JPH10267742A - Tone quality evaluating device and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To evaluate only a roaring sound recognized as a low frequency beat from among noises composed of many sounds having different tone qualities. SOLUTION: A noise signal which is obtained through a microphone 1, an acoustic sense compensater 2, an analog/digital converter 3 and a data memory 4 is inputted into a feature quantity computing device 5. In a peak detecting section 51, a sound pressure detecting section 512 and a peak judging section 513 detect the peak value of a sound pressure level from a relation between a frequency and the sound pressure level at a time when a frequency analyzing section 511 analyzes the frequency, and a pure sound being the primary factor of a beat is extracted. In a power spectrum mean value computing section 52, an effective value computing section 521 obtains a sound pressure level wave form, and a power spectrum computing section 522 analyzes the frequency, and a weight function overlaping section 523 gives importance to human sensitivity, and a mean value computing section 524 obtains the mean value of a power spectrum, and the low frequency sound pressure fluctuation of the beat is thereby extracted. A noisy degree estimation value computing section 53 estimates a noisy degree from the above extracted value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a sound quality evaluation device and a sound quality evaluation method, and more particularly to a sound quality evaluation device and a sound quality evaluation method for evaluating a sound recognized as a low frequency beat having a peak in frequency.

[0002]

2. Description of the Related Art In recent years, from the viewpoint of environmental friendliness, there has been an increasing interest in noise problems, and there has been an increasing demand for noise reduction in office equipment. Conventionally, an equivalent noise level (J
IS Z8731) is generally used. However, it is known that the equivalent noise level does not correlate well with the psychological loudness of noise generated from office equipment such as copiers and printers.

[0003] This is because when a person evaluates the annoyance of noise, the annoyance is determined for each type of sound included in the noise, not for the overall loudness. The type of sound is, for example, a low-frequency heavy sound, a high-frequency high-pitched sound, or a sound generated in an impact. Therefore, conventionally, several sound quality evaluation methods have been proposed for monotonous ones such as vehicle interior noise and air conditioner noise.

However, noises generated from office equipment such as copiers and printers are composed of noises of many timbres due to the complexity of the mechanism, and it is difficult to evaluate them with one physical quantity of the whole size. Therefore, it is necessary to decompose the sounds from which various noises are superimposed to obtain an evaluation scale that matches the psychological annoyance of each sound.

Therefore, the noise of typical copiers and printers was analyzed, and the individual constituent sounds constituting these noises were classified according to audibly recognizable timbres, and were extracted in terms of artificial sounds. The sounds extracted here are "go sound", which is low frequency random noise composed of aerodynamic noise caused by exhaust from fans, etc., "shear sound", which is high frequency random noise caused by paper rubbing, and movement of the scanner for reading documents. "Win sound" of instantaneous pure sound, "Keen sound" of pure sound due to high-speed rotation of scanner motors and electromagnetic waves, "Won won sound" consisting of pure sounds with peaks at multiple frequencies close to each other due to the beat of the drive system And "click noise" which is an impact noise caused by the paper transport system. These constituent sounds will be described below with reference to the drawings.

FIGS. 9A and 9B show typical noise waveforms during the operation of different types of copiers and printers. FIGS. 9A and 9B show changes in sound pressure level on a time axis.
(C) shows the distribution of the sound pressure level on the frequency axis.

The go sound expressed by the onomatopoeia is shown in FIG.
Is a sound distributed in a frequency range of about 100 Hz to 5 kHz indicated by oblique lines, and is a sound that is perceived as a low-frequency heavy sound. Fig. 9 (C)
Is a sound distributed in a frequency region of about 5 kHz or more, which is indicated by shading, and has a small sound pressure level as compared with the go sound, but is easily recognizable and has a harsh sound. The win sound is a sound of a periodically generated portion shown by shading in FIG. 9A and has a short generation time but a large instantaneous sound pressure level. The keen sound is a continuously generated pure sound indicated by an asterisk in FIG. 9C, and is a sound that is easily recognized when it greatly protrudes from the sound pressure level of the surrounding frequency components. The won-won sound is shown in FIG.
The sound indicated by oblique lines in (B) is a sound pressure level amplitude modulated wave that can be recognized as a low frequency beat. The rattle is an impact sound having an instantaneous peak of the sound pressure level in a portion circled in FIGS. 9A and 9B, and is recognized because the instantaneous change in the sound pressure level is large. It is a sound that is easy to do.

