JP6283890B1 - Noise spectrum distribution detection method and anti-noise volume sound quality control method - Google Patents

Abstract

The volume must be increased in listening in an environment where the noise changes greatly. Depending on the noise, it is necessary to change not only the volume but also the sound quality. Since the reproduced sound corresponding to the increased volume is mixed in the microphone that detects the mixed sound in which the noise and the reproduced sound are mixed, the control system needs to know only the intensity of the noise. A band division filter having the same characteristics is applied to each of a mixed sound signal and a reproduction signal. The maximum value of the time interval from 200 mmsec to 1 second of each filter output is detected. The difference between the mixed sound signal and the reproduction signal and the maximum value of the noise are obtained for each band. A low attack time constant and a high release time constant are applied to the maximum noise level. With the obtained signal for each band, the gain of the reproduced sound is corrected for each band. [Selection] Figure 1

Description

The terms used in the claims are the same in the specification.
Addition means + means addition – means sign-inversion addition.
Stabilization prediction of adaptive filter feedback loop Propagation of filter sound Noise characteristics of subway, aircraft cabins Vehicle structure specifications and road surface conditions and wind noise noise speed of driver seat noise Statistical characteristics of acoustic signals Probability distribution of the sum of two correlated signals Probability distribution of two uncorrelated signals with each other Probability distribution of the sum of signals Relation between auditory noise and signal

Active noise canceller for earphones and headphones Active noise canceller for driver's seat

The present invention relates to a method for detecting a maximum value of a noise signal and correcting a reproduction signal with the maximum value of the noise signal. There are multiple descriptions of how to calculate the correction amount of the playback sound.
In both cases, the magnitude of the noise is calculated from the synthesized sound of the noise and the reproduced sound and the reproduced sound.
There is no description.
There is a description about the analysis of each noise signal frequency band,
Regarding the analysis results, there is no description that the noise level in the same band is calculated from both the noise and the synthesized sound of the noise and the reproduced signal, and that the reproduced sound is controlled for each frequency band based on the result.

The present invention relates to a method for detecting a maximum value of a reproduction signal and correcting the reproduction signal based on the maximum value and the magnitude of the detected noise signal. The present invention relates to a method for controlling the magnitude of a reproduction signal having an inflection in intensity so that the intensity inflection is matched to ambient noise.
There is no description that the noise level is calculated from both the mixed signal of the noise and the reproduced sound and the reproduced signal.

The terms and symbols defined in the claims are the same in the specification.
First, while driving, in the subway commute, and under strong noise, such as in airplane seats, the volume must be increased to listen to music, lectures, and entertainment program playback. Furthermore, not only the volume but also the sound quality needs to be varied in order to achieve the minimum volume. A microphone that detects sound waves in which noise and reproduction sound are mixed contains reproduction sound corresponding to the increased volume, which is an unstable factor. To stabilize the control loop, it is necessary to know the pure noise maximum. Of course, it is necessary to detect a pure noise maximum value for noise-quality control.

Secondly, in recent years, the sound field control technology has been improved, and active noise canceller application products that mute noise in a specific area have come to the market. However, it is difficult to achieve satisfactory performance in a strong noise environment, or it requires a structure and cost that are not practical, and at the product level, it must remain within the possible range due to the relationship between performance, cost, and usability.

Thirdly, if optimal volume sound quality control according to noise is attempted, it is necessary to increase the volume according to the detected noise. You must deduct the volume. Since the relationship between the volume levels of both reproduced sound and noise changes relatively randomly with time, it is difficult to accurately estimate the volume of noise in the mixed sound.

First, regarding the method of obtaining the maximum noise value,
The present invention relates to a method for estimating the maximum amplitude of noise from a signal in which noise and a reproduction signal are mixed.
Table 1 below is a sample of calculated values indicating that there is a possibility that the maximum value of noise can be accurately detected from a signal in which noise and a reproduction signal are mixed. Table 1 relates to sine waves that are both uncorrelated. It can be estimated that the probability density near the maximum amplitude obtained by adding two uncorrelated signals to each other is the sum of the maximum values of the original two signals, which is the product of the probability densities near both maximum amplitudes. .

