JP2020024694A - Alarm sound detector - Google Patents

Abstract

To provide an alarm sound detector capable of: automatically and accurately detecting an alarm sound of a medical device with an alarm function; preventing an alarm sound from being missed by notifying a remote user terminal of an alert; and reducing mental and physical burdens on medical staff and attendants.SOLUTION: An alarm sound detector comprises: a volume determination unit 52 that determines whether the volume of a voice signal is greater than or equal to a volume threshold; a voice analysis data acquisition unit 53 for acquiring voice analysis data; a frequency determination unit 34 for determining whether the frequency of the voice analysis data is within an alarm detection range; a duration determination unit 55 for determining whether a duration that continues within the alarm detection range is equal to or longer than an alarm detection time; and an alert signal transmission unit 56 for transmitting an alert signal to a server 11, when the duration is equal to or longer than the alarm detection time.SELECTED DRAWING: Figure 1

Description

  TECHNICAL FIELD The present invention relates to an alarm sound detection device for detecting an alarm sound of a medical device with an alarm function, such as a ventilator or an artificial dialysis device.

  Conventionally, medical devices with an alarm function, such as a ventilator and an artificial dialysis device, are configured to emit an alarm sound to alert medical staff or the like when an abnormality occurs in the state of a patient or the state of a medical device. . For this reason, a patient who uses a medical device with an alarm function in a hospital has restrictions such that the medical staff can only enter a hospital room near the nurse station so that the medical staff does not miss the alarm sound.

  As a solution to such a problem, for example, Japanese Patent Application Laid-Open No. 2002-177225 discloses a centralized system in which alarm signals of a plurality of medical devices are transmitted to a nurse station using an existing nurse call line and can be monitored. A monitor system has been proposed (Patent Document 1).

JP-A-2002-177225

  However, even in the invention described in Patent Literature 1, at night or the like, the number of medical staff present at the nurse station is small, so the absence of medical staff or the case of responding to other tasks or emergency situations In some cases, the user may miss the alarm sound. In addition, when there are many inpatients, there is not always an empty room near the nurse station. Furthermore, the medical staff needs to sharpen their nerves so that they do not miss the alarm sound while performing other tasks, and there is great mental fatigue.

  When a medical device with an alarm function is used at home, the centralized monitoring system described in Patent Document 1 cannot be introduced. For this reason, while using the medical device with the alarm function, there is always a problem that someone must be present nearby and the attendant is greatly restricted from going out or going to bed.

  Furthermore, in hospitals and homes where medical equipment with alarm functions are used, there are various sounds such as the sound of televisions and radios, the ringtones of smartphones and mobile phones, the conversation of patients and visitors, and the sound of meals for serving meals. It is overflowing with. For this reason, it may be difficult for human ears to recognize the alarm sound even if it is nearby.

  The present invention has been made in order to solve such problems, and detects an alarm sound of a medical device with an alarm function automatically and with high accuracy, and notifies an alert to a remote user terminal. It is another object of the present invention to provide an alarm sound detection device that can prevent the alarm sound from being overlooked and reduce the mental and physical burden on medical staff and attendants.

  The alarm sound detection device according to the present invention detects an alarm sound of a medical device with an alarm function automatically and with high accuracy, and notifies an alert to a remote user terminal, thereby preventing the alarm sound from being overlooked. An alarm sound detection device for detecting an alarm sound of a medical device with an alarm function in order to solve a problem of reducing mental and physical burdens on medical staff and attendants, wherein the alarm sound is input via a voice input unit. The volume of the sound signal that has been set, a volume determination unit that determines whether or not the volume threshold is set to determine whether or not the volume of the alarm sound, the volume is greater than or equal to the volume threshold A voice analysis data acquisition unit that obtains a voice signal for a predetermined analysis time as voice analysis data to be analyzed from the time when the voice analysis is performed; However, a frequency determination unit that determines whether the alarm sound is within an alarm detection range corresponding to the frequency of the alarm sound, and a continuation time during which the frequency of the voice analysis data continues to be within the alarm detection range is a predetermined time. A duration determining unit for determining whether or not the alarm detection time is longer than the alarm detection time, and an alert signal for instructing a pre-registered user terminal to notify an alert when the duration is longer than the alarm detection time. And an alert signal transmitting unit that transmits the alert signal.

  Further, as one aspect of the present invention, an abnormal sound of a medical device with an alarm function is detected quickly and automatically by accurately detecting an abnormal sound of the medical device with an alarm function, and an alert is notified to a remote user terminal. In order to solve the problem of recognizing and preventing a medical accident beforehand, when the frequency of the voice analysis data is out of the alarm detection range, or when the duration is less than the alarm detection time, the frequency determination is performed. The unit determines whether or not the frequency of the voice analysis data is within an abnormal sound detection range corresponding to the frequency of the abnormal sound, and when the frequency of the voice analysis data is within the abnormal sound detection range, the continuation is performed. The time determination unit determines whether or not the continuation time during which the frequency of the voice analysis data continues to be within the abnormal sound detection range is equal to or longer than a predetermined abnormal sound detection time. When the duration time becomes the abnormal sound detection time or more, the alert signal transmission unit may transmit the alert signal to the server.

  Further, as one aspect of the present invention, in order to solve the problem of preventing noise generated around a medical device with an alarm function from being erroneously detected as an alarm sound or an abnormal sound, the volume threshold value of the audio signal is set as the volume threshold value. While setting the upper threshold and the lower threshold in the amplitude, the frequency determination unit, when specifying the frequency of the voice analysis data, of the waveform of the voice analysis data, both the upper threshold and the lower threshold May be counted, and the frequency may be specified based on the counted number.

  Further, as one aspect of the present invention, in order to solve the problem of making a user surely recognize the occurrence of an alarm sound or an abnormal sound, a predetermined retransmission setting time has elapsed after the alert signal was transmitted. The information processing apparatus may further include a retransmission determination unit configured to determine whether the alert signal has been transmitted. When the retransmission setting time has elapsed, the alert signal transmission unit may retransmit the alert signal to the server.

  Further, as one aspect of the present invention, in order to solve the problem of automatically and accurately detecting even a low volume or low frequency alarm sound, a rhythm which is a temporal change of the frequency of the voice analysis data is used. A rhythm pattern detection unit that detects a pattern, and a rhythm pattern determination unit that determines whether a rhythm pattern of the voice analysis data matches a rhythm pattern of the alarm sound, wherein the alert signal transmission unit includes: When the rhythm pattern of the voice analysis data matches the rhythm pattern of the alarm sound, an alert signal for instructing a user terminal registered in advance to notify an alert may be transmitted to the server.

  Further, the alarm sound detection device according to the present invention automatically and accurately detects even a low volume or low frequency alarm sound, and notifies an alert to a remote user terminal, thereby overlooking the alarm sound. An alarm sound detection device for detecting an alarm sound of a medical device with an alarm function in order to solve the problem of preventing mental health and physical burdens on medical staff and attendants, while preventing sound from being input. A volume determination unit that determines whether or not the volume of the audio signal input through the means is equal to or higher than a volume threshold set to determine whether or not the volume is the alarm sound; A voice analysis data acquisition unit that acquires a voice signal for a predetermined analysis time as voice analysis data to be analyzed from a time when the volume exceeds the volume threshold value; A rhythm pattern detection unit that detects a rhythm pattern that is a temporal change in the frequency of the data, a rhythm pattern determination unit that determines whether a rhythm pattern of the voice analysis data matches a rhythm pattern of the alarm sound, When the rhythm pattern of the voice analysis data matches the rhythm pattern of the alarm sound, an alert signal transmitting unit that transmits to the server an alert signal for instructing a user terminal registered in advance to notify an alert.

Further, as one aspect of the present invention, in order to solve the problem of detecting a rhythm pattern of an alarm sound with high accuracy from acoustic analysis data including noise, the rhythm pattern detection unit includes the following steps (a) to (a). By executing g), a rhythm pattern of the voice analysis data may be detected.
(A) removing measurement noise from the frequency of the voice analysis data;
(B) Emphasize relatively large frequency changes in the voice analysis data.
(C) Excluding relatively small frequency changes in the voice analysis data.
(D) detecting rising and falling of a frequency change in the voice analysis data;
(E) detecting a rising time at which the rising is first detected and a falling time at which the falling is detected first;
(F) Estimating a peak time at which the frequency of the voice analysis data reaches a peak based on the rise time and the fall time.
(G) detecting a time-series pattern indicated by the peak time as a rhythm pattern of the voice analysis data.

  Further, as one aspect of the present invention, in order to solve the problem of estimating a peak time with high accuracy, the rhythm pattern detection unit calculates an intermediate point between the rising time and the falling time as the peak time. You may.

  According to the present invention, an alarm sound of a medical device with an alarm function is automatically and accurately detected, and an alert is notified to a remote user terminal. It can reduce the mental and physical burden on people.

It is a block diagram showing a 1st embodiment of an alarm sound detector concerning the present invention. 5 is a waveform illustrating an example of an audio signal input from an audio input unit in the first embodiment. 5 is a waveform illustrating an example of voice analysis data in the first embodiment. It is a flowchart which shows the process performed by the alarm sound detection apparatus of 1st Embodiment. It is a block diagram showing a 2nd embodiment of the alarm sound detector concerning the present invention. It is a flowchart which shows the process performed by the alarm sound detection apparatus of 2nd Embodiment. It is a flowchart which shows the detail of the rhythm pattern detection processing in 2nd Embodiment. It is a graph which shows the frequency (count value) of audio | voice analysis data in 2nd Embodiment. 9 is a graph showing a frequency after a low-pass filter process in the second embodiment. 9 is a graph showing a frequency after a finite impulse response filter process in the second embodiment. It is a graph which shows the frequency after a truncation process in 2nd Embodiment. 10 is a graph illustrating a state of frequency change in FIG. 9 in the second embodiment. 9 is a graph showing frequency peak times in the second embodiment.

  Hereinafter, a first embodiment of an alarm sound detection device according to the present invention will be described with reference to the drawings.

  As shown in FIG. 1, the alarm sound detection device 1 </ b> A according to the first embodiment detects an alarm sound emitted by the medical device 10 with an alarm function and transmits an alert signal to the server 11, thereby enabling a user who has registered in advance. This is for notifying the terminal 12 of an alert.

  Specifically, the alarm sound detection device 1A is configured by a computer such as a microcomputer built in a small housing, and mainly transmits data between the voice input means 2 for inputting voice and the server 11. A communication unit 3 for performing communication, a storage unit 4 for storing the alarm sound detection program 1a and various data of the first embodiment, and an arithmetic processing unit 5 that executes various arithmetic processes and functions as each component described below. have. Hereinafter, each component will be described in detail.

  The voice input means 2 inputs voice as a voice signal. In the first embodiment, the voice input unit 2 is configured by a microphone module, and includes all sounds generated around the alarm sound detection device 1A, including an alarm sound and an abnormal sound emitted by the medical device 10 with an alarm function. Is converted into an audio signal and input. Specifically, as shown in FIG. 2, the audio signal is input as a waveform having an amplitude corresponding to the volume.

  In the present invention, the alarm sound is a concept including all alarm sounds for warning medical staff and attendants when an abnormality occurs in the state of a patient or the state of a medical device. In addition, the abnormal sound includes all abnormal sounds that occur when any mechanical abnormality occurs in the medical device 10 with an alarm function, such as an air blow sound that occurs when the suction hose is disconnected in the ventilator. It is a concept.

  The communication means 3 implements a communication function in the alarm sound detection device 1A. In the first embodiment, the communication unit 3 is configured by a wireless module compatible with a wireless LAN standard such as Wi-Fi (registered trademark). The wireless communication is performed with the server 11 via a Wi-Fi access point (not shown). The standard of the communication means 3 is not limited to wireless communication, but may be wired communication.

  Note that, in the present invention, the medical device 10 with an alarm function is any medical device that emits an alarm sound when an abnormality occurs in the state of a patient or the state of a medical device, such as a ventilator or an artificial dialysis device. This is a concept that includes medical equipment. Further, in the present invention, the server 11 may receive an alert signal from the alarm sound detection device 1A and notify an alert to the user terminal 12 registered in advance, such as a cloud server provided in a cloud environment. The concept includes all possible servers 11. Further, in the present invention, the user terminal 12 includes all terminals, such as a smartphone and a tablet, that can receive an alert from the server 11 and notify the user of the fact by voice, display, vibration, or the like. It is a concept.

  The storage means 4 is composed of a ROM (Read Only Memory), a RAM (Random Access Memory), a flash memory, etc., and stores various data, and works when the arithmetic processing means 5 executes various arithmetic processing. It functions as an area. In the first embodiment, the storage unit 4 has a program storage unit 41 and a parameter storage unit 42, as shown in FIG.

  In the program storage unit 41, the alarm sound detection program 1a of the first embodiment is installed. Then, the arithmetic processing means 5 executes the alarm sound detection program 1a, thereby causing a computer as the alarm sound detection device 1A to function as each component described later.

  The parameter storage unit 42 stores various parameters used by the alarm sound detection device 1A. In the first embodiment, as shown in FIG. 1, the parameter storage unit 42 stores a volume threshold, an alarm detection range, an alarm detection time, an abnormal sound detection range, an abnormal sound detection time, a retransmission setting time. And memorize. Hereinafter, each parameter will be described.

  The volume threshold is a parameter set to determine whether or not the volume of the audio signal corresponds to the volume of the alarm sound. The alarm detection range is a parameter for determining whether or not the frequency of voice analysis data described later corresponds to the frequency of an alarm sound. Since the frequency is the number of waves per unit time, in the first embodiment, the number of waves is set as the alarm detection range. Specifically, as the lower limit of the number of waves within the analysis time (20 milliseconds) described later, 40 times are set as the volume threshold. Thus, a frequency of 2000 Hz (= 2000 times / second = 40 times / 20 milliseconds) or more is set as the alarm detection range.

  In the first embodiment, only the lower limit is set as the alarm detection range so that various alarm sounds can be detected without omission. However, the present invention is not limited to this configuration, and the upper limit (for example, 4000 Hz) may be set together. Accordingly, only the alarm sound of the specific medical device 10 with an alarm function is detected with high accuracy, and erroneous detection of noise is reduced.

  The alarm detection time is a parameter for determining whether or not a continuation time during which the frequency of voice analysis data described later remains within the alarm detection range corresponds to an alarm sound. The abnormal sound detection range is a parameter for determining whether or not the frequency of the voice analysis data described later corresponds to the frequency of the abnormal sound. The abnormal sound detection time is a parameter for determining whether or not a continuous time in which the frequency of voice analysis data described later remains within the abnormal sound detection range corresponds to an abnormal sound. The retransmission setting time is a parameter for determining whether or not to retransmit the alert signal to the server 11.

  The arithmetic processing unit 5 is configured by a CPU (Central Processing Unit) and the like, and executes the alarm sound detection program 1a installed in the program storage unit 41, thereby, as shown in FIG. , A volume determination unit 52, a voice analysis data acquisition unit 53, a frequency determination unit 54, a duration determination unit 55, an alert signal transmission unit 56, and an alert retransmission determination unit 57. . Hereinafter, each component will be described in more detail.

  The audio signal acquisition unit 51 acquires an audio signal of a sound generated around the alarm sound detection device 1A. In the first embodiment, the audio signal acquisition unit 51 always acquires an audio signal input via the audio input unit 2.

  The volume determination unit 52 determines the volume of the audio signal. In the first embodiment, the volume determination unit 52 determines whether the volume of the audio signal acquired by the audio signal acquisition unit 51 is equal to or higher than the volume threshold set in the parameter storage unit 42. As a result, only an audio signal having a volume equal to or higher than the predetermined volume threshold is extracted as a candidate for the alarm sound.

  The voice analysis data obtaining unit 53 obtains voice analysis data to be analyzed. In the first embodiment, the voice analysis data acquisition unit 53 performs a predetermined analysis time from when the volume is determined by the volume determination unit 52 to be equal to or higher than the volume threshold, that is, when a volume equal to or higher than the volume threshold is detected. The voice signal of the minute is obtained as voice analysis data. In the first embodiment, as described above, 20 milliseconds is set as the analysis time.

  The frequency determination unit 54 determines the frequency of the audio analysis data. In the first embodiment, the frequency determination unit 54 determines whether the frequency of the audio analysis data acquired by the audio analysis data acquisition unit 53 is within the alarm detection range set in the parameter storage unit 42. I do. Thereby, only the voice analysis data having a frequency within the alarm detection range is extracted as a candidate for the alarm sound.

  In the first embodiment, an upper threshold and a lower threshold in the amplitude of the audio signal are set as the volume threshold. Then, when specifying the frequency of the voice analysis data, the frequency determination unit 54 counts only the waveform having an amplitude exceeding both the upper threshold and the lower threshold among the waveforms of the voice analysis data, as shown in FIG. The frequency is specified based on the count number.

  For example, in the example shown in FIG. 3, within the analysis time (20 milliseconds), there are nine waveforms having amplitudes exceeding both the upper threshold and the lower threshold. Therefore, when converting 9 times / 20 milliseconds into the number of times per unit time, it is 450 times / second, so that the frequency is specified as 450 Hz. Thereby, an irregular voice that exceeds only one of the upper threshold and the lower threshold is not counted, and an accurate frequency is specified, thereby preventing erroneous detection of noise.

  The duration determination unit 55 determines the duration for which the frequency of the voice analysis data continues within a predetermined range. In the first embodiment, when the frequency determination unit 54 determines that the frequency of the voice analysis data continues to be within the alarm detection range by the frequency determination unit 54, the duration is determined by the alarm detection time. It is determined whether or not this is the case. Accordingly, noise such as a single frictional sound or burst sound is determined to be voice analysis data shorter than the alarm detection time, thereby preventing erroneous detection.

  In the first embodiment, the frequency determination unit 54 and the duration determination unit 55 separately determine the frequency of the abnormal sound even when the frequency and the duration of the voice analysis data do not correspond to the frequency and the duration of the alarm sound. And has a function of determining whether the time corresponds to the duration. Specifically, when the frequency of the voice analysis data is outside the alarm detection range, or when the duration of the frequency of the voice analysis data continuing to be within the alarm detection range is less than the alarm detection time, It is determined whether or not the frequency of the voice analysis data is within the abnormal sound detection range set in the parameter storage unit 42. Then, when the frequency of the voice analysis data is within the abnormal sound detection range, the duration determining unit 55 determines that the continuous time that the frequency of the voice analysis data continues to be within the abnormal sound detection range is equal to or longer than the abnormal sound detection time. Is determined.

  The alert signal transmitting unit 56 transmits an alert signal to the server 11. In the first embodiment, information (e-mail addresses and the like) on all the user terminals 12 to be notified of an alert is registered in the server 11 in advance. The alert signal is a signal for instructing the server 11 to notify the user terminal 12 registered in advance of the alert. When the duration determining unit 55 determines that the duration for which the frequency of the voice analysis data continues within the alarm detection range is equal to or longer than the alarm detection time, the alert signal transmission unit 56 communicates via the communication unit 3. An alert signal is transmitted to the server 11.

  In the first embodiment, the alert signal transmission unit 56 determines by the duration determination unit 55 that the duration during which the frequency of the voice analysis data continues within the abnormal sound detection range is equal to or longer than the abnormal sound detection time. Then, an alert signal is transmitted to the server 11 via the communication means 3 when the server 1 is in the standby state.

  The alert retransmission determination unit 57 determines whether or not to retransmit the alert signal. In the first embodiment, the alert retransmission determination unit 57 determines whether or not the retransmission set time set in the parameter storage unit 42 has elapsed after the alert signal was transmitted by the alert signal transmission unit 56. . When the retransmission setting time has elapsed, the alert may not be transmitted to the medical staff or the like. Therefore, the alert retransmission determination unit 57 sends the alert signal to the alert signal transmission unit 56 to retransmit the alert signal to the server 11. To instruct.

  On the other hand, if the operation of canceling the alert signal is performed before the elapse of the retransmission set time, the alert retransmission determination unit 57 terminates the determination processing by determining that the user has noticed the alert. Note that the operation for canceling the alert signal includes, for example, pressing a button provided on the medical device 10 with an alarm function.

