Method and system for detecting acoustic phenomena

Abstract

The invention relates to a detection system (1) comprising N detection devices (10) distributed throughout an area to be monitored (Z), each device comprising an acoustic sensor and a processing unit; - a second operating mode in which the sounds captured by the detection device are processed, the second operating mode corresponding to a second level of power consumption higher than the first level; each detection device (10) operates by default in a first mode that corresponds to a first very low level of power consumption of the device in which the first analysis on the electrical signal delivered by the acoustic sensor is performed by the processing block, and in which the electrical signal is not permitted to be recorded; and, as soon as the first analysis detects a predefined acoustic phenomenon, the detection device (10) alerts the N-1 other detection devices (10) of the system; upon receiving the alert, each of the N-1 other detection devices (10) then switches to a second mode that triggers: - a step of transmitting, to a recording unit, the electrical signal delivered by its acoustic sensor after receiving the alert; - a step of recording, by the recording unit, the transmitted electrical signal; and - a second analysis step of analysing the recorded electrical signal in order to detect acoustic phenomena, the second analysis step implementing algorithmic processing that corresponds to a second level of power consumption higher than the first level.

Description

DESCRIPTION Method and system for detecting acoustic phenomena Technical field:

[0001] The invention lies in the field of monitoring acoustic phenomena. Previous technique:

[0002] The invention relates particularly to the monitoring of acoustic phenomena using a detection system comprising a network of acoustic sensors each coupled to a processing block, allowing the real-time detection of acoustic signatures.

[0003] Limiting the electrical energy consumed by such a system is an important issue, especially when the sensors are embedded.

[0004] Therefore, a solution is needed for such a detection system that can continuously detect acoustic signatures in real time, but with very low energy consumption. Summary of the invention:

[0005] To this end, according to a first aspect, the present invention describes a method for detecting acoustic phenomena using a detection system comprising N detection devices distributed in an area to be monitored, N > 2 and connected to each other by telecommunication links, each detection device comprising an acoustic sensor capturing sounds and delivering an electrical signal according to the captured sounds and an electrical signal processing unit, each detection device is adapted to operate alternatively according to: - a first operating mode during which the detection device performs an initial analysis to detect acoustic phenomena, corresponding to a first level of electrical consumption of the device, known as the very low power consumption detection mode; and - a second operating mode in which the sounds captured by the detection device are processed, corresponding to a second level of power consumption higher than the first level; according to which: each detection device operates by default in the first mode, in which the processing unit performs the initial analysis on the electrical signal delivered by the acoustic sensor, and in which the recording of said electrical signal is not permitted; and as soon as said initial analysis detects a predefined acoustic phenomenon, said detection device alerts the N-1 other detection devices in the system via telecommunication links; upon receiving the alert, each of the N-1 other detection devices then switches to the second mode, in which it is triggered: - the transmission, to a recording unit, of the electrical signal delivered by its acoustic sensor after receiving the alert, - the recording, by said recording unit, of the transmitted electrical signal and - a second analysis of the recorded electrical signal, to detect acoustic phenomena, implementing an algorithmic processing corresponding to a second level of electrical consumption higher than the first level.

[0006] The invention exploits the coupling between sensors and the propagation of the wave from one sensor to another to perform in-depth analyses on the signal with extremely low energy consumption.

[0007] Following the alert triggered by one of the N detection devices, the other N-1 devices can thus be switched to a second mode associated with the implementation by the system of an in-depth detection processing, and this even before the acoustic signal that triggered the alert reaches them, allowing a complete capture of the signal in question, whereas the detection device that first detected the signal could not have recorded it (this would have required the signal to be perpetually being recorded, which, in most situations, is not sustainable in terms of energy and storage).

[0008] In some embodiments, such a process will further include at least one of the following features:

[0009] - the process includes one of the following provisions: when the detection device is in said second mode: the recording is carried out by an electronic recording block of said detection device and / or the second analysis is carried out by the processing block of said detection device; when the detection device is in said second mode, it transmits the electrical signal to a remote analysis device via a teletransmission link and the second analysis is carried out by said remote analysis device;

[0010] - in the second mode, the electrical signals of the N-1 devices are recorded in a synchronized manner, the detection devices having been previously synchronized with each other and / or a synchronization being implemented by said detection devices on the basis of the alert or another signal;

[0011] - the direction of arrival of an acoustic phenomenon detected by the second analysis in the second mode is determined based on the recordings of electrical signals delivered by the sensors of several of the N-1 detection devices following the receipt of the alert;

[0012] - the relative positions of the sensors with respect to each other is determined according to at least a temporal length of the acoustic signatures sought and the medium in which the detection devices are placed, in which the sound propagates;

[0013] - the inter-sensor distance is chosen so that a sound signal can be detected by a detection device in very low power mode among the N detection devices before said sound signal reaches at least the sensor of one of the N-1 other detection devices.

