JP3148466B2 - Device for discriminating between helicopter sound and vehicle sound - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a helicopter sound detected in a monitoring area or a sound generated by a vehicle (hereinafter referred to as a sound source) (hereinafter referred to as "detected sound"). This is related to a device for determining whether the

[0002]

2. Description of the Related Art Conventionally, this type of apparatus has been disclosed in Japanese Patent Application No. Hei.
-127344, which focuses on the steepness at each frequency point of the spectrum of a helicopter sound. Within a predetermined time, if the steep spectrum exceeds a majority, it is determined that the sound is a helicopter sound. If the spectrum was low or absent, it was determined to be vehicle sound.

FIG. 5 is a block diagram of a conventional detection sound discriminating apparatus. In the following description of the operation of the apparatus and the like, description of general signal amplification / modulation / transmission methods and the like is omitted. In FIG. 5, a vibration wave generated from a sound source 1 propagates in the atmosphere, is detected as a detection sound S1 by an acoustic sensor 2, is subjected to acoustoelectric conversion in the acoustic sensor 2, and is output as an analog acoustoelectric signal S2. . The acoustoelectric signal S2 is amplified by the amplifier 3 to a required level to become an analog detection signal S3. This detection signal S3 is a spectrum composed of many harmonics. Of the spectrums, a predetermined number of extracted spectrums S4 are extracted by the spectrum extracting circuit 4 in order from the highest level in the spectrum whose level is equal to or higher than the predetermined level within a predetermined time.

From each peak point of the extracted spectrum S4, a bandwidth (hereinafter referred to as a peak value width) .DELTA.f at a position lowered by a predetermined level is obtained. If the peak value width .DELTA.f is small, the steepness is high. Peak value width Δf
Is large, it is assumed that the steepness is low. Therefore, for each of the extracted spectra S4, the steepness calculation circuit 5 obtains the steepness data S5, and the discrimination circuit 6 discriminates the one with a high steepness as a helicopter sound and the one with a low steepness as a vehicle sound. Those with a medium steepness are treated as uncertain.

When the spectrum identified as the helicopter sound exceeds the majority in the entire extracted spectrum S4, it is determined that the detection sound S1 is the helicopter sound, and when the spectrum identified as the vehicle sound exceeds the majority. Indicates that the detected sound S2 is a vehicle sound, and if none of the above, the sound is uncertain, and the final identification data S
6 and the identification result is displayed on the display device 7. Description of the discrimination value of the steepness is omitted.

[0006]

However, in the above-described conventional apparatus, the steepness of the extracted spectrum cannot be clearly determined depending on environmental conditions or the type of vehicle, and the helicopter sound, In many cases, the sound is indeterminate because it cannot be distinguished from the vehicle sound, and is not technically satisfactory.

According to the present invention, the spectrum of a helicopter sound has a high steepness at each frequency point of the spectrum in order to improve the probability of being identified as a helicopter sound or the probability of being identified as a vehicle sound. Not only that, the time variation of the frequency is small, but on the other hand, the spectrum of the vehicle sound has a low steepness at each frequency point of the spectrum, and attention is also paid to the fact that the time variation is large, and at each frequency point of the spectrum, An object of the present invention is to provide a device for discriminating between a helicopter sound and a vehicle sound that can identify a detected sound with a high probability by combining a plurality of identification items of steepness and temporal variation of each frequency of a spectrum. And

[0008]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides an apparatus for identifying whether a detection sound detected in a monitoring area is a helicopter sound or a vehicle sound. At the spectrum level of the detected sound within the time, a steepness calculating means for calculating the steepness at each frequency point of the spectrum, and calculating the temporal variation of the frequency at each frequency point of the spectrum, for a spectrum above a predetermined level. Frequency variation calculating means, first identifying means for determining whether the calculated steepness is greater than or less than a reference value, and second identifying means for determining whether the calculated temporal variation is greater than or less than the reference value. Identification means, and identifies whether the detection sound is a helicopter sound or a vehicle sound based on output signals from the first identification means and the second identification means. It is obtained by the provided and comprehensive identification means.

[0009]

According to the present invention, there is provided an apparatus for identifying whether a detection sound detected in a monitoring area is a helicopter sound or a vehicle sound, wherein the detection level within a predetermined time is equal to a predetermined spectrum level. A steepness calculating means for calculating a steepness at each frequency point of the spectrum for a spectrum having a level or higher is provided, and a first steepness is calculated from the calculated steepness.
A frequency variation calculating means for calculating a temporal variation of each frequency of the spectrum is provided, and a second identifying means is used to perform the identification based on the calculated frequency variation. By providing the comprehensive identification means capable of comprehensively identifying the identification result by the identification means and the identification result by the second identification means, it is possible to obtain an identification result with a higher probability than in the past.

[0010] In particular, when it is difficult to make a distinction due to the steepness of the extracted spectrum, a high-precision discrimination can be performed by checking the amount of change in the frequency.

[0011]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a block diagram of a device for discriminating between a helicopter sound and a vehicle sound according to an embodiment of the present invention. The vibration wave generated from the sound source 11 propagates in the atmosphere, is detected by the acoustic sensor 12 as a detection sound S11, is subjected to acoustoelectric conversion in the acoustic sensor 12, and is output as an analog acoustoelectric signal S12. The acoustoelectric signal S12 is amplified by the amplifier 13 to a required level to become a detection signal S13.

The detection signal S13 is a spectrum composed of many harmonics, and this spectrum is input to a spectrum extraction circuit 14, and among the spectrum,
Within a predetermined period of time, a predetermined number of extracted spectra S14 are extracted by the spectrum extraction circuit 14 in order from a higher level, with respect to the spectrum whose spectrum level is equal to or higher than the predetermined level. The description of the predetermined time, the predetermined level, and the predetermined number is omitted.

FIG. 2 is a block diagram of a spectrum extracting circuit. The spectrum extracting circuit 14 includes a high-pass filter (HPF) 14-1 and a low-pass filter (LP).
F) 14-2, analog / digital conversion circuit 14-3,
It comprises a linear prediction operation circuit 14-4. Here, the high-pass filter (HPF) 14-1 removes an external noise (wind noise) component, and the low-pass filter (LPF) 14-2
When the helicopter sound and the vehicle sound are present, the range is limited to the spectrum where the level increases.

The output signal S14-1 from the high-pass filter (HPF) 14-1 is a low-pass filter (LPF).
14-2, and input to a low-pass filter (LPF) 14
-2 is input to an analog / digital conversion circuit 14-3, quantized, converted into a digital audio signal S14-3, and output to a linear prediction operation circuit 14-2.
-4 is input.

FIG. 3 is a conceptual diagram of a spectrum of a detection signal according to the present invention. FIG. 3A shows an example of a spectrum of a helicopter sound, and FIG. 3B shows an example of a spectrum of a vehicle sound.
Considering the spectrum obtained by actually measuring the helicopter sound, the helicopter sound is mainly composed of a high-level rotor sound, and the component frequency of the rotor sound is, as shown in FIG. It is a spectrum. On the other hand, when examining the spectrum obtained by actually measuring the vehicle sound, the vehicle sound mainly includes an engine sound component and a caterpillar sound component, and as shown in FIG. Low.

As apparent from the comparison between FIGS. 3A and 3B, the peak value width Δf ₁ of the spectrum of the helicopter sound is always smaller than the peak value width Δf ₂ of the spectrum of the vehicle sound. Further, the peak fluctuation rate of the spectrum of the helicopter sound within a predetermined time is small, and the peak row S _R0W of the spectrum of the helicopter sound is substantially straight. On the other hand, the peak fluctuation rate of the vehicle sound spectrum is large, and the peak train S _R0W of the vehicle sound spectrum is large.
Is bent.

Therefore, in this embodiment, utilizing the characteristics of the spectrum of the helicopter sound and the vehicle sound,
The detection sound is identified as a helicopter sound or a vehicle sound. That is, the digital acoustic signal S14-3 is based on the spectrum peak information (line spectrum information).
Extract helicopter sound or vehicle sound. The linear prediction operation circuit 14-4 performs a linear prediction (LPC) operation on the digital acoustic signal S14-3, and inputs an extracted spectrum S14, which is linear prediction coefficient data, to the steepness degree calculation circuit 15 of the spectrum of the detected sound. I do.

The linear prediction operation circuit 14-4 obtains linear prediction coefficient data with higher analysis accuracy in accordance with, for example, the instantaneous maximum entropy method (IMEM method) in which the spectrum estimation accuracy is high and the operation processing time is short. .
The steepness calculation circuit 15 of the spectrum of the detected sound uses the extracted spectrum S14 which is the linear prediction coefficient data from the linear prediction operation circuit 14-4, and uses a normal method (an equation for obtaining a frequency value and a level value from the linear prediction coefficient data). ), The spectrum of the digital acoustic signal S14-3 is obtained.

The steepness calculating circuit 15 calculates the frequency, peak value width, and level of each peak in the spectrum. Here, it is assumed that the peak is larger than, for example, a first reference level obtained by adding a predetermined level to the average level of the nearby frequency component or the frequency component of the entire band limited. In addition, the peak value width is, for example, a bandwidth in which a second reference level obtained by adding a predetermined level to an average level of nearby frequency components or frequency components in the entire band-limited region crosses the peak shape.

In the case of this embodiment, the steepness of the peak which determines whether or not the spectrum is a line spectrum is represented by the peak value width thus quantified. That is, the smaller the value of the peak value width, the steeper the peak (the peak can be regarded as a line spectrum). Next, the frequency variation calculating circuit 17 measures the frequency of each peak in the spectrum at each time t, and subtracts the frequency from the previously measured peak frequency to obtain the variation (in a predetermined time period). Divide the sum of the variations by the number of measurements to find the average).

The extracted spectrum S14 is input to a steepness calculating circuit 15 and a frequency fluctuation calculating circuit 17, and the steepness calculating circuit 15 outputs steepness data S15.
The output is input to the first identification circuit 16, and the first identification circuit 16 outputs an identification signal S16. That is, the first identification circuit 16 compares the peak value width Δf with the reference value, for example, when the peak value width Δf is smaller than 0.4 Hz [FIG.
(Refer to Δf _{1 of} (a)), assuming that the steepness is high,
“H” is output, and the reference value, for example, the peak value width Δf is
When the frequency is greater than 0.4 Hz [see Δf _{2 in} FIG. 3B]
Outputs "L" as having low steepness, and inputs the signal to the first input terminal of the general identification circuit 19 as the identification signal S16.

In this way, the extracted spectrum S14
By looking at the steepness of the helicopter sound or the vehicle sound, it is possible to judge, but as described above, the steepness of the extracted spectrum depends on the environmental conditions and the like, or depending on the vehicle type. In many cases, the sound cannot be clearly determined, and cannot be clearly discriminated from a helicopter sound or a vehicle sound, and thus is uncertain. For example, due to the blunting of the leading end of the spectrum, the aggregation of a plurality of spectra at the leading end of the spectrum, and the like, there is a case where the judgment is lost only with the steepness of the extracted spectrum S14.

On the other hand, a frequency fluctuation amount S17 is output from the frequency fluctuation calculating circuit 17, and this frequency fluctuation amount S17 is input to a second identification circuit 18, and an identification signal S18 is output from the second identification circuit 18. . That is, the first discrimination circuit 16 compares the frequency variation Δf _D with the reference value, and the frequency variation Δf _D is smaller than 1.0 Hz. In other words, the spectrum peak row [S _{pROW in} FIG. If it is above, "L" is output assuming that the frequency variation is small,
If the frequency variation Δf _D is greater than 1.0 Hz, in other words, if the peak row of the spectrum (S _{pROW in} FIG. _3B ) is bent, “H” is output assuming that the frequency variation is small. , As an identification signal S18 to the second input terminal of the general identification circuit 19.

Therefore, the first input terminal of the general identification circuit 19 is "H", that is, the steepness of the extracted spectrum S14 is high, and the second input terminal of the general identification circuit 19 is "L", that is, the frequency is low. Is small, it can be determined that the sound is a helicopter sound. Further, the first input terminal of the general identification circuit 19 is “L”, that is, the steepness of the extracted spectrum S14 is low, and the second input terminal of the general identification circuit 19 is “H”, that is, the frequency variation rate. Is large, it can be determined as a vehicle sound.

Further, the first input terminal of the general identification circuit 19 is "H", that is, the steepness of the extracted spectrum S14 is high, and the second input terminal of the general identification circuit 19 is "H", that is, the frequency is high. If the rate of change of is large, it is unknown. However, when the sound source passes right above the acoustic sensor, there is a case where helicopter sound is caused by the Doppler effect, so that the measurement is performed again. In this case, in the conventional case, the sound is determined to be a helicopter sound uniformly, and it cannot be said that the sound is accurate.

The first input terminal of the general identification circuit 19 is "L", that is, the steepness of the extracted spectrum S14 is low, and the second input terminal of the general identification circuit 19 is "L", that is, the frequency is low. If the rate of change is small, it is determined to be unknown. In this case, in the traditional case,
Although the vehicle sound is determined to be uniform, this is an erroneous determination, and it is apparent that it is actually unknown by considering the frequency fluctuation rate.

In particular, in the case where it is difficult to discriminate due to the steepness of the extracted spectrum S14, highly accurate discrimination can be performed by checking the rate of change in frequency. In addition, if the spectrum identified as a helicopter sound exceeds the majority of the entire extracted spectrum S14 within a predetermined time, the detection sound S1
1 is a helicopter sound, and if the spectrum identified as a vehicle sound exceeds a majority, the detection sound S1
1 is determined to be a vehicle sound, and if none of them corresponds, it is uncertain and is output as a final identification signal S19, and the identification result is displayed on the display device 20.

Looking at a specific example of a spectrum, FIG.
As shown in (a), when the spectrum level of the detected sound within a predetermined time is a spectrum above a predetermined level La and the steepness at each frequency point of the spectrum is high, that is, a certain level from the peak point Sp. When the peak value width Δf at the position of the lowered predetermined level La is small,
Alternatively, when the frequency variation at each frequency point in the spectrum is small, that is, when the spectrum row S _ROW is substantially on a straight line, it is determined that the sound is a helicopter sound.

Further, as shown in FIG. 3B, when the spectrum level of the detected sound within a predetermined time is a spectrum above a predetermined level La and the steepness at each frequency point of the spectrum is low, that is, When the peak value width Δf at the position of the predetermined level La which is lower by a certain level from the peak point Sp is large, or when the frequency variation at each frequency point of the spectrum is large, that is, when the spectrum row S _ROW is bent, the helicopter is used. It will be judged as sound.

FIG. 4 shows a comparison between the conventional helicopter sound / vehicle sound discriminating apparatus and the helicopter sound / vehicle sound discriminating apparatus of the present invention. FIG. 4A shows a result of discrimination by a conventional helicopter sound and vehicle sound discriminating apparatus.
FIG. 4B shows the result of the discrimination between the helicopter sound and the vehicle sound according to the present invention. As shown in FIG. 4A, in the conventional case, the measurement items are only the steepness of the peak of the extracted spectrum, and when the steepness is high, the detection sound is uniformly a helicopter sound and the steepness is low. In this case, the vehicle sound is determined uniformly.

On the other hand, in the case of the present invention, in addition to the steepness of the peak of the extracted spectrum, the variation of the frequency of the extracted spectrum within a predetermined time of the peak of the extracted spectrum is taken into consideration as a measurement item. Accordingly, as shown in FIG. 4B, when the steepness of the extracted spectrum is high and the amount of change in the frequency of the extracted spectrum within a predetermined time is small, it is determined that the sound is a helicopter sound, and the steepness of the extracted spectrum is reduced. If the frequency is low and the amount of change in the frequency of the extracted spectrum within a predetermined time is large, it is determined that the sound is a vehicle sound.

Further, the steepness of the extracted spectrum is high,
In addition, when the frequency variation of the extracted spectrum within a predetermined time is large, it is determined that the sound is uncertain or a helicopter sound due to the Doppler effect. In this case, the measurement is performed again. On the other hand, the steepness of the extracted spectrum is low,
In addition, when the amount of change in the frequency of the extracted spectrum within a predetermined time is small, it is not determined as a vehicle sound as in the related art, but is determined to be unknown and unclear.

In the case where it is difficult to determine the discrimination due to the steepness of the extracted spectrum, the discrimination can be performed with high accuracy by checking the rate of change of the frequency.
Note that the present invention is not limited to the above embodiment,
Various modifications are possible based on the spirit of the present invention, and they are not excluded from the scope of the present invention.

[0034]

As described above in detail, according to the present invention, in a device for discriminating whether a detection sound detected in a monitoring area is a helicopter sound or a vehicle sound, A steepness calculating means for calculating a steepness at each frequency point of the spectrum with respect to a spectrum whose spectrum level is equal to or higher than a predetermined level by a detected sound within a time, and is identified by the first identifying means from the calculated steepness. And a frequency variation calculating means for calculating the temporal variation of each frequency of the spectrum is provided. The identification is performed by the second identifying means from the calculated frequency variation, and further, the identification result by the first identifying means is provided. Providing a comprehensive identification means capable of comprehensively identifying the identification result by the second identification means and the identification result by the second identification means, thereby obtaining an identification result with a higher probability than in the past. Door can be.

In particular, when it is difficult to make a distinction due to the steepness of the extracted spectrum, it is possible to perform a highly accurate discrimination by checking the amount of change in the frequency.

[Brief description of the drawings]

FIG. 1 is a block diagram of a device for discriminating between a helicopter sound and a vehicle sound according to an embodiment of the present invention.

FIG. 2 is a block diagram of a spectrum extraction circuit of the device for discriminating between a helicopter sound and a vehicle sound according to an embodiment of the present invention.

FIG. 3 is a conceptual diagram of a spectrum of a detection signal according to the present invention.

FIG. 4 is a diagram showing a determination result between the present invention and a conventional identification device.

FIG. 5 is a block diagram of a conventional detection sound identification device.

[Explanation of symbols]

 Reference Signs List 11 sound source 12 acoustic sensor 13 amplifier 14 spectrum extraction circuit 14-1 high-pass filter (HPF) 14-2 low-pass filter (LPF) 14-3 analog / digital conversion circuit 14-4 linear prediction operation circuit 15 steepness calculation circuit 16 first Identification circuit 17 frequency variation calculation circuit 18 second identification circuit 19 general identification circuit 20 display device S11 detection sound S12 analog acoustic electric signal S13 detection signal S14 extraction spectrum S14-1, S14-2 output signal S14-3 digital sound Signal S15 Steepness data S16, S18 Identification signal S17 Frequency variation S19 Final identification signal

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-59-72028 (JP, A) JP-A-2-275997 (JP, A) JP-A-4-104018 (JP, A) JP-A-5- 79898 (JP, A) JP-A-5-322640 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01H 3/08

Claims (1)

(57) [Claims] An apparatus for identifying whether a detection sound detected in a monitoring area is a helicopter sound or a vehicle sound, comprising: (a) a spectrum level of the detection sound within a predetermined time, which is equal to or higher than a predetermined level; For the spectrum of
A steepness calculating means for calculating a steepness at each frequency point of the spectrum, (b) a frequency fluctuation calculating means for calculating a temporal variation of a frequency at each frequency point of the spectrum, and (c) the calculated steepness is First identification means for identifying whether it is greater than or less than a reference value; (d) second identification means for identifying whether the calculated temporal variation is greater than or less than a reference value; A helicopter sound comprising: a first identification means; and a general identification means for determining whether the detection sound is a helicopter sound or a vehicle sound based on an output signal from the second identification means. For discriminating vehicle sounds.