[0008] The present invention relates to the extraction and evaluation of won-won sounds, among these constituent sounds. This won-won sound is a so-called “beat” sound in which a sound pressure of several dB to several tens dB changes with a fluctuation frequency of several Hz.
This is one of the noises generated by copying machines and printers that are particularly annoying and offensive. It is becoming indispensable to accurately evaluate the won-won sound in order to evaluate the psychological annoyance given to a person by the noise of the copying machine and the printer. Japanese Patent Application Laid-Open Nos. Hei 6-117912, Hei 3-18725, Hei 3-187
No. 26, and the like.

In the evaluation method described in Japanese Patent Application Laid-Open No. HEI 6-117912, the amount of change in sound pressure is obtained for each band decomposed by a critical band filter, and the fluctuation of the entire noise is calculated.

[0010] The evaluation of beat sounds is disclosed in
JP-A-18725 and JP-A-3-18726. In these publications, a sensor attached to a case of an electromagnetic switch senses a vibration that causes a beat sound and compares the sensor with a reference value to determine a pass / fail.

[0011]

However, Japanese Patent Application Laid-Open No.
In the case of the evaluation described in Japanese Patent No. 17912, assuming that the cause of the fluctuating sound is a beat sound due to interference between pure sounds of adjacent frequencies, the fluctuating sound can be correctly evaluated if the pure sound extends over two bands. It has the disadvantage that it cannot be done. Furthermore, the target fluctuating sound is a synthetic sound in which various fluctuating sounds are intricately intertwined, such as the exhaust noise of a car,
The fluctuating frequency around 30 Hz is considered to be the sound that fluctuates the most, but the won-won sound targeted by the noise of the copier / printer is mainly caused by a beat sound generated by pure sounds at adjacent frequencies interfering with each other. In addition, since the frequency at which the user feels uncomfortable is also 0.5 to 1.0 Hz, it is not suitable for evaluating these sounds.

In the case of beat sound evaluation described in JP-A-3-18725 and JP-A-3-18726, a pass / fail judgment is made simply based on the magnitude of the beat sound.
It cannot be said that it represents the annoyance that people feel. Furthermore, since the won-won sound is composed of unpleasant sounds as a result of complicated intertwining of noises generated from a plurality of noise sources, there is a problem that there is little advantage in evaluating the noise of a single component.

The present invention has been made in view of the above points, and recognizes a low-frequency beat from a noise composed of sounds of many timbres, in which the loudness of the sound changes periodically with time. It is an object of the present invention to provide a sound quality evaluation device and a sound quality evaluation method capable of evaluating only a won-won sound which is a very unpleasant sound.

[0014]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention provides a sound quality evaluation apparatus for evaluating a sound recognized as a low-frequency beat among constituent sounds constituting noise. Evaluation target sound acquisition means for collecting a sound to be converted into an electric signal, and peak detection means for detecting a peak value of a sound pressure level from a relationship between a frequency and a sound pressure level when the electric signal is subjected to frequency analysis. A power spectrum average value calculating means for calculating an average value of a power spectrum of an effective value waveform of the electric signal.

According to such a sound quality evaluation device, the peak detection means detects the peak value of the sound pressure level from the electric signal obtained by the evaluation target sound obtaining means. The won-won sound is a low-frequency beat generated by pure sounds of adjacent frequencies interfering with each other. Therefore, the peak value that is the maximum value of the pure sound is obtained as one index for evaluation. At the same time, the power spectrum average value calculation means obtains the average value of the power spectrum in the low frequency region where the low frequency beat is generated as one index for evaluation. These indices extract the physical characteristics of the won-won sound, which makes it possible to evaluate the psychological annoyance of the won-won sound with a simple configuration without using many frequency band filters. Become.

