Acoustic monitoring system

Abstract

To obtain a sound monitoring system having a function of executing a state estimation for a monitoring target area on the basis of a sound signal while protecting privacy.SOLUTION: A sound monitoring system comprises: a microphone that collects sound information generated in a monitoring target area; a storage unit that stores sound signals; and a control unit that extracts information other than conversation content contained in the sound information as non-voice information, generates data in which the non-voice information is associated with a time of sound collection as a sound signal and stores the data in the storage unit, executes a state estimation for the monitoring target area on the basis of the sound signal, and executes report processing for indicating a state estimation result.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to an acoustic monitoring system that estimates the state of a monitored area based on acoustic information generated in the monitored area.

Background Art

[0002] There is a fire alarm facility that determines whether there is a person who has been left behind by determining whether a signal indicating human voice is included in the sound signal collected by the sound collection means when a fire is detected, and displays the determination result (see, for example, Patent Document 1).

[0003] According to the fire alarm facility according to Patent Document 1, the state information of each fire detector and the sound signal of each sound collection means can be collected on the same signal line. Further, it is possible to display on the fire receiver that there is a person who has been left behind during a fire.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Patent Document 1 determines whether there is a person who has been left behind by determining whether a signal indicating human voice is included in the sound signal. However, depending on the installation environment, from the viewpoint of privacy protection, there may be cases where the installation of equipment that can collect signals capable of specifying conversation contents is restricted.

[0006] On the other hand, the sound signal is expected to be effectively utilized not only for determining being left behind but also, for example, when a person in each room of a facility for the elderly, a hospital, etc. falls, or when an abnormal situation occurs.

[0007] This disclosure is made to solve the above-mentioned problems and aims to provide an acoustic monitoring system that has the function of estimating the state of a monitored area based on sound signals while protecting privacy. [Means for solving the problem]

[0008] The acoustic monitoring system described herein comprises a microphone for collecting acoustic information generated in a monitored area, a storage unit for storing sound signals, and a control unit that extracts information other than conversational content contained in the acoustic information as non-audio information, generates data as sound signals associating the non-audio information with the time of sound collection and stores it in the storage unit, performs state estimation of the monitored area based on the sound signals, and performs an alarm process to notify the state estimation result. When the control unit receives a fire signal indicating that a fire has occurred, it performs state estimation by estimating the presence of people in the monitored area based on sound signals stored in the memory unit during a preset time period prior to the time the fire signal was received. It is. [Effects of the Invention]

[0009] According to this disclosure, it is possible to obtain an acoustic monitoring system that has the function of performing state estimation of a monitored area based on sound signals while protecting privacy. [Brief explanation of the drawing]

[0010] [Figure 1] This is an explanatory diagram showing the overall configuration of the acoustic monitoring system according to Embodiment 1 of this disclosure. [Figure 2] This is an explanatory diagram of an acoustic monitoring system corresponding to Example 1 in Embodiment 1 of the present disclosure. [Figure 3] This is an explanatory diagram of an acoustic monitoring system corresponding to Example 2 in Embodiment 1 of the present disclosure. [Modes for carrying out the invention]

[0011] Hereinafter, preferred embodiments of the acoustic monitoring system of this disclosure will be described with reference to the drawings. The acoustic monitoring system described herein is technically characterized by its ability to estimate the state of a monitored area based on non-audio information and to notify the user of the estimation results. As a result, while protecting privacy, it can achieve a remarkable effect of estimating the state of a monitored area based on sound signals in various applications.

[0012] Embodiment 1. The acoustic monitoring system described herein, taking privacy protection into consideration, does not record the content of conversations, and is applicable to various uses in a monitored area, such as estimating occupancy status, estimating daily routines, and estimating the presence or absence of intruders, based on sound signals. This will be explained in detail with reference to Figures 1 to 3.

[0013] Figure 1 is an explanatory diagram showing the overall configuration of an acoustic monitoring system according to Embodiment 1 of the present disclosure. The acoustic monitoring system according to Embodiment 1 comprises a microphone 10, a control unit 20, and a storage unit 30.