Table 1 shows the relationship between two sine waves (X) and sine waves (Y) that are uncorrelated with each other, and the signal (X + Y) that is the sum of them.
Table 1 shows the values calculated for the evaluation items on each horizontal axis at every 5 degrees. The evaluation range is 2160 degrees or 6 cycles for the sine wave X.
The difference between the maximum value of the sine wave (X) and the maximum value of the sine wave (X + Y) is equal to the difference between the maximum value of the sine wave (Y). Table 1 is an instantaneous value sampling every 5 degrees, so there is a slight degradation in accuracy, but it still matches the theoretical value very accurately.
The sine wave (X) and the sine wave (Y) have a one-to-root (2) frequency relationship.
Regarding the evaluation items on the horizontal axis,
PK () is the maximum amplitude of the signal in (),
AVG () is the average value of the signal in (),
RMS () is the effective value of the signal in (),
Indicates.
The vertical axis is
The results are shown when the maximum amplitude PK (Y) of the sine wave (Y) is changed from 0.01 times to 30 times the maximum amplitude PK (X) of X.
Assuming that the sine wave (X) is the playback sound and the sine wave (Y) is the noise,
This indicates that the value in the vertical column of PK (X + Y) -PK (X) matches the maximum value of PK (Y) within an error of 0.2% or less.
The calculated values are the results of sampling the sine wave (X) sampled every 5 degrees with 360 degrees as one period and calculating the range of 2160 degrees, that is, 6 periods.
6 cycles is a 200 msec time interval assuming a relatively good playback system with a minimum playback frequency of 30 Hz.
AVG (X + Y) -AVG (X) and RMS (X + Y) -RMS (X) are the difference in mean value and the difference in rms value, respectively, but they must not match PK (Y) I understand.

Table 2 shows the same evaluation results as Table 1 when the frequency of the sine wave (X) and the sine wave (Y) is one-pair route (root (route (2))). Sampling angles and evaluations for 6 cycles are the same as in Table 1. The relationship between the two frequencies is 1 to 2 ^ 0.125, but even in this condition, PK (X + Y) -PK (X) is equal to PK (Y) with an error of 0.2% or less in the evaluation interval of 200 msec. You can see that you are doing it.

Evaluation items on the horizontal axis are the same as in Table 1. The vertical axis is the same as in Table 1.

Table 3 shows that when two uncorrelated signals are M-sequence signals with an acoustic band sampling frequency of 44.1 kHz, the maximum amplitude of noise can be detected by detecting the maximum amplitude as in Tables 1 and 2. The results obtained by actual measurement are shown.
On the horizontal axis,
WN (20%) is the M-sequence signal of the reference acoustic band,
WN (20% + 4%) is the standard M-sequence signal.
The reference signal with a maximum value of 4% is a signal obtained by combining uncorrelated M-sequence signals.
100% is the maximum signed 16-bit value. For LEVEL, the numerical value (%) on the vertical axis indicates the range of the absolute value of the signal. The horizontal axis of 0.00% is the sum of the time intervals where the signal amplitude is in the range of 0.00% to 0.50%. The total evaluation time is 3 seconds.
The blanks between 4% and 16% are intervals omitted because the probability density of the amplitude has not changed. The maximum amplitude with the probability density of the reference M-sequence signal is 19.0%, and the signal that combines the 4% M-sequence signal is 22.5%, so the difference is 3.50%. Then it becomes 3.68%. Since the equivalent of 4% is noise, the maximum noise value can be measured with an error of 8%, which is practically sufficient for acoustic system control, even when the reproduced signal and noise are difficult to detect the M-sequence. It shows that. At least the reproduced signal is an acoustic signal, and the probability near the maximum value is much higher than that of the M sequence.
The above indicates that the maximum noise amplitude can be detected from the synthesized signal and the reference signal.

