source location exceeds a threshold, the output device 18 may output a display or voice to that effect as a warning. 【0112】 (5-9) Warning issued due to exceeding the volume limit of everyday noises generated within the dwelling unit (based on the location of the measurement point). The analysis device 10 calculates a logical AND of the judgment result in the judgment process described in (5-1) and the judgment result in the judgment process described in (5-2) for each type of ambient noise other than the overall noise. If the logical AND is true, that is, if the sound source is located inside the dwelling unit 2 and the volume at the location of the sensor 12 exceeds the threshold, the analysis device 10 outputs a warning using the output device 18. The warning output may be a warning sound, a warning voice, or a warning display. 【0113】 (5-10) Warning issued due to exceeding the volume limit of everyday noises (based on the location of the sound source) generated within the dwelling unit. The analysis device 10 calculates a logical AND of the judgment result in the judgment process described in (5-1) above and the judgment result in the judgment process described in (5-3) above for each type of ambient noise other than the overall noise. If the logical AND is true, that is, if the sound source is located inside the dwelling unit 2 and the volume at the sound source location exceeds a threshold, the analysis device 10 outputs a warning using the output device. The warning output may be a warning sound, a warning voice, or a warning display. 【0114】 (5-11) Warning issued due to exceeding the volume limit of ambient noise (based on measurement point and sound source location). The analysis device 10 calculates a logical AND of the judgment result in the judgment process described in (5-2) and the judgment result in the judgment process described in (5-3) for each type of ambient noise other than the overall sound. If the logical AND is true, that is, if the volume at the location of the sensor 12 exceeds the threshold and the volume at the location of the sound source exceeds the threshold, the analysis device 10 outputs a warning using the output device. The warning output may be a warning sound, a warning voice, or a warning display. 【0115】 (5-12) Warning issued due to exceeding the volume limit of everyday noises generated within the dwelling unit (based on measurement point and sound source location). The analysis device 10 calculates a logical AND of the judgment result in the judgment process in (5-1), the judgment result in the judgment process in (5-2), and the judgment result in the judgment process in (5-3) for each type of ambient noise other than the overall noise. If the logical AND is true, that is, if the sound source is located inside the dwelling unit 2, and the volume at the location of the sensor 12 exceeds the threshold, and the volume at the location of the sound source exceeds the threshold, the analysis device 10 outputs a warning using the output device. The warning output may be a warning sound, a warning voice, or a warning display. 【0116】 (5-13) Identifying the cause of noise disturbing neighbors The analysis device 10 identifies the cause of everyday noises that are likely to be transmitted to the upper floor, lower floor, or adjacent dwelling unit 2, based on the results of the acoustic analysis. The analysis device 10 outputs an audio or display from the output device 18 to indicate the cause of the noise. The identification method is, for example, one of the following (A) to (D). 【0117】 (A) The analysis device 10 determines that the volume threshold at the location of the sensor 12 has been exceeded by the determination process described in (5-2) above. Then, the analysis device 10 identifies the type of sound that exceeded the threshold as the cause. 【0118】 (B) The analysis device 10 determines that the volume threshold at the location of the sensor 12 has been exceeded by the determination process described in (5-3) above. Then, the analysis device 10 identifies the type of sound that exceeded the threshold as the cause. 【0119】 (C) The analysis device 10 determines that the logical AND in (5-9) above is true. Then, the analysis device 10 identifies the type of sound whose volume exceeds the threshold as the cause. 【0120】 (D) The analysis device 10 determines that the logical AND in (5-10) above is true. Then, the analysis device 10 identifies the type of sound whose volume exceeds the threshold as the cause. 【0121】 (5-14) Identifying the cause of noise disturbing neighbors The analysis device 10 identifies the source of everyday noises that are likely to be transmitted to the upper floor, lower floor, or adjacent dwelling unit 2, based on the results of acoustic and video analysis. The analysis device 10 outputs audio or a display from the output device 18 to indicate the identified source of noise. The identification method is, for example, one of the following (A) to (D). 【0122】 (A) The analysis device 10 determines that the volume threshold at the location of the sensor 12 has been exceeded by the determination process described in (5-2) above. If so, the analysis device 10 identifies the type of gesture estimated by the estimation process described in (3-2) above as the cause. Furthermore, the analysis device 10 identifies a specific person associated with that type of gesture as the cause. 【0123】 (B) The analysis device 10 determines that the volume threshold at the location of the sensor 12 has been exceeded by the determination process described in (5-3) above. If so, the analysis device 10 identifies the type of gesture estimated by the estimation process described in (3-2) above as the cause. Furthermore, the analysis device 10 identifies a specific person associated with that type of gesture as the cause. 