Further, according to the present invention, in a sound quality evaluation method for evaluating a sound recognized as a low-frequency beat among constituent sounds constituting noise, a sound to be evaluated is collected and an electric signal is obtained. The peak value of the sound pressure level is detected in the relationship between the frequency and the sound pressure level when the electric signal is subjected to frequency analysis, and the average value of the power spectrum of the effective value waveform of the electric signal is calculated and detected. A sound quality evaluation method comprising calculating a physical quantity of a sound recognized as a low frequency beat from the peak value and the calculated average value of the power spectrum.

In this sound quality evaluation method, the peak value of the sound pressure level of the sound to be evaluated and the average value of the power spectrum are obtained in parallel, so that the low-frequency sound pressure, which is a physical characteristic of the beat sound, is obtained. The average value of the fluctuation and the peak value of the pure tone having a close frequency which is a physical characteristic of the factor of the sound pressure fluctuation are obtained, and only the won-won sound is extracted. By obtaining a physical quantity corresponding to the psychological loudness of the won won sound from the average value and the peak value, there is no influence of the constituent sounds other than the won won sound of the noise, and the loudness of the won won sound to a person is quantitatively determined. Can be evaluated.

[0018]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing a configuration example of a sound quality evaluation device of the present invention. In this figure, the sound quality evaluation device includes a microphone 1, an audibility corrector 2, an A / D (anal
An og-to-digital converter 3, a data storage 4, and a feature value calculating device 5 are provided.

The microphone 1 collects noise generated from a copying machine or a printer and converts the noise into an electric signal. The output of the microphone 1 is connected to the input of the audibility corrector 2. The audibility corrector 2 performs audibility correction on the noise signal converted into the electric signal, and is constituted by, for example, a frequency filter imitating a frequency characteristic in audibility. Here, the A-characteristic audibility correction commonly used in noise measurement was performed. The output of the auditory sensation corrector 2 is connected to an A / D converter 3 for converting the noise signal subjected to the audibility correction into a digital signal.
This is because converting the sound signal into a digital signal keeps the measured data constant, increasing the reliability of the data, and improving the evaluation accuracy by performing calculations based on the digital signal. It is. The output of the A / D converter 3 is connected to the data storage 4. Examples of the data storage 4 include DAT (Digital Audio Tape) and MD (Mini Disc).
), HDD connected to personal computer, etc.
(HardDisk Drive) and magneto-optical recording devices. Then, the output of the data storage unit 4 is connected to the feature value calculation device 5.

The characteristic amount calculating device 5 analyzes the characteristics of the won won sound, calculates various physical quantities, and calculates the annoyance evaluation value of the won won sound by integrating them. An acoustic analysis device capable of performing editing on a time axis waveform or a frequency waveform and calculating a physical quantity such as an equivalent noise level or a loudness level, and the like, includes a peak detection unit 51, a power spectrum average value calculation unit 52, And an annoyance evaluation value calculation unit 53.

The peak detecting section 51 is connected to receive a noise signal converted into a digital signal from the data storage section 4, and the frequency analyzing section 511.
And a peak determination unit 513 that receives the sound pressure detected by the sound pressure detection unit 512 and detects the peak of the sound pressure detected by the sound pressure detection unit 512. You.

The power spectrum average value calculation section 52 is connected to receive a noise signal converted into a digital signal from the data storage 4 and receives an output of the effective value calculation section 521. , A weighting function superimposing unit 523 that receives the fluctuation frequency calculated by the power spectrum calculating unit 522 and weights the frequency, and a weighting function superimposing unit 5
Average calculation unit 534 connected to receive the output of
It is composed of

The annoyance evaluation value calculation section 53 receives the physical indices output from the peak detection section 51 and the power spectrum average value calculation section 52 to calculate annoyance evaluation, and the evaluation value calculation section 531 And a physical quantity display section 532 for displaying a physical quantity such as an annoyance evaluation value calculated in the above.

