Band-pass filter, and noise detection system using band-pass filter

Abstract

To provide a band-pass filter which considers the ease of hearing of a person with respect to sound and can be downsized and simplified.SOLUTION: A band-pass filter for attenuating or cutting off a part of sound to be an input signal consists of a first filter section having a first bandwidth and a first Q factor and a second filter section having a second bandwidth partially overlapped with the first bandwidth and a second Q factor different from the first Q factor. Frequency characteristics indicating a relation between a frequency and a gain of sound transmitted through the first filter section and the second filter section include: an up area Au in which the gain is up with a first inclination; a down area Ad in which the gain is down with a second inclination; and a gradual increase area Ai which is between the upward area Au and the downward area Ad and in which the gain is continuously gradually increased with an inclination smaller than the first inclination and the second inclination as the frequency becomes higher.SELECTED DRAWING: Figure 5

Description

[Technical field]

[0001] The present invention relates to a bandpass filter, and more particularly to a bandpass filter in which post-filter gain is corrected, and a noise detection system using this bandpass filter. [Background technology]

[0002] Conventionally, systems for detecting abnormalities in devices that emit alarm sounds, such as alarms, have been used in various situations. From the viewpoint of installation, it is preferable for such systems to be small and simple, and from the viewpoint of widespread use, it is preferable for them to be low-cost. Furthermore, when detecting abnormalities in alarm sounds, it is clear that removing noise mixed in with the alarm sounds and collecting only the alarm sounds to determine the presence or absence of an abnormality leads to high accuracy of the determination.

[0003] Generally, noise is removed by using a filter. However, when the frequency characteristic of the filter simply removes the low and high frequency sides of a frequency band having a peak, a level error occurs in the frequency characteristic of the filtered sound (frequency dependency of the gain of the filter), which is a problem. To address this issue, for example, multiple filters are used and the transmission bands of the filters are overlapped to widen the bandwidth of the bandpass filter and make the gain flat. Such a technology is described, for example, in Patent Document 1. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 11-340784 Summary of the Invention [Problem to be solved by the invention]

[0005] However, in order to improve the accuracy of the judgment, it is more preferable to consider the "audibility" of the frequency of the human ear when judging the presence or absence of an abnormality in the alarm. This is because the evaluation of an alarm is based on how it sounds to the human ear. The frequency characteristic in which the intensity of the sound corresponding to each frequency is weighted according to the audibility of the human ear is known as A-weighting. In order to further impart A-weighting to the sound having a flat frequency characteristic obtained by the filter, it is necessary to process this sound using a large number of filters. A configuration requiring a large number of filters may lead to an increase in size, complexity, and cost of the device.

[0006] The present invention has been made in consideration of the above-mentioned points, and relates to a bandpass filter that takes into consideration the ease with which humans can hear sounds and that can be made small and simple, and a noise detection system that uses this bandpass filter. [Means for solving the problem]

[0007] In order to achieve the above object, one form of the bandpass filter of the present invention is a bandpass filter that attenuates or blocks a portion of a sound that is an input signal, and includes a first filter section having a first bandwidth and a first Q value, and a second filter section having a second bandwidth that partially overlaps with the first bandwidth and a second Q value different from the first Q value, and the frequency characteristics indicating the relationship between the frequency and gain of the sound that has passed through the first filter section and the second filter section include an ascending section where the gain ascends with a first slope, a descending section where the gain descends with a second slope, and a gradually increasing region between the ascending section and the descending section where the gain continuously and gradually increases with a slope smaller than the first slope and the second slope as the frequency becomes higher.

[0008] A noise detection system according to one embodiment of the present invention includes a sound collecting unit and the above-described bandpass filter that filters sound collected by the sound collecting unit. Effect of the Invention

[0009] According to the above aspects, it is possible to provide a bandpass filter that takes into consideration the ease with which humans can hear sounds and that can be made small and simple, and a noise detection system that includes this bandpass filter. [Brief description of the drawings]