【0124】 (C) The analysis device 10 determines that the logical AND in (5-9) above is true. Then, the analysis device 10 identifies the type of gesture estimated by the estimation process in (3-2) above as the cause. Furthermore, the analysis device 10 identifies a specific person associated with that type of gesture as the cause. 【0125】 (D) The analysis device 10 determines that the logical AND in (5-10) above is true. Then, the analysis device 10 identifies the type of gesture estimated by the estimation process in (3-2) above as the cause. Furthermore, the analysis device 10 identifies a specific person associated with that type of gesture as the cause. 【0126】 (5-15) Prediction of the possibility of generating noise that may disturb neighbors. The analysis device 10 records the type of gesture identified as the cause of the noise as cause data 39 in the storage device 30 through the process described in (5-14) above. Subsequently, each time the analysis device 10 estimates the type of gesture in the future through the process described in (3-3) above, it predicts the likelihood of everyday noises being transmitted to the upper floor, lower floor, or adjacent dwelling unit 2 based on the future type of gesture and the cause data 39. For example, the analysis device 10 statistically analyzes the cause data 39 by referring to it and predicts the likelihood of everyday noises being transmitted to the upper floor, lower floor, or adjacent dwelling unit 2 due to the type of gesture in the future. The likelihood of such everyday noises being transmitted may be expressed numerically, for example, as a percentage. 【0127】 (5-16) Warning due to the high possibility of generating noise that may disturb neighbors. The analysis device 10 immediately compares the numerical value representing the probability predicted by the process described in (5-15) above with a threshold value to determine whether the numerical value exceeds the threshold. If the analysis device 10 makes a positive determination, it outputs a warning from the output device 18 based on that determination. The warning output may be a warning sound, a warning voice, or a warning display. As a result, the resident will understand that the action they took may cause trouble for their neighbor. Therefore, the resident will stop the action they took and take a different action. 【0128】 (5-17) Use of artificial intelligence The analysis device 10 may perform the above-described processing (5-1) to (5-16) by inputting the results of acoustic analysis, video analysis, or both into the artificial intelligence 26, and output the results of the processing from the artificial intelligence 26. The artificial intelligence 26 is built into the analysis device 10 in software by program 20. The artificial intelligence 26 is a model composed of an artificial neural network having a large number of parameters, such as weights or biases. The artificial neural network may be, for example, a feedforward neural network (FNN), a multilayer persetron FNN, a convolutional neural network (CNN), a 3D-CNN, a recurrent neural network (RNN), a long short-term memory (LSTM) RNN, a gated recurrent unit (GRU) RNN, an autoencoder, a transformer, a generative-opposed network (GAN), a graph neural network (GNN), or a spiking neural network (SNN). The artificial neural network may be a combination of some of these. Artificial neural networks may be neural networks derived from these. Trained models may be models that have undergone deep machine learning. Machine learning may be supervised learning, unsupervised learning, reinforcement learning, or a combination of two or more of these. 【0129】 The analysis device 10 may perform a fine-tuning process to retrain the artificial intelligence 26. For fine-tuning the artificial intelligence 26, the video data 35, posture transition data 36, ​​gesture type transition data 37, or future gesture type transition data 38 described later may be used. The analysis device 10 may also perform a real-time fine-tuning process to immediately retrain the artificial intelligence 26 using the calculation results of (5-1) to (5-15) described above. The artificial intelligence 26 may be either an inference model or a learning model. 【0130】 (6) Others The management device 70, described later, may perform the same processing as the analysis device 10 performs for each dwelling unit 2. To achieve this, the analysis device 10 instantly converts the first audio signal into an audio stream and instantly transfers the audio stream to the management device 70 via the internet 4. The management device 70 instantly converts the audio stream into an audio signal. Furthermore, the analysis device 10 converts the video signal from the surveillance camera 14 into a video stream and instantly transfers the video stream to the management device 70. The management device 70 instantly converts the video stream into a video signal. Using these audio and video signals, the management device 70 performs the same processing as the analysis device 10 performs according to program 72. Program 72 is stored in the storage device of the management device 70. The device on which the management device 70 records data is the storage device 80. 【0131】 <4. Management device> The management device 70 is installed in a data center. The management device 70 is connected to the Internet 4. The management device 70 can communicate with any of the analysis devices 10. The management device 70 is, for example, a GPU (Graphics Processing Unit) server. The management device 70 has a main board, one or more CPUs (Central Processing Units), one or more GPUs, RAM (Random Access Memory), a storage device such as an SSD (Solid State Drive) or HDD (hard disk drive), and a communication device. The main board has a bus, a bus controller, and interface circuits. The main board transmits information between the CPU, GPU, RAM, storage device, and communication device. 