Sound analysis device, sound analysis method and sound analysis program

Abstract

To enable acoustic information suitable for individual identification to be easily selected from among acoustic information included in call.SOLUTION: A sound analysis method (S1) includes an extraction step (S11) for making a computer extract specific parts of sounds emitted by a plurality of individual animals, a digitization step (S12) for making the computer digitize each piece of acoustic information in each extracted specific part, a conversion step (S13) for making the computer take a logarithm of each piece of digitized acoustic information, and an output step (S14) for allowing the computer make an output part output information based on the logarithm.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates to a voice analysis device, a voice analysis method, and a voice analysis program. [Background technology]

[0002] Various techniques have been proposed for identifying individual animals by using their calls. For example, Non-Patent Document 1 describes a method for extracting acoustic information (fundamental frequency, power, Mel-frequency cepstrum coefficients, etc.) from the calls of cattle, classifying the cattle by age based on the acoustic information, and further identifying the individual cattle classified by age. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Individual Cow Identification Considering Age-Related Changes in Cow Calls, 34th National Conference of the Japanese Society for Artificial Intelligence, 2020 Summary of the Invention [Problem to be solved by the invention]

[0004] Among the acoustic information contained in animal cries, the acoustic information suitable for individual identification varies depending on the type of animal, etc. Therefore, before performing individual identification, it is necessary to consider which acoustic information to use for individual identification. However, with conventional technology, there is no option but to look at a list of acoustic information values, making it difficult to select which acoustic information to use for identification.

[0005] One aspect of the present invention has been made in consideration of the above-mentioned problems, and aims to make it possible to easily select acoustic information contained in bird calls that is suitable for individual identification. [Means for solving the problem]

[0006] In order to solve the above problem, an audio analysis method according to one aspect of the present invention includes an extraction step in which a computer extracts specific parts of audio emitted by each of a plurality of individual animals; a quantification step in which the computer digitizes each piece of acoustic information in each extracted specific part; a conversion step in which the computer takes the logarithm of each piece of quantified acoustic information; and an output step in which the computer causes an output unit to output information based on the logarithm.

[0007] Another aspect of the present invention provides an audio analysis device that includes an extraction unit that extracts specific portions of audio emitted by each of a plurality of animals, a quantification unit that digitizes each piece of acoustic information in each extracted specific portion, a conversion unit that takes the logarithm of each piece of digitized acoustic information, and an output unit that outputs information based on the logarithm.

[0008] Another aspect of the audio analysis program of the present invention causes a computer to perform an extraction process that extracts specific parts of the audio produced by each of multiple animals, a quantification process that digitizes each piece of acoustic information in each of the extracted specific parts, a conversion process that takes the logarithm of each piece of quantified acoustic information, and an output process that outputs information based on the logarithm to an output unit. [Effects of the Invention]

[0009] According to one aspect of the present invention, acoustic information suitable for individual identification can be easily selected from acoustic information contained in cries. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a block diagram illustrating an example of a functional configuration of a voice analysis device according to an embodiment of the present invention. [Figure 2] 2A and 2B are diagrams illustrating the content of extraction processing executed by the device. [Figure 3] 10 is a table showing the results of the digitization process performed by the device. [Figure 4] 10 is a radar chart showing an example of logarithmic information output by the device. [Figure 5]10 is a flowchart illustrating a flow of a voice analysis method according to an embodiment of another aspect of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0011] <Speech analysis device 1> First, a voice analysis device 1 according to an embodiment of one aspect of the present invention will be described in detail.

[0012] [Configuration of voice analysis device 1] As shown in FIG. 1, the voice analysis device 1 according to this embodiment includes an input unit 11, an output unit 12, a storage unit 13, and a calculation unit .

[0013] [Input unit 11] Animal cry data is input to the input unit 11. The input unit 11 is composed of a microphone that collects animal cries and generates animal cry data, a communication module that receives animal cry data from other devices, a terminal for connecting a cable for receiving animal cry data from other devices, a drive that reads the animal cry data from a recording medium, etc.

[0014] [Output unit 12] The output unit 12 according to this embodiment is configured with a display device (for example, a liquid crystal display) that displays the processing results by the calculation unit 14. The output unit 12 may also be configured with a communication module that transmits the processing result data to another device, a terminal for connecting a cable for transmitting the processing result data to another device, a drive that writes the processing result data to a recording medium, etc.

[0015] [Storage unit 13] The storage unit 13 stores a voice analysis program 131. The voice analysis program 131 describes the operation of the voice analysis device 1. The storage unit 13 according to this embodiment is configured with a semiconductor memory, a hard disk, etc. Furthermore, the storage unit 13 according to this embodiment is configured to be able to store data of bird calls input to the input unit 11.

[0016] [Calculation unit 14] The calculation unit 14 includes an extraction unit 141, a digitization unit 142, a conversion unit 143, and an output processing unit 144. The voice analysis device 1 according to this embodiment further includes a generation unit 145. The calculation unit 14 according to this embodiment is configured with a processor. When this processor executes the voice analysis program 131 stored in the storage unit 13, the voice analysis program 131 causes the calculation unit 14 to function as each of the control blocks 141 to 145.

[0017] (Extraction part 141) The extraction unit 141 executes extraction processing. In the extraction processing, the extraction unit 141 extracts specific portions of sounds emitted by each of a plurality of individual animals. The extraction unit 141 according to this embodiment extracts the specific portions from the call data stored in the storage unit 13. Note that the extraction unit 141 may be configured to directly extract the specific portions from the call data input to the input unit 11. Furthermore, the extraction unit 141 according to this embodiment extracts, as the specific portion, a portion of the sound that extends linearly in the time axis direction when represented in a spectrogram (the portion sandwiched between two dashed lines L1 and L2 in FIG. 2). Note that if the call data input to the input unit 11 is composed of only the specific portions of the sound, the calculation unit 14 does not need to include the extraction unit 141.

[0018] (Digitization unit 142) The digitizing unit 142 executes a digitizing process. In the digitizing process, the digitizing unit 142 digitizes each piece of acoustic information in each extracted specific portion. The acoustic information digitized by the digitizing unit 142 according to the present embodiment includes, for example, a fundamental frequency, intensity, and formants. The fundamental frequency and intensity digitized by the digitizing unit 142 according to the present embodiment each include a maximum value, a minimum value, and an average value. The formants digitized by the digitizing unit 142 according to the present embodiment include the first formant to the fifth formant. The formants digitized by the digitizing unit 142 may not include at least one of the second formant to the fifth formant, or may include at least one of the sixth formant to the nth formant. When the digitizing unit 142 digitizes the acoustic information, a list of numerical values ​​indicating each piece of acoustic information for each individual is obtained, as shown in FIG. 3. As shown in FIG. 3, the numerical values ​​of acoustic information vary greatly depending on the type of sound. For example, there is a two-digit difference between the intensity and the formant.

[0019] (Conversion unit 143) The conversion unit 143 executes a conversion process. In the conversion process, the conversion unit 143 takes the logarithm of each piece of digitized acoustic information. As a result, the difference between the numerical values ​​of each piece of acoustic information that was significantly different appears smaller.

[0020] (Generation unit 145) The generation unit 145 converts each logarithm obtained by the conversion unit 143 into visual data. Each piece of acoustic information is converted into a logarithm. Therefore, the generation unit 145 according to this embodiment generates a radar chart connecting the plots of each logarithm. Note that the visual data may be anything that can display each logarithm in a list, and may be, for example, a bar graph, or just a plot.