Table 4 shows the maximum value of driving noise by subtracting the maximum value of the reproduction signal from the maximum value of the signal where the driving noise of the car and the reproduction sound of the car audio are mixed. It is an example of the actual measurement result that shows that the error is detected by the error. Since the driving noise of the car and the reproduced sound of the car audio are generally independent sound sources, the two are uncorrelated.
MUSIC is a substitute for playback signals, and is an M-sequence signal assuming the worst case.
LOAD-Nise is a recording signal of driving noise of an example of a driver's seat of an example. MUSIC + LoadNoise is a composite signal of both.
For LEVEL, the numerical value (%) on the vertical axis indicates the range of the absolute value of the signal. The horizontal axis of 0.00% is the sum of the time intervals where the signal amplitude is in the range of 0.00% to 0.50%.
The total time is 300msec. The maximum amplitude with Music probability density is 29.5%, and the maximum value of the synthesized signal is 55.5%, so the maximum noise amplitude estimated from the actual measurement is 26%. Since the maximum amplitude of LoadNoise is 31%, the difference has an error of 1.3dB in -16% decibel conversion at 5% ratio. Since the control amount of the sound volume with respect to noise is about 10 dB at the maximum, the sound volume feedback system does not fall into unstable operation due to gain divergence due to an error of 16%.
The fact that the maximum noise amplitude can be calculated from the combined signal and the reference signal with an error of 1.3 dB indicates that this detection result is sufficient for actual use.

The calculations and actual measurement results in Tables 1, 2 and 3 and 4 above show that pure noise is obtained by subtracting the maximum value of the reproduced signal from the maximum value of the mixed sound signal corresponding to the synthesized sound of noise and reproduced sound. Indicates that the maximum value of can be detected . However, a process of combining the maximum values of both the mixed sound signal and the reproduced signal when there is no noise is required.

Second, regarding the relationship between the spectral distribution of noise components and the control of volume sound quality,
The following relates to a method for obtaining the maximum value of each noise frequency band and controlling the sound quality of the reproduced signal for each band corresponding to the maximum value.
As described above, the maximum noise value can be calculated, but there remains a problem of how to reflect the obtained maximum noise value in the correction of the volume sound quality of the reproduction signal.

In the case of noise from the driver's seat of a car, road noise and engine noise are distributed in a low-frequency range, and in the case of wind noise, the noise is high-level. In such a case, it is not appropriate to control the volume and sound quality with simple uniform parameters over the entire band.
Wind noise is scarcely generated at low speeds, but road noise that is biased toward low frequencies is generated even at low speeds depending on road conditions.
In the case of large passenger planes used for international flights, wind noise is stronger than engine noise at the front seat, and wind noise is weaker and engine noise is greater at the rear seat. Also, the window seat and the center seat are different.
In the case of a subway, there is a strong noise with a wide band due to the fact that it is a railway, and there are also echoes in the tunnel, sometimes in a low tone, sometimes in a middle tone, sometimes in a high tone, with a strong bias, and a rapidly changing noise environment It is.
At home, vacuum cleaners and talking voices make noise. In the case of a vacuum cleaner, the sound range is wide, and the intensity of the spoken voice is distributed over medium sounds.
When listening to music, lectures, or news in a noisy environment, it is desirable to finely control according to the spectral distribution of the noise, but it is roughly divided into low, middle and high frequencies, and environmental noise in each band. By correcting the sound volume according to, the sound quality correction of the entire band can be satisfied.

Thirdly, regarding the combination with noise cancellation technology that has become popular in recent years,
The following relates to the use of the residual signal of the noise canceller.
In terms of physical phenomenon, noise cancellation can be expressed accurately.
It uses the phenomenon of wave prevention, that is, reflection, rather than muffling. Since sound waves are energy, even if waves of opposite phase are injected, there is no cancellation of the waves, but the waves cancel out in the direction of the noise, but are reversed in the direction of approach. .
This is phenomenologically the same as reflection. Wave extinguishing converts sound waves, which are energy propagating in the air, into heat. However, active control of wave extinction is difficult both theoretically and technically. As of April 2017, it is practically used. The technology to become is not established.

When the control amount is voltage or current, it is relatively easy to cancel.
A speaker or a microphone is an energy conversion system of an element that generates or captures a wave, but an electrical system that is a counterpart of a sound wave usually handles either a voltage or a current signal. Control signals of control systems that control waves including energy conversion systems are handled by voltage or current, and there are many difficult issues. For these reasons, the function of the noise canceller is not always perfect, and it is difficult to raise the degree of completion to a state that satisfies human hearing.

In particular, car audio noise cancellers, unlike earphones and headphones, are in a listening environment in a state closer to an arbitrary space. Not reached.
For the above reasons, it is effective to improve the listening environment of car audio by extracting the maximum value of the noise component from the residual signal of the noise canceller and reflecting this value in the volume sound quality control of the reproduction signal. Since the noise canceller system already has a hybrid sound microphone and advanced signal processing resources, it does not require extra hardware costs for the detection system, and it uses extra resources for the signal processing system. Therefore, effective sound quality volume control of anti-noise is enabled.

The extracted noise level is the maximum value of the noise, and the frequency component of the noise cannot be specified from this value. Therefore, by providing a filter for the required band on both the playback signal side and the mixed sound signal side, and calculating the maximum noise value from the signal after passing through the band filter, the maximum noise value for each band with that value is calculated. Adjust the playback volume accordingly. As a method for applying signal processing to each divided band, there are a time division method and a parallel processing method.

Fourth, regarding averaging of the control amount of sound quality volume control corresponding to noise,
As described above, the volume and sound quality can be controlled in response to noise.
Even when listening in a noisy environment, human hearing is so sensitive that the volume and quality of the playback sound change instantly in response to noise, which is uncomfortable for listeners. Cause stress. On the other hand, when the average processing over a long time interval is reflected, when the noise becomes stronger, it may take some time to notice that the noise is stronger than the reproduced sound, but the noise suddenly weakened Sometimes a strong playback sound that has been supported by strong noise will continue for a while. This makes me feel uncomfortable in actual use. Although it is a sensuous problem, there are individual differences, but overall, there is a strong sense of incongruity with a large playback sound in the absence of noise.
For this reason, the sound quality control of the playback sound corresponding to the detected maximum noise value is
Low-speed attack operation in the direction of increasing noise intensity,
A high-speed release operation is appropriate in the direction in which the noise intensity decreases.
Experiments applied to passenger cars have confirmed that the combination of low-speed attack and high-speed release is the best when the road surface on the highway is moving from a good place to a bad place and vice versa.
In the case of a low-speed attack, the time constant is less than 1 second, which is uncomfortable, 2 to 4 seconds is good, and 10 seconds is too long because the radio broadcast content cannot be heard.
In the case of a high-speed release, when the time constant is 3 seconds or more, there is a sense of incongruity and 1 second is good. If the running noise is unstable for the shorter 0.3 to 0.1 seconds, the correction amount will be too small to function because the running noise is weak. Regarding the relationship between the amount of control and the amount of volume correction obtained by low-speed attack and high-speed release,
If the target is an earphone, a passenger car, even a passenger car, depending on the characteristics of the car,
Determine the optimal state for the listener by design.

Fifth, it is important to relate the amplitude ratio between the reproduced signal and the reproduced sound component in the mixed signal. A coefficient for determining the size of the hybrid sound signal is determined so that the maximum amplitude of the reproduced sound and the maximum amplitude of the hybrid signal within a certain time interval in the absence of noise coincide with each other.
As this method, there are the following four methods. These selections and specific methods are determined by design.
5-1. The listener operates the switch in the absence of noise so that the maximum value of the reproduction signal matches the maximum value of the hybrid signal .
5-2. In the case of car audio, an acceleration sensor is provided so that the reproduction signal when the output of the acceleration sensor falls below a certain value and the maximum value of the mixed sound signal are matched.
5-3. In the case of car audio, the maximum value of the reproduction signal and the hybrid sound signal is matched with the vehicle stopped or the engine stopped and the accessory power supply turned on.
No. 5-4. By making the level of the reproduced signal larger than the noise for a short time, the maximum value of the reproduced signal and the maximum value of the hybrid signal are matched.