Next, the operation of the sound quality evaluation device will be described. FIG. 2 is a diagram illustrating an example of a sound pressure waveform of noise collected by a microphone. In this figure, the horizontal axis represents time, and the vertical axis represents sound pressure. Noise generated from a copying machine or a printer is observed in a sound pressure waveform as illustrated in FIG. This noise is collected by the microphone 1 and output as an analog electric signal. This electrical signal
First, it is input to the audibility corrector 2 to perform the A-characteristic audibility correction, subsequently input to the A / D converter 3 and converted into a digital signal, and then input to the data storage 4 and temporarily stored therein. Will be saved.

Next, the digital signal of the noise read from the data storage 4 is input to the feature value calculating device 5, and the peak detecting portion 51 and the power spectrum average value calculating portion 5 are provided.
2 are processed in parallel.

First, in the peak detecting section 51, the digital signal of the noise whose audibility has been corrected is subjected to frequency analysis in the frequency analyzing section 511. By this frequency analysis, a change in the sound pressure level with respect to the frequency is output. Frequency analysis unit 51
The frequency analysis result output from 1 is input to the sound pressure detection unit 512. At this time, data input to the sound pressure detection unit 512 is limited to a frequency range of 100 to 500 Hz. The reason for this is that in the frequency range of 500 Hz or more, a person recognizes not a won-won sound but a keen sound, and the degree of subjective loudness differs. Also, the area of the peak to be detected is 500 Hz.
By limiting to the following, the amount of data input to the sound pressure detection unit 512 can be significantly reduced, so that there is also an advantage that processing after the sound pressure detection unit 512 can be performed efficiently.

The sound pressure detecting section 512 receives the data in the frequency range of 100 to 500 Hz and detects the maximum sound pressure level every 10 Hz. The data of the maximum sound pressure level detected in this way is input to the peak determination
The peak in the frequency range of 0 to 500 Hz is extracted. The peak determination unit 513 regards a case where the difference between the maximum sound pressure level value and the sound pressure level value of the frequency at 10 Hz before and after the maximum sound pressure frequency is 3 dB or more is regarded as a peak, and the difference of 3 dB or more. When there is no sound pressure detection unit 512, the second largest sound pressure level is detected by the sound pressure detection unit 512, and it is similarly determined whether or not the peak is reached. This peak determination unit 51
FIG. 3 shows the result of the determination in No. 3.

FIG. 3 is a diagram showing an output example of the peak detecting section. In this figure, the horizontal axis represents frequency, and the vertical axis represents sound pressure level. From the relationship between the frequency and the sound pressure level when the frequency is analyzed by the frequency analysis unit 511, 100 to 5
The sound pressure level protruding by 3 dB or more from the sound pressure level values of the preceding and following frequencies in the frequency range of 00 Hz is determined as a peak, and the peak value is output from the peak detection unit 51 as one physical index representing a won-won sound. Then, it is sent to the noisy evaluation value calculation section 53.

On the other hand, the power spectrum average value calculation section 52
Then, the digital signal of the noise whose audibility has been corrected is input to the effective value calculation unit 521. The effective value calculation unit 521 calculates the effective value of the sound pressure of the noise every 10 milliseconds, for example, and converts it into a sound pressure level waveform. FIG. 4 shows a calculation result by the effective value calculation unit 521.

FIG. 4 is a diagram showing an example of the sound pressure level obtained by calculating the effective value of the sound pressure corrected for the auditory sense. In this figure, the horizontal axis represents time, and the vertical axis represents sound pressure level. From the change in the sound pressure level, it can be seen that the sound pressure changes in a cycle of approximately 1 Hz. The sound pressure level waveform obtained by the calculation of the effective value is input to the power spectrum calculation unit 522, and is subjected to frequency analysis. FIG. 5 shows the result of frequency analysis performed by the power spectrum calculator 522.