【0132】 The management device 70 is connected to the storage device 80. The storage device 80 is a semiconductor memory device, magnetic memory device, NAS (Network Attached Storage), data server, file server, or cloud computing system. The storage device 80 is installed in the same or a different data center as the management device 70. The storage device 80 may be connected to the management device 70 by an interface. The storage device 80 may be connected to the management device 70 via a LAN, the Internet 4, or both. The storage device 80 may be built into the management device 70. The storage device 80 may be externally attached to the management device 70. The storage device 80 may be installed at a distance from the analysis device 10. 【0133】 The management device 70 stores program 72 in its memory. By executing processes according to program 72, the management device 70 performs various functions. The functions and processes of the management device 70 described below are realized by program 72. 【0134】 <4-1. Collection and Management Functions> The management device 70 collects first volume transition data 32, position transition data 33, second volume transition data 34, posture transition data 36, ​​gesture type transition data 37, future gesture type transition data 38, and cause data 39 from multiple analysis devices 10, which are stored in storage devices 80 connected to each analysis device 10. Specifically, the management device 70 and each analysis device 10 synchronize the first volume transition data 32, position transition data 33, second volume transition data 34, posture transition data 36, ​​gesture type transition data 37, future gesture type transition data 38, and cause data 39 stored in the storage device 80. Through synchronization with each analysis device 10, the management device 70 manages the collected data 82 for each analysis device 10 in the storage device 80. The collected data 82 is a copy of the first volume transition data 32, position transition data 33, second volume transition data 34, posture transition data 36, ​​gesture type transition data 37, future gesture type transition data 38, and cause data 39. The collected data 82 exists for each analysis device 10, that is, for each dwelling unit 2. The management device 70 associates the same identification number of the analysis device 10 with the collected data 82, thereby managing the collected data 82 for each analysis device 10, that is, for each dwelling unit 2. 【0135】 <4-2. Artificial Intelligence> Artificial intelligence 74 is built into the management device 70 by software using program 72. Artificial intelligence 74 is a model composed of an artificial neural network having a large number of parameters, such as weights or biases. The artificial neural network may be, for example, a feedforward neural network (FNN), a multilayer persetron FNN, a convolutional neural network (CNN), a 3D-CNN, a recurrent neural network (RNN), a long short-term memory (LSTM) RNN, a gated recurrent unit (GRU) RNN, an autoencoder, a transformer, a generative-opposed network (GAN), a graph neural network (GNN), or a spiking neural network (SNN). The artificial neural network may be a combination of several of these. The artificial neural network may also be a neural network derived from these. The trained model may be a model that has been deeply machine-learned. Machine learning may be supervised learning, unsupervised learning, or reinforcement learning, or a combination of two or more of these. Artificial intelligence 74 is a learning model or an inference model. 【0136】 <4-3. Receiving Complaints> The management device 70 is waiting to receive complaints. Meanwhile, if a resident of any of the dwelling units 2 (hereinafter referred to as the complainant) perceives annoying noises coming from another dwelling unit 2 in their own dwelling unit 2, the complainant inputs a complaint about the noise to the analysis device 10. For example, the complainant inputs the time the annoying noise occurred to the analysis device 10. For example, the complainant inputs the direction from which the annoying noise originated, i.e., the floor above, the floor below, or the adjacent floor. For example, the complainant inputs the type of annoying noise to the analysis device 10. The input of the annoying noise may also be done using a chatbot, which is realized through communication between the analysis device 10 and the management device 70, and through cooperation between the analysis device 10 and the management device 70. 【0137】 The analysis device 10 responds to the input of a complaint by the complainant and acquires information about the complaint input by the complainant (hereinafter referred to as "complaint data"). The complaint data may include information about the time the nuisance noise occurred. The complaint data may include information about the direction from which the nuisance noise occurred. The complaint data may include information about the type of nuisance noise. The complaint data may be in text format. 【0138】 The analysis device 10 transmits the complaint data to the management device 70, and the management device 70 receives the complaint data. As a result, the management device 70 accepts the complaint. 【0139】 <4-4. Estimating the source of annoying everyday noises> The management device 70, while referring to the collected data 82, estimates the source of the nuisance noise from among the dwelling units 2 of the apartment building 1 based on the complaint data. There may be one or more sources of the nuisance noise. If there are two or more estimated sources, the management device 70 may rank them. The ranking may be in order of the likelihood of generating the nuisance noise. The ranking may be in order of the expected volume of the nuisance noise. There may also be cases where the management device 70 cannot estimate the source of the nuisance noise. 