[0021] (output processing unit 144) The output processing unit 144 executes output processing. In the output processing, the output processing unit 144 causes the output unit to output information based on the logarithm. The output processing unit 144 according to this embodiment controls the output unit 12. As a result, the output unit 12 outputs information based on the visual data. As described above, the calculation unit 14 according to this embodiment includes a generation unit 145. Furthermore, the output unit 12 according to this embodiment is configured with a display device. Therefore, the output unit 12 according to this embodiment displays a radar chart, for example, as shown in FIG. 4. This radar chart shows that when the target animal is a cow, there are large individual differences in fundamental frequency (individual identification can be easily performed by comparing fundamental frequencies).

[0022] [Effects of voice analysis device 1] In the voice analysis device 1 described above, the conversion unit 143 converts the numerical values ​​of each piece of acoustic information, which have large differences in value or digits, into values ​​with smaller differences by taking logarithms. Therefore, the voice analysis device 1 makes it easy to compare each piece of acoustic information on the same layer (radar chart), and it is easy to understand which piece of acoustic information has large individual differences. In other words, the voice analysis device 1 makes it easy to select acoustic information suitable for individual identification from the acoustic information contained in the bird's cry. As a result, individual identification can be performed easily and accurately.

[0023] <Sound analysis method S1> Next, a speech analysis method S1 according to another embodiment of the present invention will be described.

[0024] [Flow of voice analysis method S1] 5, the voice analysis method S1 includes an extraction step S11, a digitization step S12, a conversion step S13, and an output step S14. The voice analysis method S1 according to this embodiment further includes a generation step S15.

[0025] [Extraction step S11] In extraction step S11, a computer extracts specific portions of sounds emitted by each of a plurality of animals. In extraction step S11 according to this embodiment, a portion of the sound that extends linearly along the time axis when represented as a spectrogram is extracted as the specific portion. The computer that extracts the specific portion of the sound may be the above-described sound analysis device 1 or another device.

[0026] [Digitalization step S12] After extracting the specific portions of the audio, the process proceeds to a digitization step S12. In the digitization step S12, a computer digitizes each piece of acoustic information in each extracted specific portion. The computer that digitizes the acoustic information may be the above-mentioned audio analysis device 1 or another device.

[0027] [Conversion step S13] After the acoustic information is digitized, the process proceeds to conversion step S13. In conversion step S13, a computer takes the logarithm of each digitized piece of acoustic information. The computer that takes the logarithm of the acoustic information may be the above-mentioned voice analysis device 1 or another device.

[0028] [Generation step S15] After taking the logarithm of the acoustic information, the process proceeds to generation step S15. In generation step S15, the computer generates visual data based on each logarithm obtained in the conversion step. In the generation step, the computer generates a radar chart connecting the plots of each logarithm. The computer that generates the visual data may be the above-mentioned voice analysis device 1 or another device.

[0029] [Output step S14] After taking the logarithm of the acoustic information or generating the visual data, the process proceeds to output step S14. In output step S14, the computer causes the output unit to output information based on the logarithm. The computer that controls the output unit to output the information may be the voice analysis device 1 or another device. Furthermore, the output unit may be provided in the voice analysis device 1 or another device.

[0030] [Effects of voice analysis method S1] In the sound analysis method S1 described above, in the conversion step, numerical values ​​of each piece of acoustic information that differ greatly in value or digit are converted to values ​​with smaller differences by taking logarithms. Therefore, sound analysis method S1 makes it easy to compare each piece of acoustic information on the same layer, and it is easy to understand which acoustic information has the greatest individual differences. In other words, sound analysis method S1 makes it easy to select acoustic information suitable for individual identification from the acoustic information contained in the bird's cry. As a result, individual identification can be performed easily and accurately.

[0031] <Modification> The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims.

[0032] For example, some or all of the functions of the control blocks 141 to 145 can be realized by logic circuits. For example, the scope of the present invention also includes integrated circuits in which logic circuits that function as the control blocks 141 to 145 are formed. In addition, the functions of the control blocks 141 to 145 can also be realized by, for example, a quantum computer.