FIG. 5 is a diagram showing a power spectrum of a fluctuating frequency extracted by frequency analysis of a sound pressure level waveform. In this figure, the horizontal axis represents the fluctuation frequency, and the vertical axis represents the power. The power spectrum calculation unit 522 performs calculation except for a DC component and a fluctuating frequency region exceeding 5 Hz. This is because the DC component of the power spectrum is irrelevant to the sound pressure fluctuation of the won-won sound, and is not suitable for use in calculating an average value because the value is extremely large. Further, since the region exceeding 5 Hz is not recognized as a won-won sound, it is excluded from the calculation. In this way, the power spectrum of the fluctuating frequency obtained by the power spectrum calculation unit 522 is input to the weight function superimposition unit 523, where the weight function is superimposed. FIG. 6 shows an example of the weight function superimposed by the weight function superimposing unit 523.

FIG. 6 is a diagram showing an example of the weight function. In this figure, the horizontal axis represents the fluctuation frequency, and the vertical axis represents the weight. The weighting function represents the sensitivity that a person perceives with respect to the fluctuating frequency with respect to the power spectrum, and is obtained by analyzing the won-won sound of a typical copying machine or printer. According to this analysis result, it is understood that the vicinity of 0.5 to 1 Hz is felt most uncomfortable, and that the frequency is higher or lower, the loudness is gradually reduced. As shown in the drawing, 0.88 Hz is the highest sensitivity and unpleasant frequency, and the weighting function Y at this time is

[0033]

Y = 0.5 [1-cos {2π (x / 5) ^0.4 }] (1) Here, x is a variable frequency, and 0 ≦ x ≦
5

The weighting function superimposing section 523 superimposes a weighting function representing the unpleasantness felt by a person on the frequency of the power spectrum, so that the power spectrum is reduced in a variable frequency region where the user does not feel annoyed. The subjective annoyance received from the sound will be more accurately evaluated.

The power spectrum on which the weighting function is superimposed is input to the average value calculation section 524, and the average value is obtained.
FIG. 5 shows an example of the average value obtained by the average value calculation unit 524. Since FIG. 5 shows the power spectrum before the weight function is superimposed, the average value shown is also the average value of the power spectrum before the weight function is superimposed. Therefore, the average value of the power spectrum on which the weight function is superimposed by the weight function superimposing unit 523 is smaller than the average value illustrated. The average value of this power spectrum is one physical index representing the won-won sound.
It is output from the power spectrum average value calculation unit 52 and sent to the annoyance evaluation value calculation unit 53.

When the loudness evaluation value calculation section 53 receives the peak value of the sound pressure level from the peak detection section 51 and the power spectrum average value from the power spectrum average value calculation section 52, the evaluation value calculation section 531 calculates the evaluation value. calculate. The evaluation value calculation unit 531 calculates the evaluation value as a linear sum of the physical feature amounts of the noise waveform. That is, the loudness evaluation value expressing the won-won sound in a subjective amount is:

[0037]

[Mathematical formula-see original document] a (peak value) + b (power spectrum average value) + c (2) Here, a and b are coefficients and c is a constant.

The annoyance evaluation value calculated by the evaluation value calculation section 531 is sent to the physical quantity display section 532 and displayed. FIG. 7 shows a display example of the physical quantity display section 532.
FIG. 7 is a diagram illustrating a display example of the annoyance evaluation value output from the evaluation value calculation unit. According to the display example of the physical quantity display unit 532, the loudness evaluation value is displayed together with the peak value of the sound pressure level and the average value of the power spectrum, which are the physical features of the noise waveform to be obtained.