  Next, the operation of the alarm sound detection device 1A of the first embodiment will be described with reference to the flowchart of FIG.

  When detecting the alarm sound or the abnormal sound of the medical device 10 with the alarm function using the alarm sound detection device 1A of the first embodiment, the volume threshold value, the alarm detection range, An alarm detection time, an abnormal sound detection range, an abnormal sound detection time, and a retransmission setting time are set in advance. In addition, information about all the user terminals 12 to which an alert is to be notified is registered in the server 11 in advance.

  Next, when the alarm sound detection device 1A is installed around the medical device 10 with an alarm function, the audio signal acquisition unit 51 acquires the audio signal of the ambient sound (step S1). Then, it is determined by the volume determination unit 52 whether or not the volume of the audio signal is equal to or higher than the volume threshold (step S2).

  If the result of determination in step S2 is that the volume of the audio signal is lower than the volume threshold (step S2: NO), the processing from step S1 is repeated. As a result, an audio signal having no volume corresponding to the alarm sound or the abnormal sound is left as noise other than the alarm sound or the abnormal sound.

  On the other hand, if the volume of the audio signal is equal to or higher than the volume threshold (step S2: YES), the voice analysis data acquiring unit 53 determines that there is a possibility of an alarm sound or an abnormal sound, and analyzes the voice signal for a predetermined analysis time. It is acquired as data (step S3). Then, the frequency determination unit 54 determines whether or not the frequency of the voice analysis data is within the alarm detection range (Step S4).

  If the result of the determination in step S4 is that the frequency of the voice analysis data is not within the alarm detection range (step S4: NO), the process proceeds to step S6 described below as noise that is not an alarm sound. On the other hand, if the frequency of the voice analysis data is within the alarm detection range (step S4: YES), the duration determination unit 55 determines whether or not the duration that continues within the alarm detection range is equal to or longer than the alarm detection time. Is determined (step S5).

  If the result of determination in step S5 is that the duration is equal to or longer than the alarm detection time (step S5: YES), the alert signal transmitting unit 56 transmits an alert signal to the server 11 via the communication unit 3 (step S8). As a result, only an audio signal having a volume, frequency, and duration corresponding to the alarm sound of the medical device 10 with an alarm function is automatically and accurately detected as an alarm sound.

  Then, since an alert is notified from the server 11 to the user terminal 12 in response to the detection of the alarm sound, even if the user is far away from the medical device 10 with an alarm function and cannot directly hear the alarm sound, A situation in which an alarm sound is occurring can be notified. In addition, when the medical device 10 with an alarm function is used at home, the user terminal 12 is notified in real time that the alarm sound has been emitted even if the attendant is not nearby, so the physical and mental status of the attendant Burden is reduced.

  On the other hand, when the frequency of the voice analysis data is out of the alarm detection range (step S4: NO), or when the continuation of the frequency of the voice analysis data remaining within the alarm detection range is less than the alarm detection time (step S5: NO), the frequency determination unit 54 determines whether or not the frequency of the voice analysis data is within the abnormal sound detection range (Step S6).

  If the result of the determination in step S6 is that the frequency of the voice analysis data is not within the abnormal sound detection range (step S6: NO), the process returns to step S1, and the above-described processing is repeated. As a result, a sound signal having no frequency corresponding to the alarm sound or the abnormal sound is left as not the alarm sound or the abnormal sound.

  On the other hand, if the frequency of the voice analysis data is within the abnormal sound detection range (step S6: YES), the duration determining unit 55 continues for more than the abnormal sound detection time while remaining within the abnormal sound detection range. It is determined whether or not (step S7). As a result of the determination, when the duration is equal to or longer than the abnormal sound detection time (step S7: YES), the alert signal transmitting unit 56 transmits an alert signal to the server 11 via the communication unit 3 (step S8). Thus, an audio signal having a volume, frequency and duration corresponding to an abnormal sound is automatically and accurately detected as an abnormal sound.

  Then, since an alert is notified from the server 11 to the user terminal 12 in response to the detection of the abnormal sound, even if any abnormality occurs in the medical device 10 with the alarm function, the abnormality is promptly found and recognized. In addition, medical accidents are prevented. On the other hand, if the duration is not longer than the abnormal sound detection time (step S7: NO), the process returns to step S1 and repeats the above-described processing. As a result, even if the audio signal has a frequency corresponding to the abnormal sound, the sound signal having a short duration is not regarded as an abnormal sound.

  After the alert signal is transmitted to the server 11 (Step S8), the alert retransmission determination unit 57 determines whether the retransmission set time has elapsed (Step S9). As a result of the determination, if the operation of releasing the alert signal is performed before the retransmission set time elapses (step S9: NO) (step S10: YES), the process returns to step S1, and the above-described processing is repeated.

  On the other hand, if the reset operation of the alert signal has not been performed (step S10: NO) and the retransmission set time has elapsed (step S9: YES), the process returns to step S8, and the alert retransmission determination unit 57 outputs the alert signal. By instructing the alert signal transmitting unit 56 to retransmit the alert signal to the server 11, the alert signal is transmitted to the server 11. Thus, even if there is a user who does not notice the first alert, the alert is repeatedly notified until the alert signal is released, so that the occurrence of the alarm sound can be reliably recognized.

According to the alarm sound detection device 1A of the first embodiment as described above, the following effects can be obtained.
1. By automatically and highly accurately detecting the alarm sound of the medical device 10 with the alarm function and notifying the remote user terminal 12 of the alert, the occurrence of the alarm sound can be reliably transmitted, and the medical staff and attendant Physical and mental burden can be reduced.
2. The abnormal sound of the medical device 10 with the alarm function is detected automatically and with high accuracy, and an alert is notified to the remote user terminal 12, whereby the abnormality of the medical device 10 with the alarm function is quickly found and recognized, and the medical accident Can be prevented beforehand.
3. It is possible to prevent noise generated around a medical device with an alarm function from being erroneously detected as an alarm sound or an abnormal sound.
4. Since the alert is notified via the server 11, even if the user terminal 12 does not exist on the same network, the medical device 10 with the alarm function can be monitored and the alert can be notified.
5. Since the alert is notified via the server 11, even if the number of the user terminals 12 to be notified of the alert increases or decreases, it is not necessary to change the setting of the alarm sound detection device 1A.
6. Since the alarm sound detection device 1A only needs to transmit an alert signal only to the server 11, high security can be ensured.

  Next, a second embodiment of the alarm sound detection device according to the present invention will be described. Note that, in the configuration of the second embodiment, the same reference numerals are given to the same or corresponding components as those of the above-described first embodiment, and the description thereof will not be repeated.

  The feature of the second embodiment is that it has a function of detecting an alarm sound using a rhythm pattern, which is a temporal change in frequency, in addition to the function of detecting a frequency described in the first embodiment.

  Specifically, as shown in FIG. 5, in addition to the configuration of the first embodiment, the alarm sound detection device 1B of the second embodiment has various alarm functions in the parameter storage unit 42 of the storage unit 4. A rhythm pattern (temporal change in frequency) of an alarm sound unique to each medical device 10 is stored in advance. In addition, the arithmetic processing unit 5 separately functions as a rhythm pattern detection unit 58 and a rhythm pattern determination unit 59.

  The rhythm pattern detecting section 58 detects a rhythm pattern which is a temporal change of the frequency of the voice analysis data. In the second embodiment, each time the voice analysis data acquisition unit 53 acquires the voice analysis data, the rhythm pattern detection unit 58 combines the voice analysis data for a predetermined time acquired up to that time with the following process ( By executing a) to (g), a rhythm pattern of the voice analysis data is detected.

(A) Eliminate measurement noise from the frequency of voice analysis data.
(B) Emphasize relatively large frequency changes in voice analysis data.
(C) Exclude relatively small frequency changes in the voice analysis data.
(D) Detect rising and falling frequency changes in the voice analysis data.
(E) The rising time at which the rising is detected first and the falling time at which the falling is detected first are detected.
(F) Based on the rise time and the fall time, a peak time at which the frequency of the voice analysis data reaches a peak is estimated.
(G) A time-series pattern indicated by the peak time is detected as a rhythm pattern of the voice analysis data.

  Details of each step will be described later. In the second embodiment, the rising of the frequency change corresponds to the beginning of the sound, and means that the frequency is increasing. Further, the falling of the frequency change corresponds to the end of the sound, which means that the frequency is in a state of decreasing.

  The rhythm pattern determination unit 59 determines whether the rhythm pattern of the voice analysis data matches the rhythm pattern of the alarm sound. In the second embodiment, the rhythm pattern determination unit 59 compares the rhythm pattern detected by the rhythm pattern detection unit 58 with the rhythm pattern stored in the parameter storage unit 42. Then, it is determined whether or not the two coincide with each other within a predetermined allowable range.

  Next, the operation of the alarm sound detection device 1B of the second embodiment will be described with reference to the flowcharts of FIGS.

  In the second embodiment, as shown in FIG. 6, first, as in the first embodiment, when the audio signal of the surrounding sound is acquired by the audio signal acquisition unit 51 (step S1), the volume of the audio signal is reduced. The volume determination unit 52 determines whether or not the volume is equal to or higher than the volume threshold (step S2). At this time, in the second embodiment, since the erroneous detection of noise is reduced by the matching using the rhythm pattern, the volume threshold can be set small (narrow). As a result, the detection rate of an alarm sound with a small volume (small amplitude) is improved.

  Next, the voice analysis data obtaining unit 53 obtains voice signals for a predetermined analysis time (20 ms) as voice analysis data (step S3), and specifies the frequency (count value). As described above, the frequency of the audio analysis data is determined by counting only the waveform having an amplitude exceeding both the upper threshold and the lower threshold among the waveforms of the audio analysis data, and specifying the frequency based on the counted number.

  Subsequently, the rhythm pattern detection unit 58 to which the frequency (count value) of the voice analysis data is input detects a rhythm pattern that is a temporal change in the frequency of the voice analysis data (step S11). Hereinafter, specific processing contents in the rhythm pattern detection step will be described with reference to FIG.

  First, the rhythm pattern detection unit 58 removes measurement noise by using a low-pass filter (LPF: Low-Pass Filter) for the frequency (count value) of the voice analysis data as shown in FIG. 8 (step S111: Step (a)). As a result, as shown in FIG. 9, a portion where the frequency change is small is removed, and only a portion where the frequency change is small is left. In FIG. 9, the count value is multiplied by 100 in order to reduce the quantization error (rounding error) due to the low-pass filter.

  Next, the rhythm pattern detection unit 58 uses a finite impulse response (FIR) filter for the frequency (count value) after the low-pass filter processing, thereby obtaining a relatively large frequency as shown in FIG. The change is emphasized (step S112: above step (b)).

  Subsequently, the rhythm pattern detection unit 58 excludes a relatively small frequency change as shown in FIG. 11 by cutting off the frequency (count value) of the voice analysis data by a predetermined threshold value (count value). (Step S113: the above step (c)). As a result, in the next step S114, it is prevented that the rise or fall of a fine frequency change such as noise is determined, so that the rise or fall of the original alarm sound to be extracted is accurately detected. .

  In the second embodiment, as shown in FIG. 10, the threshold value of the frequency (count value) at the time of stepping is set to 1000. However, the present invention is not limited to this value, and may be increased or decreased as appropriate. You may.

  Next, the rhythm pattern detection unit 58 detects the rise and fall of the frequency change in the voice analysis data (step S114: the above step (d)). Specifically, the rhythm pattern detection unit 58 determines the frequency (count value) of the previously obtained voice analysis data and the frequency (count value) of the voice analysis data obtained this time in the count value (FIG. 11) after the truncation process. ) And compare. Then, as shown in FIG. 12, if the count value has increased, it is determined that the signal has risen. If the count value has decreased, it is determined that the signal has fallen.

  Subsequently, the rhythm pattern detection unit 58 detects the rising time at which the rising is detected first and the falling time at which the falling is detected first based on the rising and falling detected at step S114 ( Step S115: the above step (e)). Then, a peak time at which the frequency (count value) of the voice analysis data reaches a peak is estimated based on the rising time and the falling time (step S116: the above step (f)). Specifically, as shown in FIG. 13, the rhythm pattern detection unit 58 calculates an intermediate point between the rising time and the falling time as a peak time. This makes it possible to detect an error sound with a small error even for an alarm sound that has a long sound duration, has a gentle frequency peak, and is difficult to determine.

  Then, as shown in FIG. 13, the rhythm pattern detection unit 58 detects a time-series pattern indicated by the peak time calculated in step S116 as a rhythm pattern of the voice analysis data (step S117: the above step (g)). .

  Through the above processing, the rhythm pattern of the alarm sound of the medical device with the alarm function is detected with high accuracy even from acoustic analysis data including noise such as life noise.

  Returning to FIG. 6, the rhythm pattern determination unit 59 determines whether the rhythm pattern of the voice analysis data detected in step S11 matches the rhythm pattern of the alarm sound (step S12). If there is a matching rhythm pattern as a result of the determination (step S12: YES), the alert signal transmitting unit 56 transmits an alert signal to the server 11 via the communication unit 3 (step S8). As a result, a voice signal having a rhythm pattern equivalent to the rhythm pattern of the alarm sound of the medical device 10 with an alarm function is automatically and accurately detected as an alarm sound.

  On the other hand, if there is no matching rhythm pattern (step S12: NO), the process proceeds to step S4 and subsequent steps described in the first embodiment, and the volume, frequency, and duration corresponding to the alarm sound of the medical device 10 with the alarm function. Is automatically and accurately detected as an alarm sound.

According to the second embodiment as described above, the following effects are obtained in addition to the functions and effects of the first embodiment described above.
1. Even an alarm sound whose volume is low and a correct frequency cannot be measured can be detected based on a rhythm pattern.
2. It is possible to detect an alarm sound having a frequency as low as that of a living noise without erroneously detecting a living noise or the like as an alarm sound.

  Note that the alarm sound detection devices 1A and 1B according to the present invention are not limited to the above-described embodiments, and can be appropriately changed.

  For example, in each of the above-described embodiments, the alarm sound detectors 1A and 1B have an abnormal sound detection function and an alert retransmission function in addition to the alarm sound detection function. May not necessarily be provided. That is, the alarm sound detectors 1A and 1B according to the present invention need only have at least an alarm sound detection function.

  In the above-described first embodiment, when detecting an alarm sound or an abnormal sound, the determination is made using three parameters of a volume, a frequency, and a duration. However, by adjusting the value of each parameter, the detection may be performed using any one or two parameters. For example, by setting the volume threshold to zero, the determination is made substantially only by frequency and duration.

  Furthermore, in the above-described first embodiment, the alarm sound detection device 1A having only the detection function based on the frequency will be described. In the above-described second embodiment, the alarm sound having both the detection function based on the frequency and the detection function based on the rhythm pattern will be described. Although the sound detection device 1B has been described, an alarm sound detection device having only a rhythm pattern detection function may be used.

  In the above-described second embodiment, a plurality of rhythm patterns are stored in the parameter storage unit 42 so that various alarm sounds of the medical device 10 with an alarm function can be detected. It is not done. For example, when the alarm sound detection device 1B is used at home, only the rhythm pattern of the alarm sound emitted from the medical device 10 with an alarm function used in the house may be stored.

