JP5937125B2 - Sound identification condition setting support apparatus and sound identification condition setting support method - Google Patents

Description

  The present invention relates to a sound identification condition setting support device and a sound identification condition setting support method that support the setting of an identification condition for identifying the presence or absence of an abnormal sound in a collected sound.

  For example, when considering measures to remove abnormal noise emitted from power transmission lines, or when diagnosing equipment malfunctions based on abnormal noise emitted from equipment, of the sounds collected near the power transmission line or equipment. It is determined by using a pattern recognition method whether or not there is an abnormal sound. Specifically, in many cases, the collected sound is subjected to spectrum analysis, and the presence or absence of the abnormal sound in the collected sound is identified based on the characteristics of the collected sound or the frequency of the abnormal sound (see Patent Document 1).

  However, if the frequency or intensity of the abnormal sound changes over time, it is difficult to accurately identify the presence or absence of the abnormal noise based solely on the static characteristics of the sampled sound or the abnormal noise frequency. There is. In such a case, it is desirable to analyze the spectrogram of the collected sound, and to identify the presence or absence of the abnormal sound in the collected sound in consideration of the temporal change in the frequency or intensity of the collected sound or the abnormal sound. Thereby, even when the frequency or intensity of the abnormal sound changes with time, the presence or absence of the abnormal sound can be identified with high accuracy.

JP-A-11-241945

  In order to identify the presence or absence of an abnormal sound in the collected sound by the pattern recognition method using the spectrogram of the collected sound, it is necessary to extract the characteristic of the abnormal sound expressed as the spectrogram. The superiority or inferiority of the feature extraction greatly affects the identification accuracy.

  There is no known general or unified method for extracting abnormal sound features. 2. Description of the Related Art Conventionally, an operator performs feature extraction of an abnormal sound by a method of displaying a spectrogram of a collected sound on a computer monitor and finding the characteristic of the spectrogram of the abnormal sound by visually observing the spectrogram.

  However, it takes a long time to find the characteristics of the abnormal sound that enables high-accuracy identification of the abnormal sound by looking at the spectrogram displayed on the computer monitor, which imposes a heavy burden on the operator. In addition, since it is often dependent on the experience of the operator whether or not the feature of the abnormal sound that enables high-precision identification of abnormal noise can be found, high-precision identification is performed when there is no skilled worker. It is difficult to realize.

  In addition, for example, when considering a method for removing abnormal noise emitted from a transmission line, it is difficult to predict when an abnormal noise will occur and what kind of abnormal noise will occur, If it occurs, measures must be taken in a short time to remove the noise. In such a scene, it is necessary to quickly and reliably perform the feature extraction of abnormal noise, but as described above, it is difficult to realize this by the conventional technology.

  The present invention has been made in view of, for example, the above-described problems, and the problem of the present invention is that it is possible to easily and reliably perform abnormal sound feature extraction or processing corresponding thereto in a spectrogram of collected sound. An object of the present invention is to provide a sound identification condition setting support device and a sound identification condition setting support method that can reduce the burden on an operator regarding processing and can realize highly accurate identification of the presence or absence of abnormal noise.

In order to solve the above-described problem, a sound identification condition setting support device according to the present invention is a sound identification condition setting support device that supports setting of an identification condition for identifying the presence or absence of an abnormal sound, and includes a plurality of sound identification condition setting support devices. The spectrogram data indicating the spectrograms of the abnormal sample sound and the non-noise sample sound that does not include the abnormal sound are received, the predetermined frequency range in each spectrogram data is divided into a plurality of blocks, A histogram generator for generating histogram data indicating a histogram of the intensity distribution of the sound; histogram data of each block for several different sampled sounds of the plurality of different sampled sounds; and Histogram of each block for some non-sound sample sounds of all sound samples A discriminant function generator for generating a discriminant function for discriminating the presence or absence of abnormal noise, and each block for some other abnormal sound samples of the plurality of abnormal sound samples Histogram data of each block and the histogram data of each block for some other non-sound sample sounds of the plurality of non-sound sample sounds are input to the discrimination function to identify the presence or absence of allophones A correct rate calculation unit for calculating the correct rate of discrimination, which is a probability that the discrimination function correctly identifies the presence or absence of abnormal noise, and setting of the number of blocks when dividing a predetermined frequency range in each spectrogram data into a plurality of blocks As a result of executing the processing by the histogram generation unit, the discrimination function generation unit and the accuracy rate calculation unit a plurality of times, and executing the processing a plurality of times, And said identification accuracy rate selecting the number of blocks set in generating the highest classification function and the identification function as the identification condition identification condition selecting section among the plurality of discriminant function generated by the function generator It is characterized by having.

A second sound identification condition setting support device according to the present invention is a spectrogram showing a spectrogram of the abnormal sound sample or the non-natural sound sample sound in the first sound identification condition setting support device of the present invention described above. A spectrogram generation unit for generating data is provided.

  In order to solve the above-described problem, a sound identification condition setting support method according to the present invention is a sound identification condition setting support method that supports setting of an identification condition for identifying the presence or absence of abnormal noise, and includes a plurality of sound identification condition setting support methods. For each of the spectrogram data indicating the spectrograms of the abnormal sample sound and the non-noise sample sound that does not include the abnormal sound, the predetermined frequency range in the spectrogram data is divided into a plurality of blocks, and A histogram generating step for generating histogram data indicating a histogram of a sound intensity distribution; histogram data of each block for several different sampled sounds of the plurality of different sampled sounds; Hist of each block for some non-sound sample sounds of all sound samples A discriminant function generating step for generating a discriminant function for discriminating the presence or absence of abnormal noise using the ram data, and each block for some other abnormal sound samples of the plurality of abnormal sound samples Histogram data of each block and the histogram data of each block for some other non-sound sample sounds of the plurality of non-sound sample sounds are input to the discrimination function to identify the presence or absence of allophones A correct answer rate calculating step of calculating a correct answer rate that is a probability that the discriminating function correctly identifies the presence or absence of abnormal noise, and the number of blocks when dividing a predetermined frequency range in each spectrogram data into a plurality of blocks The histogram generation step, the discriminant function generation step and the accuracy rate calculation step are executed a plurality of times while changing the setting of As a result, among the plurality of discriminant functions generated in the discriminant function generation step, the discriminant function having the highest discriminant accuracy rate and the number of blocks set when generating the discriminant function are selected as the discriminating conditions. It is characterized by.