Next, in order to confirm the effects of the present invention,
The loudness due to the won-won sound was evaluated by sensory evaluation. FIG. 8 shows the result of this evaluation. FIG. 8 is a diagram showing the relationship between the loudness evaluation value and the loudness sensory value due to the won-won sound. Here, a sensory evaluation experiment was performed using seven types of printers and copiers. The evaluation was performed with a score of 0 not giving a won-won sound at all and a score of 10 given a loud and unbearable sound. The number of persons evaluated was a total of 20 men and women who had no abnormal hearing. According to the evaluation results shown in the drawing, it is understood that the annoyance evaluation value obtained by the present invention has a high correlation with the result of the annoyance evaluation by the sensory evaluation.

Although the preferred embodiment of the present invention has been described in detail, the present invention is not limited to this particular embodiment. For example, in the embodiment, a noise waveform of a copying machine or the like is converted to a microphone 1, an audibility corrector 2,
Although acquired by the A / D converter 3 and the data storage 4,
The output signal may be input to the feature value calculation device 5 using a sound level meter.

The peak detecting section 51 detects the peak value from the relationship between the frequency and the sound pressure level when the frequency of the electric signal is analyzed, but converts the analysis result obtained by the frequency analyzer 511 into an image signal. After conversion, the sound pressure level at each frequency is read from the image signal, for example, the amount of change from an adjacent frequency is calculated, and when a certain threshold value is exceeded, a calculation is performed to determine that a frequency peak exists. It can be configured by software and a computer that executes the software.

As another method of detecting a peak value in the sound pressure detecting section 512 and the peak determining section 513, the average value of sound pressure levels in a certain band is compared with the sound pressure levels of individual frequencies in the band. If the difference exceeds a certain threshold value, it is determined that there is a frequency peak, or the derivative of the frequency is calculated, and it is convex upward and there is a change amount from the starting point of the convex part to the vertex. If the threshold value is exceeded, it may be determined that there is a frequency peak.

A power spectrum average value calculating section 5 for calculating the average value of the power spectrum of the effective value of the electric signal.
2 has the weight function superimposing section 523 in the preferred embodiment, but the weight function superimposing section 523 is not always necessary and can be omitted. in this case,
The value of the coefficient b that is multiplied by the average power spectrum value when obtaining the noisy evaluation value is changed accordingly.

[0044]

As described above, according to the present invention, the maximum peak value of the sound pressure level is detected based on the relationship between the frequency and the sound pressure level when the frequency of the electric signal corresponding to the sound to be evaluated is analyzed. And a power spectrum average value calculating means for calculating the average value of the power spectrum of the effective value waveform of the electric signal. Thereby, the power spectrum average value calculation means extracts the physical characteristics of the so-called beat sound of the sound pressure fluctuation of the low frequency,
In practice, the peak detection means causes a beat sound due to interference.In practice, it becomes easy to extract the physical characteristics of two peaks having close frequencies, and the correspondence between these sounds and the psychological loudness of the sounds is difficult. By being able to do so, it is possible to provide a sound quality evaluation device that is compatible with psychological annoyance.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration example of a sound quality evaluation device of the present invention.

FIG. 2 is a diagram illustrating an example of a sound pressure waveform of noise collected by a microphone.

FIG. 3 is a diagram illustrating an output example of a peak detection unit.

FIG. 4 is a diagram illustrating an example of a sound pressure level obtained by calculating an effective value of a sound pressure corrected for auditory sense.

FIG. 5 is a diagram showing a power spectrum of a fluctuating frequency extracted by frequency analysis of a sound pressure level waveform.

FIG. 6 is a diagram illustrating an example of a weight function.

FIG. 7 is a diagram illustrating a display example of an annoyance evaluation value output from an evaluation value calculation unit.

FIG. 8 is a diagram showing a relationship between a loudness evaluation value and a loudness sensory value by a won-won sound.