First,
Since the noise component can be accurately measured and calculated from the signal in which the noise and the reproduction signal are mixed, and the spectrum distribution is also obtained, the volume sound quality of the reproduction signal optimum for the noise can be determined finely.

Second,
As a first result, even if the volume of the reproduction signal is increased according to the noise, the maximum noise value is accurately detected from the detected reproduction sound and the synthesized signal of the noise, and the detected noise intensity is increased. Since the intensity of the reproduction sound is not included or is small, the volume control of the reproduction system is accurate and does not cause the closed loop operation of the volume control system to become unstable.

Third,
An error signal when the noise canceller cannot satisfactorily cancel the noise, that is, the maximum amplitude of the noise component reaching the hearing can be obtained from the residual signal. Therefore, it is possible to correct the volume sound quality of the reproduced sound in accordance with the maximum noise value that reaches the hearing beyond the limit performance of the noise canceller.

This is an explanatory diagram for detecting the spectrum of noise, and the basic structure of this plan Explanatory diagram of playback sound volume control based on detected noise signal Explanatory diagram of a configuration example using the residual signal of the noise canceller as a hybrid sound signal Illustration of low-speed attack and high-speed release against noise changes Relationship between noise intensity, volume sound quality control and noise duration

Mobile phone with sound quality control function for noise-capable TV with volume sound quality control function for car audio noise
PA system with sound volume control function for noise

Application of the noise spectrum distribution calculated from the playback signal when the playback signal is known and the mixed sound signal that is a mixture of noise and playback sound.
Volume sound quality control program for earphones, headphones, passenger cars, TVs, PA systems that supports environmental noise.
Noise volume sound quality control program as an additional improvement function of the noise canceller function used in earphones, headphones, and passenger cars.

FIG. 1 is an explanatory diagram corresponding to claim 1 of the function of detecting the maximum noise value for each band from the reproduction signal and the mixed sound signal, which is the basic configuration of the present proposal.
+ And-indicate positive polarity and negative polarity, respectively.
INPUT is a signal input, and a music or broadcast playback signal is input.
The source is the playback signal, SAMP is the playback signal amplification function, and SP is the speaker.
MUSIC (source) is the playback sound that reaches the hybrid sound microphone, en is the noise source, the detection signal of the microphone installed in the place where noise is generated or measured as noise, NOISE (en) is the hybrid sound microphone MR is a hybrid microphone that detects the synthesized sound of noise and playback sound, MRAMP is a function of amplifying the electromotive force of the hybrid microphone, and EN + SOURCE is an amplified hybrid signal.
GAIN is a gain adjustment function for matching the maximum value of the playback signal with the maximum value of the playback signal component included in the mixed sound signal, and is controlled by CONDITIOBNER.
CTRLGAIN is the GAIN control signal. This is a function necessary to accurately detect the maximum value of the noise component when subtracting the maximum value of the reproduction signal from the maximum value of the mixed sound signal.
BPF- and BPF + are band filters for the playback signal and the mixed sound signal, respectively, and both have the same characteristics. The characteristics of BPF- and BPF + are controlled by CONDITIONEDR, which will be described later, and the noise spectrum distribution is detected in the required resolution with the required resolution by time division or parallel division. The division method, the division characteristics, and the number of divisions are determined by design. The inputs for BPF- and BPF + are SOURCE and EN + SOURCE, respectively.
BPF (SOURCE) and BPF (EN + SOURCE) are the output of the bandpass filter,
PEAK- and PEAK + are the maximum value detection functions, respectively, and PK (BPF (S)) and PK (BPF (EN, S) are the output of the maximum value detection function respectively. This is the maximum value within a certain time, for example 1 sec.This time width is determined by design according to the individual purpose.PEAK- and PEAK + are controlled by CONDITIONER by the control signal CTRLPK. The automatic detection method is determined by design.
ADD is an addition function, and the maximum noise level can be obtained by subtracting PK (BPF (S)) from PK (BPF (EN, S)) as indicated by the sign.
PK (BPF (EN)) is the maximum value of the detected noise.
The fact that PK (BPF (EN)) corresponds to the maximum noise level is as explained in Table 1, Table 2, Table 3, and Table 4.
CONDITIONER is the control function for the entire system, CTRLGAIN is the gain control signal for the playback signal, CTRLBPF is the control signal for the bandpass filter, and CTRLPK is the control signal for the maximum value detection function.
CTRLGAIN is adjusted so that the maximum value of both SOURCE and EN + SOURCE match when the noise is small enough to be ignored and when the playback sound is large and the noise can be ignored. In general, since the characteristics of the acoustic system drift due to environmental conditions such as temperature and time, it is desirable to update manually or automatically as necessary. The conditions for adjusting the gain of GAIN are determined by design.
NOISE_PATTERN is the spectrum pattern of the detected noise.
How to divide into a plurality of bands including one division of the entire range of the band is determined by design according to the purpose. As an application example of the detected noise spectrum pattern, there is volume sound quality control of reproduced sound according to the maximum noise value. Since the control amount of volume sound quality is only the noise component excluding the error of the configuration system, playback volume control according to the maximum noise value, and further, volume control of the playback sound for each band according to the maximum noise value for each band Is possible.

FIG. 2 is a configuration example of a system that performs volume sound quality control according to noise, which is simple and highly effective, when the basic configuration of FIG. 1 is divided into three parts of a low frequency range, a middle frequency range, and a high frequency range. . This corresponds to claim 3.
The same symbols as in FIG. 1 have the same functions.
COMP (B), COMP (M), and COMP (T) are volume correction functions for the low, middle, and high frequencies, respectively. In the volume correction function, the correction amount is controlled by the maximum noise value detected for each band. SUM_INP is an addition function.
BASS- and BASS + are low-band filters, MID- and MID + are mid-band filters, and TREBLE- and TREBLE + are high-band filters.
In FIG. 1, the number of band filters is arbitrary, but FIG. 2 is a three-part simultaneous measurement of a low range, a mid range, and a high range. It is an example of the most practical configuration.
GAINB, GAINM, and GAINT are gain adjustment functions that match the maximum values of the low, middle, and high frequency components of both the source and hybrid signals. The output of the gain adjustment function is SOURCEB, SOURCEM, and SOURCET, respectively.
The control of each gain is determined by CONDITIONER, and its control signal is CTRLGAIN.
BASS (SOURCEB) and BASS (EN + SOURCE) are the output of the bass filter,
MID (SOURCE) and MID (EN + SOURCE) are the output of the mid-band filter,
TREBLE (SOURCE) and TREBLE (EN + SOURCE) are the output of the high frequency band filter.
PEAKB- and PEAKB + are the maximum value detection function for the low range, PEAKM- and PEAKM + are the maximum value detection function for the middle range, and PEAKT- and PEAKT + are the maximum value detection function for the high range.
CTRLPK is a control signal for the maximum value detection function.
PK (B (S)), PK (B (EN, S)) is the maximum detection value in the low range, PK (M (S)), PK (M (EN, S)) is the maximum detection value in the mid range, PK (T (S)) and PK (T (EN, S)) are maximum detection values in the high range.
ADDB, ADDM, and ADDT are the addition functions, and PK (B (EN)), PK (M (EN)), and PK (T (EN)) are the maximum noise values in the low, mid, and high frequencies, respectively. This is a value obtained by subtracting the maximum detection value on the signal side from the maximum detection value. S / Q_B, S / Q_M, and S / Q_T are PK (B (EN)), PK (M (EN)), and PK (T (EN)) described in Tables 1 to 4, respectively. It is an averaging function for the low, middle and high frequencies, respectively, where the fast release time function is applied. Controls the volume correction of low, medium and high sounds according to the output of the averaging function.
The specific averaging method, attack time and release time are determined by design according to the application.
CONDITIONER is a control function for the entire system, and CTRL is a control signal for volume sound quality correction.