  According to the present invention, it is possible to easily and reliably perform abnormal noise feature extraction or processing corresponding thereto in a spectrogram of collected sound, and reduce the burden on the operator regarding the processing.

It is a block diagram which shows the sound identification condition setting assistance apparatus by embodiment of this invention. It is a flowchart which shows the sound identification condition setting assistance process by the sound identification condition setting assistance apparatus by embodiment of this invention. It is a flowchart which shows the sound identification condition setting assistance process following FIG. It is a wave form diagram of a strange sound sample sound. It is a graph which shows the spectrogram of an unusual sound sample sound. It is a graph which shows the histogram of the intensity distribution of the sound in one block of the spectrogram of the unusual sound sample. It is a graph which shows the histogram of the intensity distribution of the sound in other one block of the spectrogram of an unusual sound sample sound. It is a graph which shows the histogram of the intensity distribution of the sound in another one block of the spectrogram of a strange sound sample sound. It is explanatory drawing which shows the state which divided | segmented the spectrogram of the sample sound into the block of 20 blocks. It is explanatory drawing which shows the state which divided | segmented the spectrogram of the sample sound into the block of the number of 30 blocks. It is explanatory drawing which shows the calculation method of the identification correct rate of an identification function. It is explanatory drawing which shows the discrimination | determination accuracy rate of the discrimination function corresponding to each setting of the number of blocks. It is a block diagram which shows the abnormal sound determination apparatus by embodiment of this invention. It is a flowchart which shows the noise determination process by the noise determination apparatus by embodiment of this invention.

  Embodiments of the present invention will be described below with reference to the drawings.

(Outline of processing by the sound identification condition setting support device)
FIG. 1 shows a sound identification condition setting support apparatus according to an embodiment of the present invention. In FIG. 1, a sound identification condition setting support apparatus 1 according to an embodiment of the present invention is an apparatus that supports the setting of an identification condition for identifying the presence or absence of abnormal sounds in a collected sound. The collected sound is, for example, a sound collected at a place near a power transmission line or equipment. The abnormal noise is, for example, a sound in which people living around the power transmission line complain of discomfort or a sound that does not occur when the device is normal.

  The process for supporting the setting of the identification condition performed by the sound identification condition setting support device 1 (sound identification condition setting support process) can be roughly described as an identification function that can identify the presence or absence of abnormal sounds in the collected sound. It is a process to generate or select. This process is substantially equivalent to a process for extracting features of abnormal sounds.

  The sound identification condition setting support device 1 generally performs the following process as the sound identification condition setting support process. First, as preparation for processing, an operator prepares audio data of abnormal sound samples and audio data of non-natural sound samples. The abnormal sound sample sound is a sample of the collected sound including the abnormal sound. An allophone sample sound is a sample of a sampled sound that does not contain an allophone. The audio data is digital data having an audio file format such as WAV, MP3 (MPEG Audio Layer-3), AAC (Advanced Audio Coding), and the like.

  For example, the operator prepares a plurality of voice data of abnormal sample sounds and a plurality of voice data of non-sound sample sounds by the following method. In other words, the operator collects a plurality of sounds by executing recording at the same place, reproduces the collected sounds sequentially, listens to them, and determines the presence or absence of abnormal sounds in each sound. Then, the operator classifies the audio data corresponding to the sound including the abnormal sound into the sound data of the abnormal sound sample sound, and the sound data corresponding to the sound not including the abnormal sound is sampled without the abnormal sound. Classify into sound data.

  Next, the operator inputs the sound data of the abnormal sample sound and the sound data of the non-sound sample sound to the sound identification condition setting support device 1.

  The sound identification condition setting support device 1 generates spectrogram data indicating a spectrogram of speech for each of the input speech data of a plurality of abnormal sample sounds and the sound data of a plurality of non-sound sample sounds.

  Subsequently, the sound identification condition setting support device 1 uses, for example, spectrogram data of a half-tone sample sound among a plurality of peculiar sound sample spectrogram data, and the spectrogram data of the phantom sample sound for generating a discrimination function. Choose as. Further, the sound identification condition setting support device 1 uses, for example, spectrogram data of half of the non-sound sample sound as spectrogram data of the non-sound sample sound for generating the discrimination function among the spectrogram data of the plurality of non-sound sample sounds. select.

  Subsequently, the sound identification condition setting support device 1 divides the predetermined frequency range for each of the selected spectrogram data into blocks having the number of blocks set according to a predetermined rule, and the sound intensity distribution of each block is determined. Histogram data indicating a histogram is generated. The predetermined frequency range is, for example, a range from about 20 Hz to about 20000 Hz, which is a general audible range. When the abnormal frequency range is predicted, the predetermined frequency range may be limited to a narrower range including the abnormal frequency range.

  Subsequently, the sound identification condition setting support device 1 has histograms in a plurality of spectrogram data respectively selected as spectrogram data of anomalous sample sound for discriminating function generation or spectrogram data of an allogeneic sample sound for discriminant function generation. An identification function is generated using the data, and the generated identification function and the number of blocks set at this time are stored in the storage unit 18.

  Subsequently, the sound identification condition setting support device 1 uses, for example, the spectrogram data of the remaining half-tone sample sounds among the spectrogram data of the plurality of different-tone sample sounds, for example, Select as spectrogram data. In addition, the sound identification condition setting support apparatus 1 uses, for example, the spectrogram data of the remaining half-sampled sound samples of the plurality of non-sampled sample sounds, and the spectrogram of the non-sampled sample sounds for discrimination function verification. Select as data. Subsequently, the sound identification condition setting support device 1 divides the predetermined frequency range for each of the selected spectrogram data into blocks having the number of blocks set according to a predetermined rule, and the sound intensity distribution of each block is determined. Histogram data indicating a histogram is generated.

  Subsequently, the sound identification condition setting support device 1 performs histograms in a plurality of spectrogram data respectively selected as spectrogram data of the abnormal sound sample for discrimination function verification or spectrogram data of the non-sound sample sound for verification of the discrimination function. Data is input to the generated discriminant function, and each time one piece of histogram data is input, discriminating by the discriminant function is executed to determine whether the discrimination result is correct or not. As a result, the number of determination results corresponding to the number of histogram data input to the discrimination function is obtained. Subsequently, the sound identification condition setting support device 1 calculates the identification accuracy rate of the identification function using the obtained plurality of determination results, and stores the calculated identification accuracy rate in the storage unit 18. The discrimination correct answer rate is the probability that the discrimination function correctly discriminates the presence or absence of abnormal sounds.

  Subsequently, the sound identification condition setting support device 1 performs the above-described generation of the identification function and calculation of the identification accuracy rate a plurality of times while changing the setting of the number of blocks according to a predetermined rule. Accordingly, the storage unit 18 of the sound identification condition setting support device 1 stores a plurality of identification functions generated using the plurality of blocks, histogram data generated under the setting of the number of blocks, and these identification functions. Is stored.

  Subsequently, the sound identification condition setting support device 1 is based on the identification accuracy rate stored in the storage unit 18 and the identification function having the highest identification accuracy rate among the plurality of identification functions stored in the storage unit 18 and corresponding to it. The number of blocks to be selected is selected, and these are stored in the storage unit 18 as a definite identification function and a definite block number. The determined identification function and the number of determined blocks selected and stored by the sound identification condition setting support device 1 in this way are used when the abnormal sound determination device 21 (see FIG. 13) determines the presence or absence of abnormal noise in the collected sound. Used for.

  Here, the spectrogram of sound and the histogram of intensity distribution will be described with reference to FIGS. FIG. 4 shows a specific example of the waveform of the abnormal sound sample. In FIG. 4, the horizontal axis represents time, and the vertical axis represents sound intensity. FIG. 5 shows a spectrogram of the abnormal sample sound shown in FIG. In FIG. 5, the horizontal axis is time, the vertical axis is frequency, and the color indicates the intensity of sound. In FIG. 5, the sound intensity is represented in two stages of black and white, and black indicates that the sound intensity is larger than white, but the actual spectrogram graph is The intensity of sound is represented by a difference in color (for example, lightness of color) in multiple levels (for example, 256 levels).

  6 shows a histogram of sound intensity distribution in block F1 in FIG. 5, FIG. 7 shows a histogram of sound intensity distribution in block F2 in FIG. 5, and FIG. 8 shows block F3 in FIG. The histogram of the sound intensity distribution in is shown. In each of FIG. 6 to FIG. 8, the vertical axis represents frequency / total frequency, and the horizontal axis represents sound intensity (intensity). Here, the noise sample sound and the non-noise sample sound have the same length (time from the start time to the end time), and the length of the sound sample sound or the sound sample sound without noise is the same. A value obtained by dividing (for example, 1 hour) by a predetermined unit time (for example, 1 second) corresponds to “total frequency”. In addition, it is obtained by counting the number of times that the sound intensity is the same in the predetermined unit time from the start point to the end point of the abnormal sound sample or non-noise sample sound. The obtained value corresponds to “frequency”.

  In the spectrogram of the abnormal sound sample shown in FIG. 5, in the block F1, the state where the sound intensity is high is continuous from the start time to the end time of the abnormal sound sample. In the sound intensity distribution in the block F1, as shown in FIG. 6, the sound intensity at which the frequency / total frequency has a peak is P1, the distribution is concentrated on P1, and the distribution range is narrow.

  Further, in the spectrogram of the abnormal sound sample shown in FIG. 5, in the block F2, the state where the sound intensity is low is continuous from the start time to the end time of the abnormal sound sample. In the sound intensity distribution in the block F2, as shown in FIG. 7, the sound intensity at which the frequency / total frequency reaches its peak is P2, and P2 is a value smaller than P1. Further, in the sound intensity distribution in the block F2, the distribution is concentrated on P2, and the distribution range is narrow.

  Further, in the spectrogram of the noise sample sound shown in FIG. 5, in the block F3, the sound intensity is high and the sound intensity is low between the start time and the end time of the noise sample sound. Are mixed. In the sound intensity distribution in the block F3, as shown in FIG. 8, the range of the distribution is relatively wide.

  As described above, according to the histogram of the intensity distribution for each block in the spectrogram of the abnormal sound sample, the temporal change in the sound intensity for each block of the abnormal sound sample can be expressed. Therefore, by using a histogram of the sound intensity distribution for each block in the spectrogram for each sampled sound such as abnormal sound sample, non-sound sample sound, judgment sound, abnormal sound contained in the sampled sound and the collected sound It is possible to generate a discriminant function capable of accurately discriminating the presence or absence of abnormal sounds in the collected sound while taking into account temporal changes in the frequency or sound intensity.

(Configuration of sound identification condition setting support device)
As shown in FIG. 1, the sound identification condition setting support device 1 includes an arithmetic processing unit 11. The arithmetic processing unit 11 divides the spectrogram data into a plurality of blocks, and a spectrogram generation unit 12 that generates spectrogram data from speech data. A histogram generating unit 13 that generates histogram data of the sound intensity distribution in each block, a discrimination function generating unit 14 that generates a discrimination function using the histogram data of the sound intensity distribution in each block, and discrimination of the discrimination function A correct rate calculation unit 15 that calculates the correct rate, an identification function selection unit 16 that selects an identification function with the highest identification accuracy rate and the number of blocks corresponding thereto, and a predetermined frequency range of spectrogram data is divided into a plurality of blocks. Set or change the number of blocks (ie the number of divisions in a given frequency range) It comprises processing and sound identification condition setting support processing and control unit 17 for control of flow. Furthermore, the sound identification condition setting support device 1 includes a storage unit 18 that stores a definite identification function, the number of definite blocks, and various data that should be held during the sound identification condition setting support process, and a sampled noise or non-uniform sound. An input / output unit 19 is provided for inputting sound data of sound sample sounds, inputting operations, displaying the progress of processing, and the like.

  The sound identification condition setting support device 1 having such a configuration can be realized by a personal computer, for example. More specifically, the arithmetic processing unit 11 can be realized by a CPU (Central Processing Unit) built in a personal computer. Further, the spectrogram generation unit 12, the histogram generation unit 13, the discrimination function generation unit 14, the accuracy rate calculation unit 15, the identification condition selection unit 16, and the control unit 17 make the personal computer function as the sound identification condition setting support device 1. This can be realized by causing the CPU to read and execute the computer program. The storage unit 18 can be realized by a storage device (for example, a hard disk drive device, a flash memory, or the like) included in the personal computer. The input / output unit 19 can be realized by an input / output device (for example, a keyboard, a mouse, a monitor, etc.) included in the personal computer.

(Sound identification condition setting support processing)
2 and 3 show a sound identification condition setting support process performed by the sound identification condition setting support device 1. FIG. As a preparation for processing, for example, the worker prepares audio data of 100 abnormal sound samples A1 to A100 and audio data of 100 non-natural sound samples B1 to B100.

  Subsequently, the operator operates the input / output unit 19 of the sound identification condition setting support device 1 to set the sound data of the abnormal sound samples A1 to A100 and the sound data of the non-natural sound samples B1 to B100 as sound recognition conditions. Input to the setting support apparatus 1. When these sound data are input to the sound identification condition setting support device 1, the sound identification condition setting support device 1 receives the sound data of the abnormal sound samples A1 to A100 and the non-natural sound samples B1 to B100. In addition to being able to identify each one, it is possible to identify whether the sound data is the sound data of an abnormal sound sample or the sound data of an allophone sample sound. As a method for realizing this, for example, each voice data is identified with identification information for identifying each voice data, and whether each voice data is voice data of abnormal sample sound or voice data of non-sound sample sound. There is a method of attaching classification information indicating whether or not there is. This method can be easily realized, for example, by devising a method for naming the file name of the audio data.

  Subsequently, the worker inputs an instruction to start the sound identification condition setting support process to the sound identification condition setting support device 1. In response to this, the sound identification condition setting support device 1 starts the sound identification condition setting support process shown in FIGS.

  In the sound identification condition setting support process, as shown in FIG. 2, first, the spectrogram generation unit 12 generates spectrogram data for each of the sound data of the abnormal sound samples A1 to A100 and the non-natural sound samples B1 to B100. Then, the generated spectrogram data is stored in the storage unit 18 (step S1).

  Subsequently, the control unit 17 sets the number of blocks when dividing the predetermined frequency range of the spectrogram data into a plurality of blocks according to a predetermined rule (step S2). For example, in the present embodiment, as the predetermined rule, the minimum number of blocks is set to 20, the maximum value is set to 80, the number of blocks set for the first time is set to the minimum value, and the setting of the number of blocks is changed once. A rule of increasing the number of blocks by 5 every time is adopted.

  Here, FIG. 9 shows a state in which the number of blocks is set to 20, and a predetermined frequency range of the spectrogram of the collected sound is divided into blocks G. FIG. 10 shows a state in which the number of blocks is set to 30 and the spectrogram of the collected sound is predetermined. The frequency range is divided into blocks H. As can be seen from FIGS. 9 and 10, as the number of blocks increases, the width (frequency range) of each block decreases.

  Subsequently, as shown in FIG. 2, the control unit 17 randomly selects one unselected spectrogram data from the spectrogram data of the abnormal sound samples A1 to A100, and stores the spectrogram data in the storage unit 18. (Step S3). Subsequently, the control unit 17 stores, in the storage unit 18, history information indicating which abnormal sound sample data is selected from the spectrogram data of the abnormal sound samples A1 to A100. This history information is used by the control unit 17 to specify spectrogram data that has not been selected in step S3 or S10.

  Subsequently, the histogram generation unit 13 divides the predetermined frequency range of the selected spectrogram data into blocks having the number of blocks set in step S2, and generates histogram data of intensity distribution of each divided block. Each generated histogram data is stored in the storage unit 18 (step S4).

  Subsequently, the control unit 17 determines whether or not the processing for generating the histogram data for, for example, 50 spectrogram data among the spectrogram data of the abnormal sound samples A1 to A100 is finished (step S5). When the process of generating histogram data for 50 spectrogram data of the spectrogram data of the noise sample sounds A1 to A100 has not been completed (step S5: NO), the control unit 17 returns the process to step S3.

  On the other hand, when the process of generating the histogram data for 50 spectrogram data among the spectrogram data of the abnormal sound samples A1 to A100 is completed (step S5: YES), the control unit 17 performs the non-noise sample sound. One spectrogram data that has not been selected is selected at random from the spectrogram data of B1 to B100, and the spectrogram data is read from the storage unit 18 (step S6). Subsequently, the control unit 17 stores, in the storage unit 18, history information that indicates which spectrogram data of which anomalous sample sound is selected from the spectrogram data of the allomalous sample sounds B1 to B100. This history information is used by the control unit 17 to specify spectrogram data that has not been selected in the subsequent step S6 or S15.

  Subsequently, the histogram generation unit 13 divides the predetermined frequency range of the selected spectrogram data into blocks having the number of blocks set in step S2, and generates histogram data of intensity distribution of each divided block. Each generated histogram data is stored in the storage unit 18 (step S7).

  Subsequently, the control unit 17 determines whether or not the processing for generating the histogram data for, for example, 50 spectrogram data among the spectrogram data of the allophone sample sounds B1 to B100 is finished (step S8). When the process of generating the histogram data for 50 spectrogram data of the spectrogram data of the allophone sample sounds B1 to B100 has not been completed (step S8: NO), the control unit 17 returns the process to step S6.

  On the other hand, when the process of generating the histogram data for 50 spectrogram data among the spectrogram data of the allophone sample sounds B1 to B100 is finished (step S8: YES), the discriminant function generation unit 14 performs step S4. The histogram data of each block in the generated 50 abnormal sound samples and the histogram data of each block in the 50 sound sample sounds generated in step S7 are read from the storage unit 18, and these histogram data are read out. Is used to generate a discrimination function (step S9). For the generation of the discriminant function, for example, a pattern recognition method using a learning model such as a support vector machine (see, for example, JP-A-2005-352997) is employed. Note that another learning model such as a neural network may be employed for generating the discrimination function. Subsequently, the discriminant function generation unit 14 stores the generated discriminant function in the storage unit 18 together with the number of blocks corresponding thereto (for example, information such as a numerical value indicating the number of blocks currently set).

  Subsequently, the control unit 17 randomly selects one non-selected spectrogram data from the spectrogram data of the abnormal sound samples A1 to A100, and reads the spectrogram data from the storage unit 18 (step S10). Subsequently, the control unit 17 stores, in the storage unit 18, history information indicating which abnormal sound sample data is selected from the spectrogram data of the abnormal sound samples A1 to A100. The history information is used by the control unit 17 to specify unselected spectrogram data in the subsequent step S3 or S10.

  Subsequently, the histogram generation unit 13 divides the predetermined frequency range of the selected spectrogram data into blocks having the number of blocks set in step S2, and generates histogram data of intensity distribution of each divided block. Each generated histogram data is stored in the storage unit 18 (step S11).

  Subsequently, as shown in FIG. 3, the accuracy rate calculation unit 15 inputs the histogram data generated in step S11 to the discrimination function generated in step S9, and the discrimination function uses the discrimination function to detect the noise sample sound. The presence / absence of abnormal noise is identified (step S12).

  Subsequently, the correct answer rate calculation unit 15 determines whether the identification function is correct or not (step S13). In the determination in step S13, the answer is correct when the result of the identification by the identification function is “abnormal noise”, otherwise it is incorrect. Subsequently, the correct answer rate calculation unit 15 stores the determination result (for example, information such as a flag indicating the determination result) in the storage unit 18 as determination result information.

  Subsequently, the control unit 17 executes discrimination by the discrimination function using, for example, 50 spectrogram data among the spectrogram data of the abnormal sound samples A1 to A100, and finishes the process of determining the execution result of the discrimination function. It is determined whether or not (step S14). When the execution of discrimination by the discrimination function using 50 spectrogram data among the spectrogram data of the abnormal sound samples A1 to A100 and the determination of the result are not finished (step S14: NO), the control unit 17 The process returns to step S10 in FIG.

  On the other hand, when the execution of the discrimination by the discrimination function using 50 spectrogram data of the spectrogram data of the unusual sound samples A1 to A100 and the determination of the result are finished (step S14: YES), the control unit 17 However, one non-selected spectrogram data is randomly selected from the spectrogram data of the allophone sample sounds B1 to B100, and the spectrogram data is read from the storage unit 18 (step S15). Subsequently, the control unit 17 stores, in the storage unit 18, history information that indicates which spectrogram data of which anomalous sample sound is selected from the spectrogram data of the allomalous sample sounds B1 to B100. This history information is used by the control unit 17 to specify spectrogram data that has not been selected in the subsequent step S6 or S15.

  Subsequently, the histogram generation unit 13 divides the predetermined frequency range of the selected spectrogram data into blocks having the number of blocks set in step S2, and generates histogram data of intensity distribution of each divided block. Each generated histogram data is stored in the storage unit 18 (step S16).

  Subsequently, the correct answer rate calculation unit 15 inputs the histogram data generated in step S16 to the discrimination function generated in step S9, and uses the discrimination function to identify the presence or absence of abnormal sounds in the non-abnormal sound sample. Execute (Step S17).

  Subsequently, the correct answer rate calculation unit 15 determines whether the discrimination function is correct or not (step S18). In the determination in step S18, the answer is correct when the result of the identification by the identification function is “no abnormal sound”, otherwise it is incorrect. Subsequently, the correct answer rate calculation unit 15 stores the determination result in the storage unit 18 as determination result information.

  Subsequently, the control unit 17 has performed identification using an identification function using, for example, 50 spectrogram data among the spectrogram data of the allophone sample sounds B1 to B100, and has finished the process of determining identification based on the identification function It is determined whether or not (step S19). When the execution of the discrimination by the discriminant function using the 50 spectrogram data of the spectrogram data of the allophone sample sounds B1 to B100 and the determination of the result are not finished (step S19: NO), the control unit 17 The process returns to step S15.

  On the other hand, when the execution of the discrimination by the discrimination function using the 50 spectrogram data of the spectrogram data of the allophone sample sounds B1 to B100 and the determination of the result are finished (step S19: YES), the control unit 17 Shifts the process to step S20.

  By the processes in steps S3 to S19, half of the spectrogram data randomly selected from the spectrogram data of the abnormal sound samples A1 to A100 and the spectrogram data of the allophone sample sounds B1 to B100 are randomly selected. The discriminant function is generated using the half of the spectrogram data selected in the above, and whether or not the execution result of the discriminant function is correct is determined by determining the remaining half of the spectrogram data of the sampled sounds A1 to A100 and the remaining half of the spectrogram data. That is, the other half of the spectrogram data of the non-sound sample sounds B1 to B100 is used.

  Subsequently, in step S20, the control unit 17 determines whether or not the generation of the identification function and the determination of the execution result (steps S3 to S19) have been repeated a plurality of times (for example, 200 times). If the generation of the discriminant function and the determination of the execution result have not been repeated 200 times (step S20: NO), the control unit 17 returns the process to step S3 and repeats the spectrogram data random selection from the beginning. All the history information indicating the data selection history is erased (initialized).

  On the other hand, when the generation of the identification function and the determination of the execution result are repeated 200 times (step S20: YES), the control unit 17 shifts the process to step S21.

  Through the processes in steps S3 to S20, 200 combinations of 50 different sample sounds used for generating the discriminant function and 200 combinations of 50 noise samples used for generating the discriminant function are provided. Thus, 200 discriminant functions are generated using these 200 combinations. Further, by the processing of steps S3 to S20, the combination of 50 different sampled sounds used for determining the execution result of the discriminant function, and 50 non-sound sampled sounds used for determining the execution result of the discriminant function. 200 combinations are generated, and the identification results are determined for each identification function using these 200 combinations. When Steps S3 to S20 have been executed 200 times, the storage unit 18 stores 20000 pieces of determination result information. Thus, by repeatedly generating the identification function and determining the execution result and obtaining a large amount of determination result information, it is possible to improve the reliability of the identification accuracy rate calculated in step S21 later.

  Subsequently, in step S21, the accuracy rate calculation unit 15 uses 20,000 pieces of determination result information obtained in steps S3 to S20, and uses one discriminant function generated with one currently set number of blocks. The identification correct answer rate is calculated, and the calculated identification correct answer rate is stored in the storage unit 18.

  Here, FIG. 11 shows an example of the method of calculating the identification correct rate in step S21. In FIG. 11, “◯” indicates that the determination result in step S13 or S18 is correct, and “X” indicates that the determination result in step S13 or S18 is incorrect. “-” Indicates that the spectrogram data of the sample sound is used for generating the discriminant function, and has not been used for discriminating correctness by the discriminant function in step S13 or S18. As shown in FIG. 11, in the process of calculating the identification correct rate in step S21, first, the identification correct rate is calculated for the spectrogram data of the sample sounds A1 to A100 and B1 to B100. The classification accuracy rate for one spectrogram data is the sum of the number of correct answers and the number of incorrect answers in the identification result obtained by inputting the histogram data of the spectrogram data into the identification function. It can be determined by dividing. Next, the average value of the identification accuracy rates of the spectrogram data of the sample sounds A1 to A100 and B1 to B100 is calculated, and this average value is used as the identification accuracy rate of the identification function generated by the number of one block currently set. Used as

  Subsequently, as illustrated in FIG. 3, the control unit 17 determines whether or not the processing of steps S2 to S21 has been completed for all the settings of the number of blocks (step S22). When the execution of steps S2 to S21 has not been completed for all settings of the number of blocks (step S22: NO), the control unit 17 returns the processing to step S2 in FIG. In step S2, the control unit 17 changes the number of blocks in accordance with the predetermined rule, and then the control unit 17, the histogram generation unit 13, and the correct answer rate calculation unit 15 execute the processes of steps S3 to S21.

  On the other hand, when the execution of the processing of steps S2 to S21 is completed for all the settings of the number of blocks (step S22: YES), the control unit 17 shifts the processing to step S23. According to the predetermined rule regarding the setting of the number of blocks described above, there are 13 types of setting of the number of blocks. Therefore, the processing of steps S2 to S21 is repeated 13 times while the number of blocks is changed. The discrimination accuracy rate of the discrimination function generated by the number of street blocks is calculated. When the execution of the processing of steps S2 to S21 is completed for the setting of 13 types of blocks, the storage unit 18 stores 13 identification accuracy rates respectively corresponding to the number of blocks.

  Here, FIG. 12 shows an example of the identification accuracy rate of the identification function corresponding to the number of blocks set according to the predetermined rule. According to FIG. 12, the identification accuracy rate of 50 blocks is the highest at 97%.

  Subsequently, in step S23 in FIG. 3, the identification condition selection unit 16 reads out a plurality of identification accuracy rates stored in the storage unit 18 and identifies the highest identification accuracy rate from among them. Subsequently, the identification condition selection unit 16 selects the number of blocks corresponding to the identification function having the identified identification accuracy rate, and stores the selected number of blocks in the storage unit 18 as the determined number of blocks.

  Subsequently, all the spectrogram data of the abnormal sound samples A1 to A100 and the non-natural sound samples B1 to B100 generated under the setting of the number of blocks stored as the determined number of blocks. The discrimination function is newly generated using all the histogram data in, and the generated discrimination function is stored in the storage unit 18 as a definite discrimination function (step S24). The determined block number and the determined identification function are used for abnormal sound determination processing by the abnormal sound determination device 21 described later.

  As described above, according to the sound identification condition setting support device 1 according to the embodiment of the present invention, an identification function that can accurately identify the presence or absence of an abnormal sound in a collected sound can be quickly and easily performed with a small work load. Can be generated. That is, since the generation of the discrimination function by the sound discrimination condition setting support device 1 is automatically performed as long as voice data is input, the task of generating the discrimination function, that is, the task corresponding to the feature extraction of abnormal noise, Compared with the conventional feature extraction operation of finding the features of abnormal sounds while looking at the spectrogram of the collected sound displayed on the computer monitor, it can be greatly simplified. Therefore, the working time can be shortened and the burden on the worker can be reduced. Also, if a sample sound is prepared, the sound identification condition setting support device 1 can automatically generate a discrimination function, so that a discrimination function having a high accuracy discrimination ability can be obtained regardless of the experience of the operator. Can be generated.

  In the sound identification condition setting support process in the above-described embodiment, the case where the generation of the identification function and the determination of the identification result of the identification function (steps S3 to S20) are repeated 200 times, for example, is described in the present invention. Not limited to this, the generation of the discrimination function and the determination of the discrimination result of the discrimination function may be performed once to 199 times or 201 times or more.

  Further, in the sound identification condition setting support process in the above-described embodiment, after selecting the number of confirmed blocks, the sampled sound samples A1 to A100 generated under the setting of the number of blocks selected as the number of confirmed blocks and A case where a discrimination function is generated again using all histogram data in all spectrogram data of the allophone sample sounds B1 to B100, and the generated discrimination function is selected as a definite discrimination function (step S24 in FIG. 3). As an example, the present invention is not limited to this, and the discriminant function generated in step S9 in FIG. 2 may be selected as the deterministic discriminant function under the setting of the number of blocks selected as the deterministic block number. .

  Further, in the sound identification condition setting support processing in the above-described embodiment, an example in which the spectrogram data of the sample sound used for generating the discrimination function and the spectrogram data of the sample sound used for determining the discrimination result of the discrimination function are respectively selected at random. However, the present invention is not limited to this. For example, before starting the sound identification condition setting support process, the operator classifies the collected sample sounds into a sample sound used for generating the identification function and a sample sound used for determining the identification result by the identification function, The sound identification condition setting support device 1 selects the spectrogram data of the sample sound used for generating the discrimination function and the spectrogram data of the sample sound used for the determination of the discrimination result of the discrimination function according to the classification of the operator. May be.

(Configuration of abnormal sound determination device)
FIG. 13 shows an abnormal sound determination device that determines the presence or absence of abnormal noise in the collected sound using the determined identification function and the number of determined blocks generated by the sound identification condition setting support device 1 according to the embodiment of the present invention. . In FIG. 13, the abnormal sound determination device 21 includes an arithmetic processing unit 30. The arithmetic processing unit 30 divides the spectrogram data into a plurality of blocks, a spectrogram generation unit 31 that generates spectrogram data from audio data, The histogram generation unit 32 that generates histogram data of the sound intensity distribution, the histogram data in each block is input to the identification function, the noise identification unit 33 that identifies the presence or absence of abnormal noise, and the flow of abnormal noise determination processing And a control unit 34 that performs control and the like. Further, the abnormal sound determination device 21 has a storage unit 35 for storing various data to be held during the abnormal sound determination processing, and input / output for performing input of sound data of the sound to be determined, operation input, display of processing results, and the like. Part 36.

  The abnormal sound determination device 21 can be realized by a personal computer, for example. Specifically, the arithmetic processing unit 30 can be realized by a CPU built in a personal computer. The spectrogram generation unit 31, the histogram generation unit 32, the abnormal sound identification unit 33, and the control unit 34 can be realized by causing the CPU to read and execute a computer program. The storage unit 35 can be realized by a storage device such as a hard disk drive device of a personal computer. The input / output unit 36 can be realized by an input / output device such as a keyboard or a monitor included in the personal computer. Note that the sound identification condition setting support device 1 and the abnormal sound determination device 21 may be operated simultaneously by a single personal computer.

(Abnormal sound judgment processing)
FIG. 14 shows an abnormal sound determination process performed by the abnormal sound determination device 21. The abnormal sound determination device 21 can determine the presence or absence of abnormal noise, for example, for a collected sound over a relatively long period of time (for example, 24 hours) collected at a location close to a power transmission line or equipment. The entire collected sound that is the target of the abnormal sound determination is referred to as a sound to be determined. In the abnormal sound determination processing by the abnormal sound determination device 21, the sound data of the sound to be determined is divided into several sections of a relatively short time (for example, 1 hour), and the presence or absence of the abnormal sound is identified for each section.

  In preparation for the abnormal sound determination process, the operator inputs the confirmed identification function and the number of determined blocks generated by the sound identification condition setting support device 1 to the abnormal sound determination device 21 and stores them in the storage unit of the abnormal sound determination device 21. 35. Further, the worker inputs the sound data of the sound to be determined to the abnormal sound determination device 21 and stores it in the storage unit 35 of the abnormal sound determination device 21. Subsequently, when the operator inputs an instruction to start the abnormal sound determination process to the abnormal sound determination apparatus 21, the abnormal sound determination apparatus 21 starts the abnormal sound determination process.

  As shown in FIG. 14, in the abnormal sound determination process, first, the control unit 34 reads the definite identification function and the definite block number from the storage unit 35 (step S <b> 31).

  Subsequently, the control unit 34, in the sound data of the sound to be determined, the first section (section closest to the start time of the sound data of the sound to be determined) among the sections that have not yet been identified for the presence or absence of abnormal sound. Is extracted (step S32).

  Subsequently, the spectrogram generation unit 31 generates spectrogram data of the voice data extracted in step S32 (step S33).

  Subsequently, for the spectrogram data generated in step S33, the histogram generation unit 32 divides a predetermined frequency range into blocks of the determined number of blocks, and generates histogram data of sound intensity distribution for each block (step S34). .

  Subsequently, the abnormal sound identification unit 33 inputs the histogram data of each block to the definite identification function, executes the identification of the presence or absence of abnormal noise, and stores the identification result in the storage unit 35 (step S35).

  Subsequently, the control unit 34 determines whether or not the identification of the presence or absence of abnormal sounds has been completed for all sections of the sound data of the sound to be determined (step S36). If the identification of the presence / absence of abnormal sound has not been performed for all sections of the sound data of the sound to be determined (step S36: NO), the process returns to step S32, and the control unit 34 performs the sound corresponding to the next section. Data is extracted from the sound data of the sound to be determined, and the spectrogram generating unit 31, the histogram generating unit 32, and the abnormal sound identifying unit 33 execute the processes of steps S33 to S35.

  On the other hand, when the identification of the presence / absence of abnormal sound has been completed for all sections of the sound data of the sound to be determined (step S36: YES), the control unit 34 identifies the presence / absence of abnormal sound for the sound to be determined. The result is displayed on a monitor or the like for each section (step S37). By looking at the monitor, the operator can immediately grasp the presence or absence of abnormal noise in the sound to be determined. In addition, since the presence / absence of abnormal noise in the judged sound can be grasped for each section, the collected sound used as the judged sound is relatively long, and the abnormal sound included therein is relatively short. Even in this case, it is possible to quickly find an abnormal sound in the collected sound.

  It should be noted that the present invention can be appropriately changed without departing from the spirit or idea of the invention that can be read from the claims and the entire specification, and the sound identification condition setting support device and the sound identification condition accompanying such a change. A setting support method is also included in the technical idea of the present invention.

DESCRIPTION OF SYMBOLS 1 Sound identification condition setting assistance apparatus 11 Operation processing part 12 Spectrogram generation part 13 Histogram generation part 14 Discrimination function generation part 15 Correct answer rate calculation part 16 Discrimination condition selection part 17 Control part 18 Storage part 19 Input / output part 21 Abnormal sound determination apparatus 30 Arithmetic processing unit 31 Spectrogram generation unit 32 Histogram generation unit 33 Abnormal sound identification unit 34 Control unit 35 Storage unit 36 Input / output unit

Claims (3)

A sound identification condition setting support device that supports setting of an identification condition for identifying the presence or absence of abnormal noise,
It receives spectrogram data showing the spectrograms of multiple abnormal sample sounds that contain abnormal sounds and non-noisy sample sounds that do not contain abnormal sounds, and divides the predetermined frequency range in each spectrogram data into multiple blocks. A histogram generation unit that generates histogram data indicating a histogram of the sound intensity distribution in each block;
Histogram data of each block for some noise sample sounds of the plurality of noise sample sounds, and each of the noise data of some noise samples of the plurality of noise sample sounds An identification function generator for generating an identification function for identifying the presence or absence of abnormal noise using block histogram data;
Histogram data of each block for some other abnormal sound samples of the plurality of abnormal sound samples, and some other non-noise samples of the non-natural sound samples Accurate rate calculation that inputs the histogram data of each block for sound into the discrimination function to identify the presence or absence of abnormal noise, and calculates the discrimination accuracy rate that is the probability that the discrimination function correctly identifies the presence or absence of abnormal noise And
While changing the setting of the number of blocks when dividing the predetermined frequency range in each spectrogram data into a plurality of blocks, the processing by the histogram generation unit, the discrimination function generation unit and the accuracy rate calculation unit is executed a plurality of times. As a result of executing the process a plurality of times, among the plurality of discriminant functions generated by the discriminant function generator, the discriminant function having the highest discrimination accuracy rate and the number of blocks set when generating the discriminant function are obtained. A sound identification condition setting support device, comprising: an identification condition selection unit that selects the identification condition . The sound identification condition setting support apparatus according to claim 1, further comprising a spectrogram generation unit that generates spectrogram data indicating a spectrogram of the abnormal sound sample sound or the non-natural sound sample sound. A sound identification condition setting support method for supporting setting of an identification condition for identifying the presence or absence of abnormal noise,
For each piece of spectrogram data showing the spectrograms of a plurality of abnormal sample sounds including abnormal sounds and a plurality of non-noise sample sounds not including abnormal sounds, the predetermined frequency range in the spectrogram data is divided into a plurality of blocks. A histogram generation step of generating histogram data indicating a histogram of the sound intensity distribution in each block;
Histogram data of each block for some noise sample sounds of the plurality of noise sample sounds, and each of the noise data of some noise samples of the plurality of noise sample sounds An identification function generating step for generating an identification function for identifying the presence or absence of abnormal noise using the histogram data of the block;
Histogram data of each block for some other abnormal sound samples of the plurality of abnormal sound samples, and some other non-noise samples of the non-natural sound samples Accurate rate calculation that inputs the histogram data of each block for sound into the discrimination function to identify the presence or absence of abnormal noise, and calculates the discrimination accuracy rate that is the probability that the discrimination function correctly identifies the presence or absence of abnormal noise A process,
While changing the setting of the number of blocks when dividing the predetermined frequency range in each spectrogram data into a plurality of blocks, the histogram generation step, the discrimination function generation step and the accuracy rate calculation step are executed a plurality of times, As a result of executing the process of a plurality of times, among the plurality of discriminant functions generated in the discriminant function generating step, the discriminant function having the highest discrimination accuracy rate and the number of blocks set when generating the discriminant function are A sound identification condition setting support method characterized by selecting as an identification condition.