FIGS. 9A and 9B show typical noise waveforms during the operation of copiers and printers of different models, wherein FIGS.
Shows the change of the sound pressure level on the time axis, and (C) shows the distribution of the sound pressure level on the frequency axis.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Microphone 2 Hearing corrector 3 A / D converter 4 Data storage device 5 Feature amount calculating device 51 Peak detecting unit 52 Power spectrum average value calculating unit 53 Noisy evaluation value calculating unit 511 Frequency analyzing unit 512 Sound pressure detecting unit 513 Peak judgment Unit 521 Effective value calculation unit 522 Power spectrum calculation unit 523 Weight function superposition unit 534 Average value calculation unit 531 Evaluation value calculation unit 532 Physical quantity display unit

Claims (13)

[Claims] 1. A sound quality evaluation device for evaluating a sound recognized as a low frequency beat in constituent sounds constituting noise, wherein a sound to be evaluated is collected and converted into an electric signal. Acquisition means, peak detection means for detecting a peak value of a sound pressure level from a relationship between a frequency and a sound pressure level when the electric signal is subjected to frequency analysis, and an average value of a power spectrum of an effective value waveform of the electric signal. And a power spectrum average value calculating means for calculating. 2. The sound quality evaluation device according to claim 1, wherein said peak detecting means sets a frequency range of a detected sound pressure peak to 500 Hz or less. 3. A peak detecting means, comprising: a frequency analyzing section for frequency-analyzing the electric signal; and a sound for detecting a sound pressure level based on a relationship between a frequency and a sound pressure level when the frequency analyzing section performs frequency analysis. The sound quality evaluation device according to claim 1, further comprising: a pressure detection unit; and a peak determination unit that extracts a peak value from the detected sound pressure level. 4. The power spectrum average value calculating means calculates an effective value of the electric signal to obtain a sound pressure level waveform, and a frequency analysis of the sound pressure level waveform to calculate the sound pressure level. The sound quality evaluation device according to claim 1, further comprising: a power spectrum calculation unit that calculates a power spectrum of the waveform; and an average value calculation unit that calculates an average value of the power spectrum. 5. The sound quality evaluation device according to claim 4, wherein the average value calculation unit excludes a DC component from a target of the average value calculation of the power spectrum. 6. The sound quality evaluation device according to claim 4, wherein the average value calculation section sets a power spectrum in a frequency domain of 5 Hz or less as a target of the average value calculation. 7. The power spectrum average value calculating unit includes a weight function superimposing unit that superimposes a weight function representing a subjective discomfort level of a person on a frequency on the power spectrum output from the power spectrum computing unit. The sound quality evaluation device according to claim 4, further comprising: 8. An annoyance evaluation value calculating means for calculating an annoyance evaluation value based on two physical indices output from the peak detection means and the power spectrum average value calculation means. The sound quality evaluation device according to claim 1. 9. A sound quality evaluation method for evaluating a sound recognized as a low frequency beat in constituent sounds constituting noise, wherein a sound to be evaluated is collected and converted into an electric signal. The peak value of the sound pressure level is detected in the relationship between the frequency and the sound pressure level when the signal is subjected to the frequency analysis, and the average value of the power spectrum of the effective value waveform of the electric signal is calculated. Calculating a physical quantity of a sound recognized as a low-frequency beat from the average value of the obtained power spectrum. 10. The step of detecting the peak value,
10. The sound quality evaluation method according to claim 9, wherein a frequency range in which a peak value is detected is set to 500 Hz or less. 11. The sound quality evaluation method according to claim 9, wherein in the step of calculating the average value of the power spectrum, the average value is calculated excluding the power spectrum of the DC component. 12. The method according to claim 9, wherein the step of calculating the average value of the power spectrum calculates the average value of the power spectrum in a frequency region of 5 Hz or less.
The described sound quality evaluation method. 13. The step of calculating an average value of the power spectrum includes performing a calculation of the average value after superimposing a weighting function representing a subjective discomfort level of a person on a frequency on the power spectrum. The sound quality evaluation method according to claim 9, wherein: