Abnormal state detection system, abnormal state detection method, and abnormal state detection program

Abstract

To provide an abnormal state detection system, an abnormal state detection method, and an abnormal state detection program capable of detecting the abnormal state of a vehicle without preparing scream data in advance.SOLUTION: An abnormal state detection system comprises: a voice abnormality detection unit that includes a model generation unit for generating a model of voice data of a person in a vehicle under normal circumstances as a normal-state model from voice data including the voice of the person, and detects abnormalities in the voice based on the current voice data and the normal-state model of the person; a vehicle body abnormality detection unit that detects abnormalities in the body of the vehicle based on acceleration data of the vehicle; and an abnormal state determination unit that determines whether or not the vehicle is in an abnormal state based on the detection result of the voice abnormality detection unit and the detection result of the vehicle body abnormality detection unit.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present disclosure relates to an abnormal condition detection system, an abnormal condition detection method, and an abnormal condition detection program. [Background technology]

[0002] Generally, a technology for detecting an abnormal state of a vehicle such as an accident is known. Another known technology is a technology for detecting the occurrence of an accident by detecting the scream of a person in the vehicle. For example, Patent Document 1 describes a technology in which a model of a scream, which is a model learned from features of a plurality of types of screams, is prepared in advance, and whether an input voice is a scream is determined based on features extracted from an input voice, an acoustic model set, and a likelihood. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2017-187676 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, when preparing data on screams in advance as in the technology described in Patent Document 1, it is necessary to prepare data on various screams, which results in a large amount of data. Also, screams of patterns that are not prepared in advance cannot be recognized as screams. Therefore, the conventional technology is not sufficient to detect abnormal conditions of the vehicle.

[0005] The present disclosure aims to provide an abnormal condition detection system, an abnormal condition detection method, and an abnormal condition detection program that can detect an abnormal condition in a vehicle without preparing scream data in advance. [Means for solving the problem]

[0006] In order to achieve the above-mentioned objective, the abnormal condition detection system disclosed herein includes a model generation unit that generates a model of a person's normal voice data as a normal state model from voice data including the voice of the person riding in a vehicle, a voice anomaly detection unit that detects voice abnormalities based on the person's current voice data and the normal state model, a body anomaly detection unit that detects abnormalities in the body of the vehicle based on acceleration data of the vehicle, and an abnormal condition determination unit that determines whether or not the vehicle is in an abnormal state based on the detection result of the voice anomaly detection unit and the detection result of the body anomaly detection unit.

[0007] In addition, in order to achieve the above-mentioned objective, the abnormal condition detection method disclosed herein includes a model generation process in which a processor provided in the abnormal condition detection system generates a model of the normal voice data of a person riding in a vehicle from voice data including the voice of the person as a normal state model, detects a voice abnormality based on the person's current voice data and the normal state model, detects a body abnormality of the vehicle based on acceleration data of the vehicle, and outputs a determination result as to whether or not the vehicle is in an abnormal state based on the detection result of the voice abnormality and the detection result of the body abnormality.

[0008] In addition, in order to achieve the above-mentioned object, the abnormal condition detection program of the present disclosure is configured to cause a processor provided in an abnormal condition detection system to execute a process that includes a model generation process that generates a model of a person's normal voice data from voice data including the voice of the person riding in the vehicle as a normal condition model, detects a voice abnormality based on the person's current voice data and the normal condition model, detects a body abnormality of the vehicle based on acceleration data of the vehicle, and outputs a determination result as to whether or not the vehicle is in an abnormal condition based on the detection result of the voice abnormality and the detection result of the body abnormality. Effect of the Invention

[0009] According to the present disclosure, an abnormal state of a vehicle can be detected without preparing scream data in advance. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram for explaining an overview of an abnormal condition detection system. [Diagram 2] 1 is a circuit diagram illustrating an example of a hardware configuration of an abnormal condition detection system. [Diagram 3] FIG. 1 is a configuration diagram illustrating an example of a configuration of an abnormal condition detection system. [Figure 4] FIG. 2 is a diagram for explaining the function of the abnormal condition detection system. [Diagram 5] FIG. 13 is a diagram for explaining a data frequency / data accumulation measurement unit. [Figure 6] 10 is a flowchart showing an example of the flow of an abnormal state detection process. [Figure 7] 13 is a flowchart illustrating an example of the flow of an audio anomaly detection process. [Figure 8] 10 is a flowchart showing an example of the flow of a vehicle body abnormality detection process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. Note that the technology of the present disclosure is not limited to the following embodiments.

[0012] As shown in Fig. 1, an abnormal condition detection system 1 of this embodiment is provided inside a vehicle body 3 of a vehicle 2. The abnormal condition detection system 1 is a system that determines whether or not the vehicle 2 is in an abnormal condition based on an abnormality in the voice of a person riding in the vehicle 2 and an abnormality in the vehicle body 3. Note that in this embodiment, "persons riding in the vehicle 2" specifically refers to all people present inside the vehicle body 3, and includes people other than the driver.

[0013] An example of the hardware configuration of the abnormal condition detection system 1 of this embodiment is shown in Fig. 2. As shown in Fig. 2, the abnormal condition detection system 1 of this embodiment includes an acceleration sensor 40, a control unit 50, a storage unit 52, a sound collector 54, a sound recording device 56, and a communication I / F (Interface) 60. The acceleration sensor 40, the control unit 50, the storage unit 52, the sound collector 54, the sound recording device 56, and the communication I / F 60 are connected via a bus 69 such as a system bus or a control bus so as to be able to exchange various information with each other.

[0014] The control unit 50 includes a CPU (Central Processing Unit) 51, and controls the entire abnormal condition detection system 1. The CPU 51 of the present embodiment is an example of a processor of the present disclosure.

[0015] The storage unit 52 stores the voices of people riding in the vehicle 2, a current model generation unit 28 and a display area 62 (see FIG. 3 ), which will be described in detail later, etc. The storage unit 52 is realized by a storage medium such as a hard disk drive (HDD), a solid state drive (SSD), or a flash memory.

[0016] The sound collector 54 has a function of collecting the voices of people riding in the vehicle 2, and is, for example, a microphone. The recording device 56 has a function of recording the voices collected by the sound collector 54, and the recorded voices are stored in the memory unit 52. In addition, the acceleration sensor 40 has a function of detecting the acceleration of the body 3 of the vehicle 2.

[0017] The communication I / F 60 communicates various types of information with devices outside the abnormal condition detection system 1 and devices outside the vehicle 2 via wireless communication or wired communication.

[0018] The configuration of the abnormal condition detection system 1 of this embodiment will be described further with reference to Fig. 3. As shown in Fig. 3, the abnormal condition detection system 1 of this embodiment includes a voice abnormality detection unit 10, a vehicle body abnormality detection unit 12, and an abnormal condition determination unit 14.

[0019] The voice anomaly detection unit 10 includes a voice acquisition unit 20, a voice extraction unit 22, a spectrum creation unit 24, a data frequency / data accumulation measurement unit 26, a current model generation unit 28, a multiplexer 30, a normal state model generation unit 32, a difference detection unit 34, and a multiplexer 36.

[0020] The voice acquisition unit 20 includes the above-mentioned sound collector 54 and recording device 56. As shown in FIG. 4, the voice acquisition unit 20 has a function of acquiring voice data including the voice of the person 6 riding in the vehicle 2. In reality, the sounds collected by the sound collector 54 of the voice acquisition unit 20 also include sounds other than the voice emitted by the person 6 riding in the vehicle 2. Examples of sounds other than the voice emitted by the person 6 include sounds emitted by the vehicle 2 itself and environmental sounds. In this embodiment, sounds collected by the sound collector 54 of the voice acquisition unit 20 other than the voice of the person 6 are referred to as "noise".

[0021] The voice extraction unit 22 has a function of extracting voice data of the person 6 riding in the vehicle 2 from the voice data acquired by the voice acquisition unit 20. In other words, the voice extraction unit 22 has a function of removing noise from the voice data acquired by the voice acquisition unit 20. The method of extracting the voice of the person 6, i.e., human voice, from the voice data acquired by the voice extraction unit 22 is not particularly limited, and any known method can be used. For example, a filter that passes only the frequency of the human voice may be applied.

[0022] The spectrum creation unit 24 has a function of creating a waveform histogram for the voice data output from the voice extraction unit 22. The method by which the spectrum creation unit 24 creates the waveform histogram is not particularly limited, and any known method can be used. For example, the waveform histogram may be created by performing FFT (Fast Fourier Transform) and DFT (Discrete Fourier Transform) on the voice data. This allows the frequency characteristics of the voice data to be understood, and the characteristics of the voice data become clear. As shown in FIG. 4, the characteristics of the waveform histogram of voice data in normal times and the waveform histogram of an emergency such as an accident are different. The frequency of the voice data corresponds to the pitch of the voice, and the amplitude corresponds to the volume of the voice. As shown in FIG. 4, the waveform histogram of normal times includes many waveforms with relatively small amplitudes and relatively low frequencies. On the other hand, a scream tends to be louder and higher than a normal voice. Therefore, an emergency includes waveforms with large amplitudes and high frequencies. In this way, when the waveform histogram contains features that deviate significantly from the features contained in the waveform histogram in normal times, it can be determined that the voice data contains a scream.

[0023] The data frequency / data accumulation level measurement unit 26 has a function of measuring the frequency and accumulation level of voice data. The data frequency / data accumulation level measurement unit 26 measures the frequency and accumulation level of voice data to determine whether the voice data is correct and sufficient. The data frequency / data accumulation level measurement unit 26 of this embodiment divides a section into a predetermined time interval, measures the frequency of data for each section, and adopts a section in which the frequency of data is higher than a predetermined threshold as the valid period of voice data. In the example shown in FIG. 5, the periods 70_1, 70_2, and 70_3 are valid periods of voice data because the data frequency is high. On the other hand, the period 72 is outside the valid period of voice data because the data frequency is low. In other words, the voice data of the period 72 is excluded from the creation of the normal state model. Note that a case where the data frequency is low is a case where the person 6 does not speak much. Therefore, when the data frequency is low, it is preferable that the abnormal state detection system 1 encourages the person 6 to speak.

[0024] The current model generation unit 28 has a function of generating a current voice data model (referred to as a "current model" in this embodiment) using current voice data. The method by which the current model generation unit 28 generates the current model is not particularly limited. For example, the current model may be generated as a voice feature classifier by performing machine learning using a waveform histogram of the voice data as input without using teacher data for the model, using a known method.

[0025] The multiplexer 30 has a function of receiving the current model generated by the current model generation unit 28 as an input, and outputting data in which voice data with a low data frequency is excluded as outliers from the voice data under normal conditions, based on the measurement results of the data frequency / data accumulation measurement unit 26.

[0026] The normal state model generating unit 32 has a function of generating a model of the normal state voice (normal state model) of the person 6. The method by which the normal state model generating unit 32 generates the normal state model is not particularly limited, and may be the same as the method by which the current model generating unit 28 generates the current model, for example.

[0027] The difference detection unit 34 has a function of sequentially comparing the current model generation unit 28 with the normal state model generation unit 32 and outputting the comparison result. As described above, the difference detection unit 34 compares the waveform histogram of the current model generation unit 28 with the waveform histogram of the normal state model generation unit 32, and when the difference exceeds a predetermined threshold, outputs information indicating that fact. In other words, when the difference detection unit 34 detects a scream, it outputs information indicating that fact.

[0028] When information indicating that a scream has been detected is input from the difference detection unit 34, the multiplexer 36 has a function of outputting a detection result of "abnormality present" to the abnormal state determination unit 14. On the other hand, when information indicating that a scream has been detected is not input from the difference detection unit 34, the multiplexer 36 has a function of outputting a detection result of "no abnormality" to the abnormal state determination unit 14. Specifically, the multiplexer 36 in this embodiment outputs "1" to the abnormal state determination unit 14 if "abnormality present" and "0" to the abnormal state determination unit 14 if "no abnormality" is detected.

[0029] In this manner, the audio anomaly detection unit 10 can output a detection result indicating that an anomaly has occurred when the person 6 screams.

[0030] On the other hand, the vehicle body abnormality detection unit 12 includes an acceleration sensor 40, a difference detection unit 42, and a multiplexer 44. The acceleration sensor 40 has a function of outputting acceleration data of the vehicle 2, as described above.

[0031] The difference detection unit 42 has a function of sequentially outputting the change in the acceleration data as a difference. The difference becomes large when the vehicle 2 has a collision, the vehicle suddenly brakes, etc. Therefore, when the change (difference) in the acceleration data exceeds a predetermined threshold, the difference detection unit 42 outputs information indicating that an abnormality in the vehicle body 3 has been detected.

[0032] The multiplexer 44 has a function of outputting a detection result indicating "abnormality present" to the abnormal state determination unit 14 when information indicating that an abnormality in the vehicle body 3 has been detected is input from the difference detection unit 42. On the other hand, when information indicating that an abnormality in the vehicle body 3 has not been input from the difference detection unit 42, the multiplexer 44 has a function of outputting a detection result indicating "no abnormality" to the abnormal state determination unit 14. Specifically, the multiplexer 44 in this embodiment outputs "1" to the abnormal state determination unit 14 when "abnormality present" and "0" when "no abnormality" is detected.

[0033] In this manner, when an abnormality occurs in the vehicle body 3 due to a collision of the vehicle 2 or the like, the vehicle body abnormality detection unit 12 can output a detection result indicating that an abnormality has occurred.

[0034] The abnormal state determination unit 14 has a function of determining whether or not the vehicle 2 is in an abnormal state based on the detection result of the voice abnormality detection unit 10 and the detection result of the vehicle body abnormality detection unit 12, and outputting the determination result. As an example, the abnormal state determination unit 14 of this embodiment is an OR circuit, and outputs the determination result that the vehicle 2 is in an abnormal state to the outside of the voice abnormality detection unit 10 when at least one of the detection result of the voice abnormality detection unit 10 indicating that there is an abnormality and the detection result of the vehicle body abnormality detection unit 12 indicating that there is an abnormality.

[0035] Next, the operation of the abnormal condition detection system 1 of this embodiment will be described with reference to the drawings. Fig. 6 shows a flowchart showing an example of the flow of the abnormal condition detection process executed by the CPU 51 of the control unit 50. In this embodiment, the abnormal condition detection process shown in Fig. 6 is executed by the CPU 51 executing an abnormal condition detection process program (not shown) stored in the control unit 50. As an example, in this embodiment, when the engine of the vehicle 2 is started, the abnormal condition detection process shown in Fig. 6 is executed.

[0036] 6, the CPU 51 starts a process of detecting an abnormality in an audio signal by the audio abnormality detection unit 10. Specifically, the CPU 51 starts an abnormal state detection process (see FIG. 7), the details of which will be described later.

[0037] In the next step S102, the CPU 51 starts a process of detecting an abnormality in the vehicle body 3 by the vehicle body abnormality detection unit 12. Specifically, the CPU 51 starts a vehicle body state detection process (see FIG. 8), the details of which will be described later.

[0038] In the next step S104, it is determined whether or not the vehicle body 3 is in an abnormal state. Depending on the detection results of steps S100 and S102, as described above, the detection result is output from the abnormal state determination unit 14. If the detection result output from the abnormal state determination unit 14 indicates that the vehicle body 3 is not in an abnormal state, the determination in step S104 is negative, and the process proceeds to step S108. On the other hand, if the detection result indicates that the vehicle body 3 is in an abnormal state, the determination in step S104 is positive, and the process proceeds to step S106.

[0039] In step S106, the CPU 51 outputs, via the communication I / F 60, information to the outside that the vehicle body 3 is in an abnormal state.

[0040] In the next step S108, the CPU 51 determines whether or not to end the abnormal condition detection process. As an example, in this embodiment, when the engine of the vehicle 2 is turned off, the abnormal condition detection process is ended. Therefore, until the engine of the vehicle 2 is turned off, the determination in step S108 is negative, the process returns to step S104, and the processes in steps S104 and S106 are repeated. On the other hand, when the engine of the vehicle 2 is turned off, the determination in step S108 is positive, and the abnormal condition detection process shown in FIG. 6 is ended.

[0041] Next, the above-mentioned audio abnormal state detection process will be described with reference to Fig. 7. Fig. 7 shows a flowchart illustrating an example of the flow of the audio abnormal state detection process.

[0042] In step S120 of FIG. 7, the CPU 11 causes the voice acquisition unit 20 to start acquiring voice data.

[0043] In the next step S122, the CPU 51 determines whether or not to generate a normal state model. In this embodiment, a normal state model is generated at a predetermined timing. Therefore, it is determined whether or not it is the predetermined timing. If a normal state model is not to be generated, the determination in step S122 is negative, and the process proceeds to step S126. On the other hand, if a normal state model is to be generated, the determination in step S122 is positive, and the process proceeds to step S124.

[0044] In step S124, the CPU 51 causes the normal model generating unit 32 to generate a normal model, as described above.

[0045] In the next step S126, the CPU 51 causes the current model generating unit 28 to generate a current model, as described above.

[0046] In the next step S128, the CPU 51 causes the difference detection section 34 to detect the difference between the normal model and the current model, as described above.

[0047] In the next step S130, the CPU 51 causes the detection result of the audio anomaly detection unit 10 to be output to the abnormal state determination unit 14.

[0048] In the next step S134, the CPU 51 determines whether or not to end the audio abnormality detection process. As described above, until the engine of the vehicle 2 is turned off, the determination in step S132 is negative, the process returns to step S122, and the processes in steps S122 to S130 are repeated. On the other hand, if the engine of the vehicle 2 is turned off, the determination in step S132 is positive, and the process proceeds to step S134.

[0049] In step S134, the CPU 51 ends the acquisition of voice data by the voice acquisition unit 20. When the process of step S134 ends, the voice anomaly detection process shown in FIG.

[0050] The above-mentioned vehicle body abnormal state detection processing will be further described with reference to Fig. 8. Fig. 8 shows a flowchart showing an example of the flow of the vehicle body abnormal state detection processing.

[0051] In step S140 in FIG. 8, the CPU 11 causes the acceleration sensor 40 to start acquiring acceleration data.

[0052] In the next step S142, the CPU 51 causes the difference detection unit 42 to detect the difference in acceleration, as described above.

[0053] In the next step S144, the CPU 51 causes the abnormal state determination unit 14 to output the detection result of the abnormal state determination unit 14.

[0054] In the next step S146, the CPU 51 determines whether or not to end the vehicle body abnormality detection process. As described above, until the engine of the vehicle 2 is turned off, the determination in step S144 is negative, the process returns to step S142, and the processes in steps S142 and S144 are repeated. On the other hand, if the engine of the vehicle 2 is turned off, the determination in step S146 is positive, and the process proceeds to step S148.

[0055] In step S148, the CPU 51 ends acquisition of acceleration data by the acceleration sensor 40. When the process of step S148 ends, the vehicle body abnormality detection process shown in FIG.

[0056] As described above, the abnormal condition detection system 1 of this embodiment includes a normal state model generation unit 32 that generates a model of normal state voice data of the person 6 riding in the vehicle 2 as a normal state model from voice data including the voice of the person 6, and includes a voice anomaly detection unit 10 that detects voice abnormalities based on the current voice data of the person 6 and the normal state model. The abnormal condition detection system 1 also includes a vehicle body anomaly detection unit 12 that detects an abnormality in the vehicle body 3 of the vehicle 2 based on acceleration data of the vehicle 2. Furthermore, the abnormal condition detection system 1 includes an abnormal state determination unit 14 that determines whether or not the vehicle 2 is in an abnormal state based on the detection result of the voice anomaly detection unit 10 and the detection result of the vehicle body anomaly detection unit 12.

[0057] According to the above configuration, a normal voice of the person 6 is acquired while riding in the vehicle 2, and a scream can be detected based on the normal voice. Therefore, according to the abnormal condition detection system 1 of this embodiment, an abnormal condition of the vehicle can be detected without preparing scream data in advance.

[0058] The following supplementary notes are further disclosed regarding the above embodiment. (Appendix 1) a model generation unit that generates a model of a person's normal voice data as a normal-time model from voice data including the voice of the person riding in the vehicle, and a voice anomaly detection unit that detects anomalies in the voice based on the person's current voice data and the normal-time model; a vehicle body abnormality detection unit that detects an abnormality in a vehicle body of the vehicle based on acceleration data of the vehicle; an abnormality determination unit that determines whether or not the vehicle is in an abnormal state based on a detection result of the voice abnormality detection unit and a detection result of the vehicle body abnormality detection unit; An abnormal condition detection system equipped with

[0059] (Appendix 2) The voice anomaly detection unit further includes a current model generation unit that generates a model of the current voice data of the person as a current model from the current voice data of the person, and determines that the voice is abnormal when it is determined that the current voice data of the person includes a scream based on a comparison result between the current model and the normal time model. 2. The abnormal condition detection system of claim 1.

[0060] (Appendix 3) The voice anomaly detection unit includes: a waveform histogram obtained from voice data in a normal state; If it is determined that the current voice data of the person includes a scream based on the comparison result with the waveform histogram obtained from the current voice data, the voice is determined to be abnormal. 3. The abnormal condition detection system of claim 1 or 2.

[0061] (Appendix 4) The abnormal state determination unit determines that the vehicle is in an abnormal state when at least one of a detection result of the voice abnormality detection unit that a voice is abnormal and a detection result of the vehicle body abnormality detection unit that a vehicle body is abnormal. 4. The abnormal condition detection system according to claim 1.

[0062] (Appendix 5) The processor in the abnormal condition detection system A model generation process is included in which a model of the normal voice data of a person riding in a vehicle is generated as a normal model from voice data including the voice of the person, and an abnormality in the voice is detected based on the current voice data of the person and the normal model; Detecting an abnormality in a body of the vehicle based on acceleration data of the vehicle; Based on the detection result of the abnormality in the voice and the detection result of the abnormality in the vehicle body, a determination result as to whether or not the vehicle is in an abnormal state is output. Method for detecting abnormal conditions.

[0063] (Appendix 6) The processor in the abnormal condition detection system A model generation process is included in which a model of the normal voice data of a person riding in a vehicle is generated as a normal model from voice data including the voice of the person, and an abnormality in the voice is detected based on the current voice data of the person and the normal model; Detecting an abnormality in a body of the vehicle based on acceleration data of the vehicle; Based on the detection result of the abnormality in the voice and the detection result of the abnormality in the vehicle body, a determination result as to whether or not the vehicle is in an abnormal state is output. An abnormal condition detection program to execute processing. [Explanation of symbols]

[0064] 1. Abnormal condition detection system 51 CPU 10. Audio anomaly detection unit 12 Vehicle body abnormality detection unit 14 Abnormal state determination section 28 Current Model Generation Unit 32 Normal state model generation unit

Claims

1. a model generation unit that generates a model of a person's normal voice data as a normal-time model from voice data including the voice of the person riding in the vehicle, and a voice anomaly detection unit that detects anomalies in the voice based on the person's current voice data and the normal-time model; a vehicle body abnormality detection unit that detects an abnormality in a vehicle body of the vehicle based on acceleration data of the vehicle; an abnormality determination unit that determines whether or not the vehicle is in an abnormal state based on a detection result of the voice abnormality detection unit and a detection result of the vehicle body abnormality detection unit; An abnormal condition detection system equipped with

2. The voice anomaly detection unit further includes a current model generation unit that generates a model of the current voice data of the person as a current model from the current voice data of the person, and determines that the voice is abnormal when it is determined that the current voice data of the person includes a scream based on a comparison result between the current model and the normal time model. The abnormal condition detection system of claim 1 .

3. The voice anomaly detection unit includes: a waveform histogram obtained from voice data in a normal state; If it is determined that the current voice data of the person includes a scream based on the comparison result with the waveform histogram obtained from the current voice data, the voice is determined to be abnormal. The abnormal condition detection system of claim 1 .

4. The abnormal state determination unit determines that the vehicle is in an abnormal state when at least one of a detection result of the voice abnormality detection unit that a voice is abnormal and a detection result of the vehicle body abnormality detection unit that a vehicle body is abnormal. The abnormal condition detection system of claim 1 .

5. The processor in the abnormal condition detection system A model generation process is included in which a model of the normal voice data of a person riding in a vehicle is generated as a normal model from voice data including the voice of the person, and an abnormality in the voice is detected based on the current voice data of the person and the normal model; Detecting an abnormality in a body of the vehicle based on acceleration data of the vehicle; Based on the detection result of the abnormality in the voice and the detection result of the abnormality in the vehicle body, a determination result as to whether or not the vehicle is in an abnormal state is output. Method for detecting abnormal conditions.

6. The processor in the abnormal condition detection system A model generation process is included in which a model of the normal voice data of a person riding in a vehicle is generated as a normal model from voice data including the voice of the person, and an abnormality in the voice is detected based on the current voice data of the person and the normal model; Detecting an abnormality in a body of the vehicle based on acceleration data of the vehicle; Based on the detection result of the abnormality in the voice and the detection result of the abnormality in the vehicle body, a determination result as to whether or not the vehicle is in an abnormal state is output. An abnormal condition detection program to execute processing.

Images