[0010] [Figure 1] 10A and 10B are diagrams for explaining a level error of a sound transmitted through a bandpass filter. [Diagram 2] FIG. 13 is a diagram for explaining how to obtain a flat frequency characteristic by using two band-pass filters. [Diagram 3] FIG. 1 is a diagram illustrating a known filter used to make flat frequency characteristics closer to A-characteristics. [Figure 4] FIG. 1 is a functional block diagram for explaining a noise detection system according to an embodiment of the present invention. [Diagram 5] 2 is a diagram for explaining the frequency characteristics of sound data that has passed through the bandpass filter 100 of the present embodiment. FIG. [Figure 6] 1A, 1B, and 1C are diagrams illustrating the conditions for realizing the bandpass filter of this embodiment. [Figure 7] 1A and 1B are diagrams for explaining a cutoff region and a criterion for determining the cutoff region according to the present embodiment; [Figure 8] FIG. 1 is a circuit diagram of a bandpass filter created in an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The drawings used in this embodiment are intended to explain the configuration and technical concept of the present invention, and are not intended to specifically limit the configuration or shape of the present invention.

[0012] [overview] First, before describing the present embodiment, the purpose and outline of the present embodiment will be described. Fig. 1 is a diagram for explaining a level error that occurs when a sound that is an input signal is filtered by one bandpass filter. The level error refers to a change in the level (gain) of a sound input to a bandpass filter as an electric signal such as a voltage value, and the gain of the sound output from the bandpass filter, depending on the frequency, and refers to a component that can become an error when measuring the input voltage value.

[0013] FIG. 1 shows the frequency characteristics of a sound in a specified frequency range. Consider using a single band-pass filter to pass sounds in the range of band A and to block sounds with frequencies lower and higher than band A. In this case, it is desirable to obtain a flat signal by eliminating the level error described above in addition to filtering noise. For this purpose, it is conceivable to filter the sound with multiple filters and obtain a flat frequency characteristic by removing peaks in band A. FIG. 2 shows the frequency characteristics of a sound obtained using two band-pass filters (BPF1 and BPF2). In FIG. 2, the thin solid line shows the frequency characteristics of the sound that passed through BPF1, the dashed line shows the frequency characteristics of the sound that passed through BPF2, and the thick solid line shows the frequency characteristics of the sound that passed through BPF1 and BPF2.

[0014] Here, consider evaluating a sound with the frequency characteristics shown in Figure 2. Sound is evaluated, for example, based on whether the volume of the alarm sounded by an alarm installed at a railway crossing is appropriate and whether its audibility satisfies preset conditions. In such evaluations, by considering that there are frequencies that are easy and difficult for the human ear to hear, and weighting the frequencies that are easy for humans to hear, it becomes possible to evaluate the alarm with higher accuracy. The frequency characteristics weighted based on the ease of hearing for humans are known as "A characteristics" and are specified in JIS C1509.

[0015] FIG. 3 shows an example of a filter used to make the frequency characteristic of a sound A-weighted. In the example shown in FIG. 3, six filters F1 to F6 are connected in series. Such a configuration leads to an increase in size and complexity of the sound filter, and further leads to high costs. The object of this embodiment is to provide a bandpass filter that can impart A-weighted characteristics to a sound by simply passing the sound through a bandpass filter after creating sound data with a flat frequency characteristic, without further using multiple filters as shown in FIG. 3.

[0016] [Embodiment] (Noise Detection System) FIG. 4 is a functional block diagram for explaining a noise detection system 101 including the band-pass filter 100 of this embodiment. The band-pass filter 100 is a band-pass filter that attenuates or blocks a part of the sound that is an input signal, and includes at least two filter units, that is, a filter unit 11 that is a first filter unit (indicated as BPF (Band-pass filter) in the figure) and a filter unit 12 that is a second filter unit. The filter unit 11 and the filter unit 12 are cascaded. The filter unit 11 has a first bandwidth and a first Q value (Q1), and the filter unit 12 has a second bandwidth that partially overlaps with the first bandwidth and a second Q value (Q2) that is different from the first Q value. In FIG. 4, the center frequency of the BPF 11 is indicated as f1, and the center frequency of the BPF 12 is indicated as f2. In this embodiment, an example in which such a band-pass filter is applied to filtering an alarm from an alarm device installed near a railroad crossing will be described.

[0017] The noise detection system 101 includes a sound collection unit 13 that collects sounds including warnings at a railroad crossing and inputs the collected sounds to the bandpass filter 100. The sound collection unit 13 may be a known microphone. The microphone can convert sounds into electrical signals (hereinafter referred to as "sound data") and input the collected sounds to the bandpass filter 100. The microphone is configured, for example, by incorporating a sound sensor or a sensor board inside a resin case. The sound sensor of this embodiment is a sound pressure sensor, and a vibration power generation element (not shown) may be used for the sound pressure sensor. The vibration power generation element generates power according to the volume of the warning sound, so that the volume (sound pressure) of the warning sound can be detected from the amount of power generated. Furthermore, the vibration power generation element includes an electret and can be operated by supplying the power generated by itself to the electret. In this way, there is no need to separately supply power to the sound pressure sensor, and the management of the noise detection system 101 can be made easier.

[0018] The microphone may generate an analog electric signal, or may include an analog / digital converter that digitizes the analog signal. Furthermore, this embodiment is not limited to inputting collected sound data to the bandpass filter 100 in real time, and sound data recorded in a file may be input and processed.

[0019] Furthermore, in this embodiment, where it is preferable to reduce the size, simplification, and cost of the configuration, a configuration including a vibration power generation element may be used as a power source (not shown) for the sound collection unit 13 or the bandpass filter 100. In this way, even when the noise detection system is installed outdoors, such as near a railroad crossing alarm, there is no need to install a power source, and power can be supplied to the sound collection unit 13 and the bandpass filter 100 by utilizing the vibrations generated by the running train.

[0020] In the noise detection system 101 of the present embodiment described above, the bandpass filter 100 may be attached near the crossing together with the sound collection unit 13, or the sound collection unit 13 may be attached near an alarm installed at a high place, and the bandpass filter 100 may be attached below the sound collection unit 13, and sound may be transmitted from the sound collection unit 13 to the bandpass filter 100 wirelessly or by wire. In this way, the sound collection unit 13 efficiently collects sound near the alarm, and the bandpass filter 100, which is more frequently subjected to maintenance than the sound collection unit 13, can be installed within reach of an operator. In addition, in this embodiment, the noise detection system 101 may be equipped with an abnormality detection unit (not shown). The abnormality detection unit judges whether the alarm is abnormal or normal from the gain of the sound data that has passed through the bandpass filter 100. Then, this result may be transmitted to a server device that monitors the alarm and the crossing via an internetwork or the like. Furthermore, in this embodiment, sound data collected by the sound collection unit 13 may be transmitted to such a server device, and the sound data may be processed by a bandpass filter on the server device side.

[0021] (Bandpass filter) FIG. 5 is a diagram for explaining the frequency characteristics of sound data that has passed through the bandpass filter 100 of this embodiment. The horizontal axis of FIG. 5 indicates the frequency of the sound data to which the bandpass filter 100 is applied, and the vertical axis indicates the gain of the sound data that is input to and passed through the bandpass filter 100. A curve L1 in FIG. 5 indicates the frequency characteristics of sound data that has passed only through the filter unit 11, and a curve L2 indicates the frequency characteristics of sound data that has passed only through the filter unit 12. A curve L4 indicates the frequency characteristics of sound data that has passed through the bandpass filter 100 in which the filter unit 11 and the filter unit 12 shown in FIG. 4 are cascade-connected. A curve L3 is a curve that indicates A characteristics. In FIG. 5, the bandpass filter 100 blocks sound data on the lower frequency side than the curve L4 where the gain rises, and attenuates the sound data in the frequency range where the gain rises and falls. In this range, the bandpass filter 100 may include a frequency band that transmits 100% of the sound data without attenuation.

[0022] As is clear from FIG. 5, the frequency characteristics of the bandpass filter 100 of this embodiment include an ascending portion Au where the gain ascends with a first slope, a descending portion Ad where the gain descends with a second slope, and a gradually increasing region Ai between the ascending portion Au and the descending portion Ad where the gain continuously and gradually increases with the frequency increasing with a slope smaller than the first slope and the second slope.

[0023] That is, the frequency characteristic of the bandpass filter 100 includes an ascending portion Au that rises exponentially from the low frequency side. The frequency characteristic also includes a descending portion Ad that falls exponentially toward the high frequency side. The slope of the ascending portion Au and the descending portion Ad is obtained by differentiating the curve. The frequency characteristic also includes a gradually increasing region Ai between the ascending portion Au and the descending portion Ad. In the gradually increasing region Ai, the gain increases gradually from the low frequency side toward the high frequency side with a slope smaller than that of the ascending portion Au and the descending portion Ad, and this rising curve follows L3 indicating the A characteristic. However, FIG. 5 is a schematic diagram, and the difference between the sound data and the A characteristic is allowed within a predetermined range even in the gradually increasing region Ai. The degree of the allowable error will be exemplified later.

[0024] In addition, the characteristics of the bandpass filter 100 of this embodiment are classified into a "transmission region" and a "cutoff region" according to the degree of transmission and attenuation of the input sound data, in addition to the above-mentioned ascending portion Au, descending portion Ad, and gradually increasing region Ai. In the bandpass filter 100 of this embodiment, the gradually increasing region occupies at least a part of the transmission region.

[0025] According to this embodiment, the bandpass filter 100 is designed to match the gradual increase region with the frequency band of the alarm, and the sound data on the higher and lower frequency sides than the gradual increase region Ai, which becomes noise, can be removed (blocked) as noise or attenuated. Furthermore, this embodiment can weight the sound data in the gradual increase region Ai according to the ease of human hearing. In this way, when determining whether or not there is an abnormality in the alarm, for example, a case where the alarm is sounded at a predetermined volume but is difficult for humans to hear can be detected and improved. Since the purpose of the alarm is to inform humans of danger, it is important to consider such characteristics of the human ear when checking the alarm. Furthermore, this embodiment can obtain frequency characteristics that follow the A-characteristics by passing the sound through the bandpass filter 100, and no circuit is required for further filtering. This embodiment is advantageous in miniaturizing, simplifying, and reducing the cost of the configuration related to the judgment of the sound data.

[0026] 6(a), 6(b) and 6(c) are diagrams for explaining the conditions for realizing the bandpass filter 100 of the present embodiment described above. All of Fig. 6(a), 6(b) and 6(c) are diagrams showing the frequency characteristics of the bandpass filter 100, with the horizontal axis showing frequency and the vertical axis showing gain. However, Fig. 6(a), 6(b) and 6(c) have different frequency bands of the gradually increasing regions, and the gradually increasing region Ai1 shown in Fig. 6(a) is included in the frequency band from 600 Hz to 850 Hz. The gradually increasing region Ai2 shown in Fig. 6(b) is included in the frequency band from 850 Hz to 1000 Hz, and the gradually increasing region Ai3 shown in Fig. 6(c) is included in the frequency band from 1000 Hz to 1200 Hz.

[0027] The frequency corresponding to the gradual increase region, i.e., the transmission region, can be set by appropriately determining the center frequency and Q value of filter section 11 and filter section 12. The center frequency (f0) and Q value (Q) of filter section 11 and filter section 12, as well as the cutoff region and the transmission region including the gradual increase region, are summarized in Tables 1, 2, and 3 below for the examples of Figures 6(a), 6(b), and 6(c).

[0028] FIG. 7 is a diagram for explaining the transmission range, the cutoff range, and the criteria for determining them shown in the following table. The horizontal axis of FIG. 7 indicates the frequency of the sound data transmitted through the bandpass filter 100, and the vertical axis indicates the gain of the sound data. The solid line in FIG. 7 indicates the characteristics of the bandpass filter, and the dashed line indicates the A-characteristic. The transmission range shown in FIG. 7 is a frequency band in which the bandpass filter 100 can obtain a gain equal to or greater than a predetermined value according to the frequency, and can be determined by the design of the bandpass filter. The gradually increasing region included in the transmission range may include a preset error with respect to the A-characteristic. For example, the "error ±0.4 dB" written in the transmission range column in Table 1 indicates the deviation (error) between the allowable frequency characteristic and the A-characteristic in the transmission range. In the example of FIG. 7, the error e1 is a positive error, and the error e2 is a negative error.

[0029] The cutoff range refers to the frequency band lower than the frequency fL2 at which the gain is attenuated by an amount equal to or greater than D1 with respect to the lowest frequency fL1 in the transmission range, and the frequency band higher than the frequency fH2 at which the gain is attenuated by an amount equal to or greater than D2 with respect to the highest frequency fH1. For example, the "attenuation of 6 dB or more" in the cutoff range column in Table 1 indicates the attenuation amounts D1 and D2.

[0030] Table 1 TIFF2025062724000002.tif41146 Table 2 TIFF2025062724000003.tif36127 Table 3 TIFF2025062724000004.tif41146

[0031] From the above results, it has been found that in this embodiment, by setting the following ranges, it is possible to set a gradual increase region Ai in the target transmission range by combining the characteristics of filter section 11 and filter section 12 as follows in the range of 0 to 2000 Hz. The ratio (f2 / f1) of the center frequency f2 of filter unit 12 to the center frequency f1 of filter unit 11 is set to 1.4 or more and 1.8 or less. Also, the ratio (Q2 / Q1) of the Q value Q2 of filter unit 12 to the Q value Q1 of filter unit 11 is set to 1.1 or more and 1.5 or less.

[0032] In this embodiment, the filter units 11 and 12 are not limited to a specific circuit configuration as long as they satisfy the above characteristics. The order of connection from the input terminals to the output terminals of the filter units 11 and 12 may be reversed. However, the above ratios of f2 / f1 and Q2 / Q1 are not related to the order of connection, but are performed such that the filter unit with a center frequency on the relatively low frequency side is the first filter unit, and the filter unit with a center frequency on the relatively high frequency side is the second filter unit. EXAMPLES

[0033] Next, an example of the embodiment described above will be described. FIG. 8 is a circuit diagram of a bandpass filter created in this example. The bandpass filter shown in FIG. 8 is configured by cascading a filter section (shown as BPF in the figure) 110 and a BPF 120. Both of the filter sections 110 and 120 are multiple feedback bandpass filters. The filter section 110 includes an operational amplifier 116 that configures a feedback circuit together with a resistive element 113, a capacitive element 114 connected in parallel with the resistive element 113, a capacitive element 115 connected between the resistive element 113 and the capacitive element 114, and resistive elements 111 and 112 connected between the capacitive element 114 and an input terminal I. Similarly, the filter section 120 includes resistive elements 121, 122, and 123, capacitive elements 124 and 125, and an operational amplifier 126.

[0034] In this embodiment, the above circuit elements are designed as follows to configure a bandpass filter. Resistor element 111 51kΩ Resistor element 112 12kΩ Resistor element 113 100kΩ Resistor element 121 16kΩ Resistor element 123 75kΩ Capacitor 114 10nF Capacitor 115 10nF Capacitor 124 10nF Capacitor 125 10nF

[0035] In this embodiment configured as described above, the filter section 110 has a center frequency f1 of 511 Hz, a Q value Q1 of 1.6, and the filter section 120 has a center frequency f2 of 903 Hz, a Q value Q2 of 2.1. f2 / f1 is 1.76, and Q2 / Q1 is 1.31, which are within the ranges set in the embodiment. In addition, the inventors have confirmed that the bandpass filter of this embodiment provides a gradual increase region along the A-characteristic in the transmission range from 600 Hz to 800 Hz. [Explanation of symbols]

[0036] 1, 2, 100... Bandpass filter (BPF) 11, 12, 110, 120... Filter section 13...Sound collecting section 101 Noise Detection System 111, 112, 113, 121, 122, 123... Resistance elements 114, 115, 124, 125... Capacitive elements 116, 126... operational amplifiers

Claims

1. A bandpass filter that attenuates or blocks a portion of the sound that is the input signal, a first filter section having a first bandwidth and a first Q value, and a second filter section having a second bandwidth that partially overlaps the first bandwidth and a second Q value different from the first Q value, A bandpass filter, wherein frequency characteristics indicating the relationship between the frequency and gain of sound transmitted through the first filter section and the second filter section include an ascending section in which the gain ascends with a first slope, a descending section in which the gain descends with a second slope, and a gradually increasing region between the ascending section and the descending section in which the gain continuously and gradually increases with a slope smaller than the first slope and the second slope as the frequency becomes higher.

2. 2. The bandpass filter according to claim 1, wherein a ratio f2 / f1 of a center frequency f2 of the second filter section to a center frequency f1 of the first filter section is 1.4 or more and 1.8 or less, and a ratio Q2 / Q1 of a Q value Q2 of the second filter section to a Q value Q1 of the first filter section is 1.1 or more and 1.5 or less.

3. The bandpass filter according to claim 1 , wherein the frequency characteristic of the gradually increasing region is along an A-weighting curve.

4. The bandpass filter according to claim 1 , wherein a power source that supplies power to the first filter section and the second filter section includes a vibration power generating element.

5. A sound collection unit; A bandpass filter according to any one of claims 1 to 4, which filters the sound collected by the sound collecting unit; a noise detection system.

6. The noise detection system according to claim 5 , wherein the sound collector includes a sound sensor that detects the sound, the sound sensor including a vibration power generation element.

Images