1A Alarm sound detection device (first embodiment)
1B Alarm Sound Detection Device (Second Embodiment)
1a Alarm sound detection program 2 Voice input means 3 Communication means 4 Storage means 5 Arithmetic processing means 10 Medical equipment with alarm function 11 Server 12 User terminal 41 Program storage unit 42 Parameter storage unit 51 Audio signal acquisition unit 52 Volume determination unit 53 Audio analysis Data acquisition unit 54 frequency determination unit 55 duration determination unit 56 alert signal transmission unit 57 alert retransmission determination unit 58 rhythm pattern detection unit 59 rhythm pattern determination unit

Claims (8)

An alarm sound detection device that detects an alarm sound of a medical device with an alarm function,
A volume determination unit that determines whether the volume of the voice signal input via the voice input unit is equal to or greater than a volume threshold set to determine whether the volume is the volume of the alarm sound,
A voice analysis data acquisition unit that obtains a voice signal for a predetermined analysis time as voice analysis data to be analyzed from when the volume is equal to or greater than the volume threshold,
A frequency determination unit that determines whether the frequency of the voice analysis data is within an alarm detection range corresponding to the frequency of the alarm sound,
A duration determining unit that determines whether or not the duration of the frequency of the voice analysis data that continues within the alarm detection range is equal to or longer than a predetermined alarm detection time,
When the duration is equal to or longer than the alarm detection time, an alert signal transmitting unit that transmits an alert signal instructing to notify an alert to a user terminal registered in advance to a server,
An alarm sound detection device comprising: If the frequency of the voice analysis data is out of the alarm detection range, or if the duration is less than the alarm detection time, the frequency determination unit may determine that the frequency of the voice analysis data is abnormal corresponding to the frequency of the abnormal sound. Determine whether it is within the sound detection range,
When the frequency of the voice analysis data is within the abnormal sound detection range, the duration determining unit determines that the duration of the frequency of the voice analysis data continuing to be within the abnormal sound detection range is a predetermined abnormal sound detection time. Judge whether it is more than time,
The alarm sound detection device according to claim 1, wherein the alert signal transmission unit transmits the alert signal to the server when the duration time is equal to or longer than the abnormal sound detection time.   While setting an upper limit threshold and a lower limit threshold in the amplitude of the audio signal as the volume threshold, the frequency determination unit, when specifying the frequency of the audio analysis data, of the waveform of the audio analysis data, 3. The alarm sound detection device according to claim 1, wherein only a waveform having an amplitude exceeding both the upper threshold and the lower threshold is counted, and the frequency is specified based on the count. 4. After the alert signal is transmitted, has a retransmission determination unit to determine whether a predetermined retransmission set time has elapsed,
4. The alarm sound detection device according to claim 1, wherein the alert signal transmitting unit retransmits the alert signal to the server when the retransmission set time has elapsed. 5. A rhythm pattern detection unit that detects a rhythm pattern that is a temporal change in the frequency of the voice analysis data,
A rhythm pattern determination unit that determines whether a rhythm pattern of the voice analysis data matches a rhythm pattern of the alarm sound,
The alert signal transmitting unit, when a rhythm pattern of the voice analysis data matches a rhythm pattern of the alarm sound, transmits an alert signal instructing a user terminal registered in advance to notify an alert to a server. The alarm sound detection device according to any one of claims 1 to 4. An alarm sound detection device that detects an alarm sound of a medical device with an alarm function,
A volume determination unit that determines whether the volume of the audio signal input via the audio input unit is equal to or greater than a volume threshold set to determine whether the volume is the alarm sound,
A voice analysis data acquisition unit that obtains a voice signal for a predetermined analysis time as voice analysis data to be analyzed from when the volume is equal to or greater than the volume threshold,
A rhythm pattern detection unit that detects a rhythm pattern that is a temporal change in the frequency of the voice analysis data,
A rhythm pattern determination unit that determines whether the rhythm pattern of the voice analysis data matches the rhythm pattern of the alarm sound,
When the rhythm pattern of the voice analysis data matches the rhythm pattern of the alarm sound, an alert signal transmitting unit that transmits to the server an alert signal instructing to notify an alert to a user terminal registered in advance,
An alarm sound detection device comprising: The alarm sound detection device according to claim 5, wherein the rhythm pattern detection unit detects a rhythm pattern of the voice analysis data by performing the following steps (a) to (g);
(A) removing measurement noise from the frequency of the voice analysis data;
(B) emphasizing relatively large frequency changes in the audio analysis data;
(C) excluding relatively small frequency changes in the audio analysis data;
(D) detecting rising and falling of a frequency change in the voice analysis data;
(E) detecting a rise time when the rise is first detected and a fall time when the fall is detected first;
(F) estimating a peak time at which the frequency of the voice analysis data reaches a peak based on the rise time and the fall time;
(G) detecting a time-series pattern indicated by the peak time as a rhythm pattern of the voice analysis data. The alarm sound detection device according to claim 7, wherein the rhythm pattern detection unit calculates an intermediate point between the rising time and the falling time as the peak time.