[0014] Microphone 10 collects acoustic information generated in the monitored area 1. Specific examples of monitored area 1 include areas within elderly care facilities, hospitals, schools, and buildings. Microphone 10 is installed in a location capable of collecting acoustics from within the monitored room, such as inside or outside each room or living space.

[0015] The control unit 20 collects acoustic information generated in the monitored area 1 via the microphone 10. By performing analysis processing on the collected acoustic information using known sound source separation techniques such as sound frequency analysis, it is possible to separate the collected sound into human voices and ambient sounds. This allows for the classification and extraction of sounds containing human voices and conversations as speech information, and ambient sounds such as noises and everyday sounds as non-speech information.

[0016] In other words, the control unit 20 according to this embodiment 1 does not perform speech recognition processing to identify the content of a conversation, taking privacy protection into consideration, and instead extracts noises and everyday sounds other than conversation content that arise from human actions as non-speech information.

[0017] The control unit 20 generates data associating the extracted non-audio information with the recording time of the audio information including the non-audio information as an audio signal, and stores it in the storage unit 30. The recording time mentioned here indicates the generation time when ambient sounds such as sounds other than conversation content occur, which is the time when such ambient sounds are recorded. It is possible to store not only the audio signal but also the audio information including the voice information in the storage unit 30.

[0018] Furthermore, the control unit 20 executes state estimation of the monitoring target area 1 based on the audio signal, and executes a reporting process to notify the state estimation result. Also, by accumulating the audio signal in the storage unit 30, when the same audio signal is accumulated a predetermined number of times, it may be stored as a state estimation pattern of the monitoring target area.

[0019] Specifically, based on the audio signal stored in the storage unit 30, the control unit 20 can also generate a state estimation pattern from a combination of a fixed time and a fixed ambient sound, such as there being sounds of the bed or sounds of eating utensils or eating, or sounds of water in the shower or toilet, etc., at a fixed time zone, and store it in the storage unit 30.

[0020] Next, a specific example of the state estimation executed by the control unit 20 in various monitoring target areas will be described in detail according to the embodiments.

[0021] <Example 1: A case where state estimation is executed by estimating the living rhythm of a resident> In Example 1, taking a senior facility as an example, a case where state estimation is performed by estimating the living rhythm of the residents in each living room based on audio information will be described.

[0022] In senior facilities, the elderly residents may fall in their living rooms, and it is important to prevent falls or detect falls early. To prevent falls or detect falls early, regular rounds by caregivers and night inspections are carried out.

[0023] For example, falls can occur at times directly related to each resident's daily routine, such as when they wake up, before going to bed, or when they go to the toilet alone.

[0024] Therefore, in Embodiment 1, the control unit 20 estimates the state of the monitored area 1 based on the acoustic information collected for each room, which is the monitored area 1, and the sound signal generated from the time of sound collection. This estimate includes, for example, the timing of each resident waking up, going to bed, and going to the toilet, as part of their daily rhythm, and notifies the resident of the timing when falls are likely to occur as a result of the state estimation.

[0025] Furthermore, if the control unit 20 generates a state estimation pattern, it may issue a warning notification if it receives acoustic information that does not match the state estimation pattern. For example, if the control unit 20 has been able to estimate the daily behavior pattern of a resident who goes to the toilet at 11 pm from the acoustic information, and it detects an acoustic signal different from that acoustic information that is not stored in the memory unit 30, it can determine that an accident such as a fall has occurred to the resident and issue a notification.

[0026] As a result, managers, such as those in charge of elderly care facilities, can provide appropriate support and monitoring by conducting patrols in accordance with each resident's daily rhythm and at times when falls are most likely to occur. Furthermore, if an unusual pattern of condition is observed, managers can pay attention, which can prevent falls or, if a fall does occur, allow for early detection.

[0027] Figure 2 is an explanatory diagram of an acoustic monitoring system corresponding to Embodiment 1 of Embodiment 1 of the present disclosure. In Figure 2, microphones 10 are installed on the outside of the doors of eight rooms A to H in an elderly care facility. Here, the eight rooms A to H correspond to monitoring areas 1(1) to 1(8), respectively.