【0140】 <4-5. Cases where estimation is impossible> If the management device 70 is unable to estimate the source of the nuisance noise, it transmits this information to the analysis device 10, which then receives the information. The analysis device 10 then outputs an audio or display message via the output device 18 indicating that it was unable to estimate the source of the nuisance noise. Such an output helps the complainant realize their mistake. 【0141】 <4-6. Cases where estimation is possible> When the management device 70 estimates the source of a nuisance noise, it transmits this information to the analysis device 10, which receives the information. The analysis device 10 then outputs an audio or display from the output device 18 indicating that the source of the nuisance noise has been estimated. This indication may be in a format that does not identify the dwelling unit of the source. Furthermore, the management device 70 transmits a message to the analysis device 10 advising attention to the source, which receives the information. The analysis device 10 then outputs an audio or display from the output device 18 advising attention to the source. This indication may be in a format that does not identify the dwelling unit of the source. The output from the analysis device 10 gives the complainant a sense of reassurance that the occurrence of nuisance noises will be suppressed. 【0142】 <4-7. Cases where estimation is possible> When the management device 70 estimates the source of a nuisance noise, it sends a warning signal to the analysis device 10 of the dwelling unit 2 that is estimated to be the source of the nuisance noise. When the analysis device 10 receives the warning signal, it outputs a warning via the output device 18. The warning output may be a warning sound, a warning voice, or a warning display. The content of the warning may be, for example, that the resident is causing a nuisance to their neighbors with excessive noise. Such a warning will make the resident of the dwelling unit 2, which is estimated to be the source of the nuisance noise, aware that they have been generating relatively loud noises. As a result, the resident will change their behavior or lifestyle and reduce the amount of noise they generate that is a nuisance to others. 【0143】 <4-8. Recording of Complaint Data and Estimation Results> Regardless of whether the source of the nuisance noise can be estimated, the management device 70 associates the complaint data with the estimation result and records the complaint data and estimation result in the storage device 80. The data generated in the storage device 80 by such recording is log data 84. Log data 84 consists of complaint data and the estimation result associated with it. Note that there is log data 84 for each piece of complaint data, and log data 84 is recorded in the storage device 80 each time the management device 70 estimates the source of the nuisance noise based on the complaint data. The estimation result includes at least information on whether or not the source of the nuisance noise could be estimated, and, if the source of the nuisance noise was estimated, the identification number of the dwelling unit 2 as the source. 【0144】 <5. Summary> The embodiments disclosed herein are for illustrative purposes only and are not intended to limit the scope of the invention. The scope of the invention should be interpreted by the terms of the claims. [Explanation of symbols] 【0145】 10 Analysis device 12 sensors 14 surveillance cameras 70 Management device

Claims

[Claim 1] An audio input means that receives the audio signal from one or more sensors installed or positioned in the dwelling unit of an apartment building, which convert everyday sounds into electrical audio signals, An acoustic analysis means for analyzing the audio signal input by the audio input means, A video input means that receives video signals from a camera installed or positioned in the dwelling unit and used to film the inside of the dwelling unit, A video analysis means for analyzing the video signal input by the video input means, A means for utilizing the analysis results of the acoustic analysis means and the analysis results of the video analysis means, Equipped with, The aforementioned video analysis means Predictive means for predicting the type of future gesture taken by a person or individual shown in the video signal input by the video input means. It has, The aforementioned means of use is, Based on the analysis results of the acoustic analysis means and the analysis results of the video analysis means, a means for identifying the cause of everyday noises that are likely to be transmitted to the neighbors of the dwelling unit, A second prediction means predicts the likelihood of everyday noises that may be transmitted to the neighbors of the dwelling unit, based on the type predicted by the prediction means and the cause identified by the identification means. has Analysis device. [Claim 2] The aforementioned means of use is, A determination means for determining whether or not the prediction made by the second prediction means is highly likely, If the determination means makes a positive determination, the output means outputs a warning by voice or display, has The analysis apparatus according to claim 1. [Claim 3] The aforementioned acoustic analysis means, the aforementioned video analysis means, or the aforementioned utilization means are realized by artificial intelligence. The analysis apparatus according to claim 1 or 2.

Images