[0014] According to another aspect, the invention describes a system for detecting acoustic phenomena comprising N detection devices adapted to be distributed in an area to be monitored, N > 2 and connected to each other by telecommunication links, each detection device comprising an acoustic sensor, adapted to capture sounds and deliver an electrical signal according to the captured sounds, and an electrical signal processing block, each detection device is adapted to operate alternatively according to: - a first operating mode during which the detection device performs an initial analysis to detect acoustic phenomena, corresponding to a first level of electrical consumption of the device, known as the very low power consumption detection mode; and - a second operating mode in which the sounds captured by the detection device are processed, corresponding to a second level of power consumption higher than the first level; in which: each detection device is adapted to operate by default in the first mode, in which the processing unit is adapted to perform the first analysis on the electrical signal delivered by the acoustic sensor, and in which the recording of said electrical signal is not permitted; and as soon as said first analysis detects a predefined acoustic phenomenon, said detection device is adapted to alert the N-1 other detection devices in the system via telecommunication links; upon receiving the alert, each of the N-1 other detection devices is adapted to switch to the second mode, in which the following operations are triggered: - the transmission, to a recording unit, of the electrical signal delivered by its acoustic sensor after receiving the alert, - the recording, by said recording unit, of the transmitted electrical signal and - a second analysis of the recorded electrical signal, to detect acoustic phenomena, implementing an algorithmic processing corresponding to a second level of electrical consumption higher than the first level.

[0015] In some embodiments, such a system will further include at least one of the following features:

[0016] - the acoustic phenomenon detection system includes one of the following provisions: when the detection device is in said second mode: the recording is carried out by an electronic recording block of said detection device and / or the second analysis is carried out by the processing block of said detection device; when the detection device is in said second mode, it is adapted to transmit the electrical signal to a remote analysis device via a teletransmission link and the second analysis is carried out by said remote analysis device;

[0017] - in the second mode, the electrical signals of the N-1 devices are recorded in a synchronized manner, the detection devices having been previously synchronized with each other and / or a synchronization being implemented by said detection devices on the basis of the alert or another signal;

[0018] - the system is adapted to determine the direction of arrival of an acoustic phenomenon detected by the second analysis in the second mode based on the recordings of electrical signals delivered by the sensors of several of the N-1 detection devices following the receipt of the alert. Brief description of the figures:

[0019] The invention will be better understood and other features, details and advantages will become clearer from the following description, given by way of non-limiting example, and from the accompanying figures, given by way of example.

[0020] Figure 1 is an illustration of an acoustic phenomenon monitoring system in one embodiment of the invention;

[0021] Figure 2 is a schematic representation of a detection device in one embodiment of the invention;

[0022] Figure 3 represents the steps of a method for monitoring acoustic phenomena in one embodiment of the invention.

[0023] Identical references may be used in different figures when they designate the same or comparable elements. Detailed description:

[0024] Figure 1 represents a system 1 for monitoring acoustic phenomena within a predefined area of ​​interest to be monitored, named Z. System 1 comprises a network including N detection devices 10, N being an integer greater than or equal to 2.

[0025] The Z-zone is, depending on the case, in one, two or three dimensions and can take various shapes and sizes.

[0026] In this case, the monitored area Z is a submarine area, and the monitored acoustic phenomena are submarine phenomena. The acoustic sensors 101 are, for example, hydrophones.

[0027] The detection devices 10 are distributed throughout the monitored area Z according to a predefined configuration, depending on the signals to be measured: in the example considered, there are 4 detection devices, arranged relative to each other so that each is located on a respective vertex of the same tetrahedron. The sensors are at a known distance from each other and rigidly connected. The orientation of the entire assembly is also known, for example, by using an inertial measurement unit or other positioning system.

[0028] Each detection device 10 includes in the first embodiment considered and as shown in Figure 2, in particular an acoustic sensor 101, a processing block 102, a control block 104, a memory 105, a power supply block 106 and a telecommunications block 107 (the sensor 101 and the processing block 102 are generally in the same place, integrated in the same hardware element); in the first embodiment considered and as shown in Figure 2, it also includes a recording block 103.