[0028] By using microphone 10 instead of surveillance cameras, the feeling of being monitored can be reduced for each resident. Furthermore, since microphone 10 is installed outside the living space, it does not pick up people's voices or conversations, but mainly captures everyday sounds caused by people's actions, thus protecting the privacy of each resident.

[0029] The control unit 20 extracts information other than conversation content from the acoustic information of each of the rooms A to G, which is picked up via the microphone 10, as non-audio information, generates data as an audio signal as it associates the non-audio information with the time of pickup, and stores it in the storage unit 30.

[0030] Next, the control unit 20 collects lifestyle sound information originating from the residents of each room in the monitored area, based on the sound signals for each room stored in the memory unit 30. Specifically, the control unit 20 classifies the sounds based on the sound signals for each room stored in the memory unit 30 into categories such as the sounds of opening and closing windows and curtains, the sounds of people sleeping, the sounds of people bathing, and the sounds of flushing toilets, and collects the lifestyle sound information in association with the time of occurrence.

[0031] Next, the control unit 20 performs state estimation by estimating the daily rhythm of each resident based on the collected daily sound information. Specifically, the control unit 20 estimates the timing of activities such as waking up, going to bed, napping, bathing, and using the toilet as a daily rhythm for each resident from the collected daily sound information.

[0032] Furthermore, the control unit 20 can improve the accuracy of estimating each resident's daily rhythm by estimating the daily rhythm based on acoustic information collected over a wider range of days.

[0033] The control unit 20 then performs an alarm process to inform the staff of the elderly care facility of the estimated results of each resident's daily rhythm. As a result, the staff of the elderly care facility can support or patrol residents at the appropriate time based on the estimated results of each resident's daily rhythm, thereby preventing falls and detecting falls early, and properly monitoring the elderly.

[0034] In the above-described Example 1, we illustrated a case where the person in charge provides support and conducts patrols to prevent falls based on the estimated results of the daily rhythm. However, the use of the estimated results of the daily rhythm is not limited to this example.

[0035] For example, the person in charge can use the estimated results of the residents' daily routines to understand when they should take their medication, when they should eat, etc., and this information can be effectively used to provide support tailored to each resident and to prevent aspiration.

[0036] Furthermore, while the above-described Example 1 focused on elderly care facilities, it is also possible to apply this to other facilities such as hospitals.

[0037] <Example 2: A case where state estimation is performed by estimating the presence or absence of an intruder> Example 2 describes a case where a school is used as an example to explain how to estimate the state of a school building by estimating the presence or absence of an intruder based on acoustic information.

[0038] In facilities such as schools and buildings, staff may conduct patrols at night or other times to check for any unusual incidents, such as the intrusion of suspicious individuals.

[0039] Therefore, in Embodiment 2, the control unit 20 estimates the occurrence of sounds such as footsteps of an intruder, sounds of doors opening and closing, and sounds of glass breaking, based on acoustic information collected within the school building, and when suspicious sounds are estimated, it notifies that there is an intruder as a result of the state estimation.

[0040] As a result, patrol security can be made more efficient and manpower-efficient, allowing for monitoring of facilities such as schools and buildings at night and enabling a quick response in the event of an emergency.

[0041] Figure 3 is an explanatory diagram of an acoustic monitoring system corresponding to Example 2 in Embodiment 1 of this disclosure. In Figure 3, microphones 11 to 14 are installed at appropriate locations in five classrooms 1 to 5, the teacher's room, the music room, and the corridor in an elementary school. Here, the five classrooms 1 to 5 correspond to monitoring areas 1(1) to 1(5), the teacher's room corresponds to monitoring area 1(6), the music room corresponds to monitoring area 1(7), and the corridor corresponds to monitoring area 1(8).

[0042] Figure 3 illustrates a case where five microphones 11 are placed in monitoring areas 1(1) to 1(5), and in the teachers' room, microphone 12 is placed in monitoring area 1(6), microphone 13 is installed in monitoring area 1(7), and four microphones 14 are installed in monitoring area 1(8).

[0043] Note that the control unit 20 and storage unit 30 are omitted from Figure 3.

[0044] The control unit 20 performs state estimation by estimating the presence or absence of an intruder based on the magnitude of the sound pressure level of the sound signal stored in the memory unit 30 during a pre-set intruder alert period.

[0045] In other words, if the control unit 20 detects a sound signal with a sound pressure level above a preset level during the intruder alert period, it will analyze the sound signal with a sound pressure level above the set value and extract whether it is audio information such as people shouting, or non-audio information such as footsteps of an intruder, the sound of doors opening and closing, or the sound of glass breaking. The control unit can then indicate which type of information the collected sound signal represents and issue a notification to the administrator to alert them.

[0046] This allows us to estimate that a sound that should not normally occur is being generated, and to determine that there may be an abnormal situation.

[0047] The control unit 20 then performs an alarm process to notify that there is a possibility of an abnormal situation occurring. As a result, security personnel at facilities such as schools and buildings can quickly grasp the possibility of an abnormal situation occurring, take immediate and appropriate action, and monitor the facility so that they can respond quickly even if an abnormal situation occurs at night.

[0048] <Example 3: A case in which state estimation is performed by estimating the position and direction of movement of the sound source> In the previous embodiment 2, we described a case in which the presence or absence of an intruder is estimated from the magnitude of the sound pressure level of the sound signal stored in the memory unit 30. In contrast, embodiment 3 describes a case in which the location of the intruder corresponding to the sound source and the direction of the intruder's movement are estimated using acoustic information picked up by multiple microphones.

[0049] In Figure 3 above, microphones 11 to 14 are installed at multiple different locations. Therefore, the control unit 20 can perform state estimation by estimating the position and direction of movement of the sound source from the transition state of the sound pressure levels of the sound signals for each of the multiple microphones 11 to 14 stored in the memory unit 30.

[0050] For example, the control unit 20 estimates that the intruder is located at the position of the microphone that collected the sound signal with the highest sound pressure level among the sound signals from multiple microphones 11 to 14. Furthermore, the control unit 20 can estimate the intruder's direction of movement from the state in which the position of the microphone that collected the sound signal with the highest sound pressure level changes over time.

[0051] The control unit 20 then performs an alarm process to notify the estimated location and direction of movement of the suspicious person. As a result, security personnel at facilities such as schools and buildings can quickly grasp, based on the location and direction of movement, that there may be an abnormal situation caused by an intruder, and can take appropriate action immediately.

[0052] In particular, according to Example 3, security personnel can know the location and direction of movement of an intruder as a state estimation result, making it easier to take appropriate measures to properly protect the facility from intruders.

[0053] In the above-described Example 3, the case where it is applied to Figure 3 was explained, but it can also be applied to other monitored areas, such as Figure 2. For example, when applied to the elderly care facility shown in Figure 2, in addition to estimating the location and direction of movement of an intruder, it can also quickly estimate the location and direction of movement of elderly people who wander around at night and trigger an alarm.

[0054] In other words, it becomes possible to quickly estimate the location and direction of movement of sound sources according to the characteristics of the area being monitored, and to take appropriate action based on the estimation results.

[0055] <Example 4: A case where state estimation is performed by estimating the presence of a person> Patent Document 1, described as prior art, discloses a technique for determining whether or not there are people trapped in the escape route by determining whether or not the sound signals obtained from each sound-gathering means contain signals indicating human voices when a fire signal is generated to indicate that a fire has occurred.

[0056] However, after a fire signal is generated, sounds such as emergency broadcasts, alarm bells, and the operation of various fire prevention equipment may occur, making it difficult to accurately extract signals indicating the voices of people who were unable to escape. Therefore, in the acoustic monitoring system according to Example 4, a technique for estimating whether or not a person was present in the monitored area from sound signals collected immediately before a fire signal is generated will be described.

[0057] Although the acoustic monitoring system in Example 4 is described as being connected to a fire alarm system that monitors for fires and emits signals to alert of fires, it can also be implemented without connecting to a fire alarm system. For example, one possible method is to extract an alarm sound indicating a fire from the collected acoustic information, analyze and extract the sound signal immediately before the alarm sound is generated, and estimate that a person was present in the monitored area when voice information is extracted.

[0058] The control unit 20 is further equipped with a receiving function that can receive fire signals from the fire system, and stores sound signals in the storage unit 30 even before a fire signal is generated from the fire system. For example, the control unit 20 can sequentially update the sound data to be stored in the storage unit 30 at a predetermined time interval.

[0059] A specific example of a pre-set time interval is 10 minutes. In this case, the control unit 20 can refer to the sound signals stored in the memory unit 30 during the 10 minutes immediately preceding the time the fire signal was received.

[0060] The control unit 20 then performs state estimation by estimating that a person was present in the monitored area during the 10 minutes immediately preceding the fire signal reception, if the sound signals from the most recent 10 minutes include sounds caused by human activity.

[0061] Sounds generated by human actions include everyday sounds described in Example 1, and sound sources described in Examples 2 and 3.

[0062] In this way, the control unit 20 can estimate the presence of people within the monitored area during a preset time window prior to the reception of the fire signal. By performing the estimation process before the reception of the fire signal, it is possible to suppress the adverse effects on estimation accuracy caused by sounds such as emergency broadcasts, alarm bells, and the operation of various fire prevention equipment that occur after the fire signal is generated.

[0063] Furthermore, the control unit 20 does not extract human voices, but rather extracts sounds generated as a result of human actions to estimate the presence of a person. Therefore, it can estimate the presence of a person even in situations where no voice is being produced, or where voices cannot be accurately recognized.

[0064] Furthermore, if the control unit 20 can estimate the presence of a person within the monitored area, it can perform an alarm process to provide approximate location information of the person, based on the location of the corresponding microphone.

[0065] Therefore, when a fire signal is received, it is possible to quickly estimate the presence or absence of people within the monitored area, which can assist in appropriate rescue and evacuation guidance activities.

[0066] As described above, the acoustic monitoring system according to Embodiment 1 has a function to estimate the state of the monitored area based on non-audio signals generated in the monitored area. Furthermore, as explained in detail in Examples 1 to 4, the acoustic monitoring system according to Embodiment 1 can perform alarm processing based on state estimation in various applications.

[0067] As a result, it is possible to realize an acoustic monitoring system that can appropriately monitor a target area based on sound signals, according to its purpose, while protecting privacy. In particular, it is expected to be effectively utilized in the following applications.

[0068] By estimating the residents' daily routines, measures such as fall prevention and occupancy confirmation can be implemented. • By estimating the source of sound during nighttime security operations, it is possible to monitor for intruders, broken windows, and other abnormal situations. In the event of a fire, by estimating the presence of people before the fire signal is received, information that can quickly assist in rescue operations and evacuation guidance can be issued.

[0069] In other words, the acoustic monitoring system according to this embodiment 1 can perform appropriate alarm processing based on necessary state estimation for various applications such as crime prevention, disaster prevention, and monitoring residents. Furthermore, while the acoustic monitoring system according to this embodiment 1 can function as a standalone system, its simple system configuration also has the advantage of being easily retrofitted to existing equipment such as automatic fire alarm systems. [Explanation of Symbols]

[0070] 1 monitoring area, 10-14 microphones, 20 control unit, 30 memory unit.

Claims

[Claim 1] A microphone that captures acoustic information generated in the monitored area, A memory unit that stores sound signals, A control unit extracts information other than the conversation content contained in the acoustic information as non-audio information, generates data as an audio signal as relating the non-audio information and the time of sound acquisition and stores it in the storage unit, performs state estimation of the monitored area based on the audio signal, and performs an alarm process to notify the state estimation result. Equipped with, When the control unit receives a fire signal indicating that a fire has occurred, it performs state estimation by estimating the presence of a person in the monitored area based on the sound signal stored in the memory unit during a predetermined time period prior to the time the fire signal was received. Acoustic monitoring system.

Images