[0033] Furthermore, each process executed by each of the control blocks 141 to 145 may be executed by AI (Artificial Intelligence). In this case, the AI ​​may operate in the calculation unit 14, or may operate in another device (for example, an edge computer or a cloud server) different from the voice analysis device.

[0034] Furthermore, embodiments obtained by appropriately combining the technical means disclosed in the different embodiments are also included in the technical scope of the present invention.

[0035] 〔summary〕 The audio analysis method according to aspect 1 of the present invention includes an extraction step in which a computer extracts specific portions of audio sounds emitted by a plurality of individual animals; a quantification step in which the computer digitizes each piece of acoustic information in each extracted specific portion; a conversion step in which the computer takes the logarithm of each piece of quantified acoustic information; and an output step in which the computer causes an output unit to output information based on the logarithm.

[0036] A voice analysis method according to aspect 2 of the present invention may be a method in accordance with aspect 1 above, further comprising a generating step in which the computer generates visual data based on each logarithm obtained in the converting step.

[0037] A voice analysis method according to a third aspect of the present invention may be a method in accordance with the second aspect, wherein in the generating step, the computer generates a radar chart connecting plots of each logarithm.

[0038] The audio analysis method according to aspect 4 of the present invention may be a method in any of aspects 1 to 3 above, in which, in the extraction step, a part of the audio that extends linearly in the time axis direction when represented in a spectrogram is extracted as the specific part.

[0039] The audio analysis device according to aspect 5 of the present invention comprises an extraction unit that extracts specific parts of the audio produced by each of a plurality of animals, a quantification unit that digitizes each piece of acoustic information in each of the extracted specific parts, a conversion unit that takes the logarithm of each piece of digitized acoustic information, and an output unit that outputs information based on the logarithm.

[0040] The audio analysis program according to aspect 6 of the present invention is configured to cause a computer to execute an extraction process that extracts specific portions of audio sounds emitted by a plurality of individual animals, a quantification process that digitizes each piece of acoustic information in each extracted specific portion, a conversion process that takes the logarithm of each piece of quantified acoustic information, and an output process that outputs information based on the logarithm to an output unit. [Explanation of symbols]

[0041] 1. Voice analysis device 11 Input section 12 Output section 13 Storage section 131 Voice Analysis Program 14 Arithmetic section 141 Extraction part 142 Quantification Department 143 Conversion Unit 144 Output Processing Unit 145 Generation part S1 Audio analysis method S11 Extraction step S12 Quantification step S13 Conversion Steps S14 Output Step S15 Generation step

Claims

1. An extraction step in which a computer extracts specific portions of sounds emitted by each of the plurality of individual animals; a digitization step in which a computer digitizes each piece of acoustic information in each extracted specific portion; a conversion step in which the computer takes the logarithm of each digitized acoustic information; an output step in which the computer causes an output unit to output information based on the logarithm; Including, Audio analysis methods.

2. The method further includes a generating step in which the computer generates visual data based on each logarithm obtained in the converting step. The speech analysis method according to claim 1 .

3. In the generating step, the computer generates a radar chart connecting the plots of the respective logarithms. The speech analysis method according to claim 2 .

4. In the extraction step, a portion of the audio that extends linearly in a time axis direction when expressed in a spectrogram is extracted as the specific portion. The speech analysis method according to claim 1 .

5. an extraction unit that extracts specific portions of sounds emitted by each of the plurality of individual animals; a digitization unit that digitizes each piece of acoustic information in each extracted specific portion; a conversion unit that takes the logarithm of each digitized acoustic information; an output unit that outputs information based on the logarithm; Equipped with Audio analysis device.

6. On the computer, an extraction process for extracting specific portions of sounds emitted by each of the plurality of animals; a digitization process for digitizing each piece of acoustic information in each extracted specific portion; A conversion process to take the logarithm of each digitized acoustic information; an output process for outputting information based on the logarithm to an output unit; Execute Voice analysis program.

Images