[0029] The power supply unit 106, including for example batteries or cells, is suitable for supplying electrical power to the acoustic sensor 101, the processing unit 102, the recording unit 103, the control unit 104, the memory 105, the power supply unit 106 and the telecommunications unit 107.

[0030] The telecommunication block 107 of a detection device 10 comprises a telecommunication transmitter and receiver and is adapted to implement telecommunication links, in particular with the telecommunication block 107 of each other detection device 10. These telecommunication links are, for example, wireless links (e.g., UHF radio) or wired links on wired telecommunication media connecting the devices 10. In one embodiment, the telecommunication block 107 of at least one of the detection devices 10 is further adapted to communicate with a remote electronic monitoring module (not shown), for example via satellite link or other.

[0031] Each detection device 10 is adapted to work alternately according to a first mode, corresponding to detection processing requiring very low consumption, called very low consumption detection mode and a second mode corresponding to the implementation, by the system 1, of detection processing requiring a higher electrical energy consumption than the very low consumption detection mode, called deep detection mode and described below.

[0032] For illustrative purposes only, the power consumption of the detection device 10 in very low power detection mode corresponds to a power consumption of the device less than or equal to 100 pW (microwatts), this mode being continuously active so as to be constantly "listening" for relevant information in the signal. The instantaneous power consumption for detection processing on a signal obtained by the device in the second mode (i.e., in the more focused phases of processing and confirming detections) is at least 100 times higher and is in the range of 10 mW to 1 W.

[0033] The recording block 103 is adapted to record - only in the second mode (deep detection mode) and not in very low power detection mode - the electrical signals at the output of the sensor 101 in the memory 105. In one embodiment, it is adapted to also record - only in deep detection mode - at least some of the results of the processing block 102 in the memory 105.

[0034] The detection devices 10 are adapted to implement the steps, which are their respective responsibilities, of the process 200 now described with reference to Figure 3.

[0035] In step 201, each detection device 10 analyzes, by default, continuously or periodically with a period T shorter than the duration of the phenomena to be detected, its acoustic environment in very low power detection mode, and this continues until it detects a significant acoustic event or receives an alert from the N-1 other detection devices 10. Thus: - the sensor 101 captures ambient acoustic signals and delivers, to the processing block 102, an electrical signal based on these acoustic signals; - the processing block 102 performs electrical signal processing to detect if one (or more) acoustic phenomenon(ies) of interest is present in the captured acoustic signals.

[0036] The acoustic phenomena of interest are predefined and application-dependent. For example, in the case of monitoring marine wildlife, the acoustic phenomena of interest will be the characteristic sounds of the animals being monitored (whales, dolphins, etc.). In the case of monitoring marine vessels, they will be the sound signatures of different types of ships and underwater devices, and so on. The data defining these reference acoustic phenomena of interest have been previously stored in memory 105 and are used in step 101 by processing block 102.

[0037] When a detection device 10 is in this very low power detection mode, it is impossible to record signals from the acoustic sensor 101 (for example, the control block 104 prohibits the use of the recording block 103 of the device 10).

[0038] In one embodiment, the signal processing performed by the processing block 102 in this very low-power detection mode is, for example, a low-complexity detection process, thus consuming little electrical energy. For example, it is a processing method such as an analog filter bank coupled with expert rules or a simple neural network.

[0039] As soon as the processing block 102 of a detection device 10 (named device 10) a(ierte) detects an acoustic phenomenon of interest during this processing, it notifies the control block 104 which then triggers the sending, via the telecommunication block 107, of an alert message to the other detection devices 10 (named 10 _ autre ).

[0040] In step 202, as soon as a detection device 10 receives, via its communication block 107, an alert message sent by one of the (N-1) other detection devices 10 of system 1, this detection device 10 switches from the very low power detection mode to the second mode (deep detection mode). The switchover is triggered, for example, by its control block 104. The detection processing corresponding to the first mode is stopped.

[0041] In this deep detection mode, within the detection device 10 that received the alert, the recording block 103 then records the electrical signal from the sensor and the processing block 102 performs at least one processing on all or part of the recorded electrical signal: this processing involves an analysis of the recorded signal (post-analysis) using one or more complex algorithms for searching and identifying acoustic phenomena consuming more electrical energy than the processing carried out in the very low power consumption detection mode: for example, pattern analysis type algorithms on FFT, which are implemented based on convolutional neural networks or others (classical networks, Long Short Term Memories, expert rules...), and have an energy cost too high to operate continuously when 'embedded'.

[0042] In one embodiment, as soon as a detection device 10 aierte has detected a phenomenon, in addition to sending an alert message, it also switches to the second mode, here the deep detection mode, in addition to the other detection devices.

[0043] At the end of step 202, the processing block 104 of one or more of the N-1 other devices in turn identifies the acoustic phenomenon already detected by the device that triggered the alert, and confirms it as a significant phenomenon or, conversely, concludes that it is not significant and, if confirmed, characterizes it in greater detail than the latter did via its analysis in very low power mode. During this processing, reference definition data for significant sound phenomena (for the application detection data used) previously stored in memory 105, are used for example when comparing with data extracted from the recorded signal.

[0044] Then, if the detection of a significant acoustic phenomenon is confirmed, for example: - the signature determined by the analysis in deep detection mode is stored in memory 105; and / or - a transmission, to the supervision module, of the alert accompanied by at least some of the results of the in-depth detection is carried out.

[0045] In one embodiment of electrical signal recording, the data defining this electrical signal are recorded synchronously between the N-1 detection devices: they are time-stamped (the reference time being the same for all detection devices) and stored with their time stamp. To achieve this, the N detection devices have been synchronized in a prior step, or the alert issued in step 201 (or any other agreed signal transmitted between them) serves as the synchronization cue. Synchronization calibration may have taken place beforehand, for example, in the case of radio alerts.

[0046] Following the characterization of the detected phenomenon by each detection device 10 (or when no further acoustic phenomena are detected by that detection device), the switch from in-depth detection mode to very low-power detection mode is triggered by the control block 104 of that detection device. The N detection devices then each repeat step 201.

[0047] In step 203 (optional), the signal recorded by one or more (for example, at least three) of the detection devices and relating to the same detected acoustic phenomenon is used, after synchronization, to measure the time delays between the arrival times of the signals at each of these devices, and to deduce the direction of arrival of the signal (the relative positions of the devices being known). These calculations are performed, for example, by the control block 104 of one of the detection devices or by a processor of the supervisory module, after collection of the electrical signals relating to the acoustic phenomenon.

[0048] In a second embodiment, the second mode includes a recording-only mode, in which the recording block 103 of the detection device 10 records signals from the acoustic sensor 101 of the device. In step 202, the detection device 10 switches to this recording mode following reception of the alert message, the in-depth detection signal processing being carried out later, in delayed time, on the recorded signals (by the processing block 102 of the detection device or otherwise as described below), for example after recovery of the detection devices.

[0049] In a third embodiment, the detection devices of the monitoring system 1 (or at least some of them) are, compared to their structure shown in Figure 2, only equipped with very low-power detection and alerting capabilities enabling them to perform step 201, but do not have a recording block 103, nor in-depth detection capabilities. Instead, a recording and in-depth detection device for the monitoring system 1, equipped with memory and a processor, is used, which performs the recording and in-depth detection tasks described in the first embodiment as performed in step 202 by the recording block 103 and the processing block 102. The recording and in-depth detection device is connected to each of these detection devices (via its communication block 107) by a wireless or wired telecommunications link.For example, it is located at the center of the tetrahedron whose corners correspond to the locations of the devices, or it is integrated into one of the detection devices 10.

[0050] In this third embodiment, after step 201, in a step 202_bis (replacing step 202 described above), as soon as a detection device (without recording block 103, nor deep detection capabilities), named 10_bis, receives, via its communication block 107, an alert message sent by one of the (N-1) other detection devices of system 1, this detection device 10_bis switches from the very low power detection mode to the second mode: in this second mode, it then transmits the electrical signal from its sensor to the recording and deep detection device; the latter then records the received signal and it will perform at least one signal processing described above as performed for deep detection (in a variant, this device records in its memory the signals which are transmitted to it, the deep detection processing being performed later, after retrieval of the memory).

[0051] In one variant, the recording block is kept in all detection devices and only the "deep detection" part is centralized.

[0052] The transitions between the first and second modes correspond, regardless of the embodiments, to a switch in the detection structure; the peripheral detection devices may or may not be equipped with advanced recording and / or analysis capabilities.

[0053] In one embodiment, the shape of the network is determined according to the intended application.

[0054] For example, if it is a question of detecting a source of acoustic phenomenon which can be placed anywhere in the monitored area Z, a network of four detection devices 10 will be used, the four corresponding sensors being arranged for example in a tetrahedral structure (not necessarily regular, cf. submarine instrumentation for example).

[0055] The structure can be anything other than tetrahedral, for example planar for analyzing sources that may come from a 2D surface.

[0056] In the case where detection takes place in a linear waveguide (pipeline, for example), only two sensors, each in its respective detection device 10, are required. The distance between the sensors depends on the characteristic time of the signature (or signatures) of interest being sought and the medium through which the sound propagates.

[0057] Regardless of the network shape, it is advisable to choose the distance between the sensors so that the signal can be detected by a detection device in very low power mode and the corresponding alert can be received by at least one of the other devices before the signal reaches that other device, which will allow the acoustic signature recorded in its entirety to be available for advanced analysis later on the signals recorded following the switch to the second mode.

[0058] The invention therefore describes a solution for detecting acoustic phenomena that limits the consumption of electrical energy, based on a switch from a first operating mode corresponding to very low consumption detection to a second operating mode associated at least with triggering recording for in-depth detection, requiring much more energy, and conversely, a switch, associated with a recording prohibition, from the second mode to the very low consumption detection mode.

[0059] Each detection device continuously analyzes the received acoustic signals in very low-power detection mode so that it can wake up the entire network if an abnormal event occurs. Since the "wake-up" occurs almost instantaneously via wireless or wired telecommunication links, the other N-1 detection devices can then be switched to deep detection mode. Implemented by system 1 even before the acoustic signal reaches them, this allows for complete capture of the signal in question, whereas the detection device that first detected the signal could not have recorded it (this would have required it to be perpetually recording, which is not feasible in terms of energy and storage). The captures are then analyzed afterward by the embedded system using advanced algorithms (e.g., CNNs) whose energy cost is too high to run continuously, in order to confirm the alert. Thus, the invention, based on the coupling between the sensors and the propagation of the wave from one sensor to another, makes it possible to perform in-depth analyses of the signal with ultra-low energy consumption.

[0060] The invention thus proposes a network of high-resolution acoustic sensors (with high temporal and spectral resolution, capable of sampling frequencies on the order of 500 kHz and a spectral sensitivity of 1 Hz to 250 kHz) that enables high-quality (24-bit with a high signal-to-noise ratio) multi-channel recording of events, even single and very short ones, in order to identify them and determine the orientation of their emission source with high precision. Furthermore, always-on detection is performed at ultra-low power within the solution.

[0061] One embodiment of underwater monitoring has been described above, but the invention can be implemented for monitoring acoustic phenomena in media other than water: in air, vacuum, gas, or a waveguide medium in general. This could be, for example, metal, as seen in pipelines. In the case of one-dimensional phenomena, two sensors are sufficient: the first triggers the second, which can capture the shock or leak signal as soon as it appears and in its entirety, thus facilitating and improving classification. For example, the invention is implemented to monitor factories where machine defects can occur; in this case, detection is performed in 3D.

[0062] At least some of the steps (e.g., those performed by the processing block and / or the control block and / or the recording block) can be implemented by executing software instructions on a processor in the detection device 10 (or in the recording and deep detection device), with the software instructions stored in the device's memory. Alternatively, they can be implemented by dedicated hardware, typically a digital integrated circuit, either specific (ASIC) or based on programmable logic (e.g., FPGA / Field Programmable Gate Array).

Claims

CLAIMS 1. Method for detecting acoustic phenomena using a detection system (1) comprising N detection devices (10) distributed in an area to be monitored (Z), N >2 and connected to each other by telecommunication links (11), each detection device (10) comprising an acoustic sensor (101) capturing sounds and delivering an electrical signal as a function of the captured sounds and an electrical signal processing block (102), each detection device (10) is adapted to operate alternately according to: - a first operating mode during which the detection device (10) carries out a first analysis to detect acoustic phenomena, corresponding to a first level of electrical consumption of the device, called very low consumption detection mode; and - a second operating mode during which processing of the sounds captured by the detection device is carried out, corresponding to a second level of electrical consumption higher than the first level; according to which: each detection device (10) operates by default in the first mode, in which the processing block (102) carries out the first analysis on the electrical signal delivered by the acoustic sensor, and in which the recording of said electrical signal is not authorized; and as soon as said first analysis detects a predefined acoustic phenomenon, said detection device (10) alerts the N-1 other detection devices (10) of the system via the telecommunications links (11); upon receipt of the alert, each of the N-1 other detection devices (10) then switches to the second mode, in which it is triggered: - the transmission, to a recording block, of the electrical signal delivered by its acoustic sensor after receipt of the alert, - the recording, by said recording block, of the transmitted electrical signal and - a second analysis of the recorded electrical signal, to detect acoustic phenomena, implementing algorithmic processing corresponding to a second level of electrical consumption higher than the first level.

2. Method for detecting acoustic phenomena according to claim 1, comprising one of the following provisions: - when the detection device is in said second mode: the recording is carried out by an electronic recording block of said detection device and / or the second analysis is carried out by the processing block of said detection device; - when the detection device is in said second mode, it transmits the electrical signal to a remote analysis device via a teletransmission link and the second analysis is carried out by said remote analysis device.

3. Method for detecting acoustic phenomena according to claim 1 or 2, according to which, in the second mode, the electrical signals of the N-1 devices are recorded in a synchronized manner, the detection devices having been previously synchronized with each other and / or a synchronization being implemented by said detection devices on the basis of the alert or another signal.

4. Method for detecting acoustic phenomena according to any one of the preceding claims, according to which the direction of arrival of an acoustic phenomenon detected by the second analysis in the second mode is determined as a function of the recordings of the electrical signals delivered by the sensors of several of the N-1 detection devices following receipt of the alert.

5. Method for detecting acoustic phenomena according to any one of the preceding claims, according to which the relative positions of the sensors between them are determined as a function of at least one temporal length of acoustic signatures sought and of the environment in which the detection devices are arranged, in which the sound propagates.

6. Method for detecting acoustic phenomena according to any one of the preceding claims, according to which the inter-sensor distance is chosen so that a sound signal can be detected by a detection device in very low consumption mode among the N detection devices before said sound signal arrives at least at the sensor of one of the N-1 other detection devices.

7. System for detecting acoustic phenomena (1) comprising N detection devices (10) adapted to be distributed in an area to be monitored (Z), N >2 and to be connected to each other by telecommunication links (11), each detection device (10) comprising an acoustic sensor (101), adapted to capture sounds and deliver an electrical signal according to the captured sounds, and an electrical signal processing block (102), each detection device (10) is adapted to operate alternately according to: - a first operating mode during which the detection device (10) carries out a first analysis to detect acoustic phenomena, corresponding to a first level of electrical consumption of the device, called very low consumption detection mode; and - a second operating mode during which processing of the sounds captured by the detection device is carried out, corresponding to a second level of electrical consumption higher than the first level; in which: each detection device (10) is adapted to operate by default in the first mode, in which the processing block (102) is adapted to carry out the first analysis on the electrical signal delivered by the acoustic sensor, and in which the recording of said electrical signal is not authorized; and as soon as said first analysis detects a predefined acoustic phenomenon, said detection device (10) is adapted to alert the N-1 other detection devices (10) of the system via the telecommunication links (11); upon receipt of the alert, each of the N-1 other detection devices (10) is adapted to then switch to the second mode, in which the following operations are triggered: - the transmission, to a recording block, of the electrical signal delivered by its acoustic sensor after receipt of the alert, - the recording, by said recording block, of the transmitted electrical signal and - a second analysis of the recorded electrical signal, to detect acoustic phenomena, implementing algorithmic processing corresponding to a second level of electrical consumption higher than the first level.

8. System for detecting acoustic phenomena (1) according to claim 7, comprising one of the following provisions: when the detection device is in said second mode: the recording is carried out by an electronic recording block of said detection device and / or the second analysis is carried out by the processing block of said detection device; when the detection device is in said second mode, it is adapted to transmit the electrical signal to a remote analysis device via a remote transmission link and the second analysis is carried out by said remote analysis device.

9. Acoustic phenomena detection system (1) according to claim 7 or 8, wherein, in the second mode, the electrical signals of the N-1 devices are recorded in a synchronized manner, the detection devices having been previously synchronized with each other and / or a synchronization being implemented by said detection devices on the basis of the alert or another signal.

10. Acoustic phenomenon detection system (1) according to any one of claims 7 to 9, adapted to determine the direction of arrival of an acoustic phenomenon detected by the second analysis in the second mode as a function of the recordings of the electrical signals delivered by the sensors of several of the N-1 detection devices following receipt of the alert.

Images