FIG. 3 is an explanatory diagram corresponding to claim 2 when the residual signal of the noise canceller is a mixed sound signal.
Since the noise canceller is out of the scope of the present plan, a detailed configuration and description of this mechanism will be omitted. In the case of FIG. 2, the mixed sound signal uses the amplified signal of the mixed sound microphone, but in the case of FIG. 3, the mixed sound signal uses the remaining signal without canceling the noise. Since the remaining signal includes a reproduced sound component and a noise component, this signal is used as a hybrid sound signal.
Two microphones are used, one for detecting noise from a noise source and the other for detecting mixed sound near the hearing. When there are many noise sources, a plurality of source noise detection microphones are generally used for a plurality of noise sources. FIG. 3 shows a single case. There are many noise canceller systems, and some systems do not have a noise source microphone, but the presence and number of noise source microphones are outside the scope of this proposal.

The source noise microphone is MN. MNAMP is its signal amplification function. TRN is the transmission constant of the path from the noise source to the hybrid sound microphone. EN is its output. ITRS is the inverse of the transfer constant from amplification by SAMP, conversion to sound waves by SP and reaching MR. This loop has the role of canceling the noise from the sound source that reaches the hybrid microphone. The more accurate the TRN and ITRS, the higher the noise cancellation performance, but this loop includes a sound wave energy propagation system, so there is always a solution due to the presence of standing waves and mechanical resonance. Not always.
In particular, it is difficult to improve the accuracy of the inverse transfer constant ITRS, and when there are many noise sources, it is difficult to solve this problem under the condition of low cost and low power consumption.
FEEDBACK is the output of ITRS, and the noise cancellation system is configured so that the sound that reaches MR via SAMP and SP cancels NOISE (en).
NOISE (en) is the noise detected when the source noise reaches the hybrid sound microphone through the propagation path.

SAMP is a function for amplifying a composite signal of a reproduction signal and a noise cancellation signal. The difference between SAMP and Fig. 1 or 2 is that the regenerative signal and the noise canceling signal are input.
TRS is the transfer constant until the playback signal source reaches the hybrid microphone. SOURCE is the output signal. If the TRS accuracy is sufficiently high, its output will match the SOURCE contained in the MRAMP output.
ADDSS is an addition function, and its output is the system residual component ERR. The noise cancellation function controls TRN and ITRS so that this residual signal ERR is minimized.
GAINB, GAINM, and GAINT are gain adjustment functions that adjust the maximum values of the low, medium, and high frequency playback signal components included in SOURCE and ERR, respectively. CTRLGAIN is the control signal. The outputs of GAINB, GAINM, and GAINT are SOURCEB, SOURCEM, and SOURCET, respectively. SOURCEB, SOURCEM, and SOURCET are the BASS-, MID-, and TREBLE- band splitting filter inputs, respectively. ERR becomes the input of BASS +, MID +, TREBLE + band division filter. Subsequent signal processing is the same as in FIG.
Claim 3 uses this residual signal as a hybrid sound signal. The mixed sound signal includes a reproduced sound and a noise component that reach the auditory sense.

Fig. 4 shows the average of low-speed attack time and high-speed release time, which is a function that reduces the sense of incongruity of the change in volume sound quality when changing the volume sound quality of the playback sound in response to the maximum noise value change. It is explanatory drawing of conversion.
The horizontal axis TIME is the time axis, the vertical axis E_NOISE is the maximum noise value, and COMPENSATION is the volume sound quality correction amount. Hereinafter, the correction amount is simply a volume sound quality correction amount.
t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10, t11 are the noise change points,
P0, P1, P2, P3a, P4, P5a, P6, P7, P8a, P9, P10a, and P11 indicate control point change points, respectively.
From t1 to t2, the noise suddenly increases from 9, and from t2 to t3, the noise decreases rapidly. In response to this change, the correction amount gradually increases from P1 to P2, but the correction amount decreases rapidly from P2 to P3a. Therefore, the correction responds dullly to noise changes that return to the original state in a short time.
When the noise suddenly increases from t4 to t5 and continues until t6, the correction amount increases slowly from P4 to P5a, and the state continues until P6. Furthermore, the noise increases for a short time from t6 to t7 t8, but the correction amount does not change as much as p6 p7 p8a.
The noise suddenly decreases from t9 to t10, but the correction amount suddenly returns from p9 to p10a.
It has been confirmed experimentally that the amount of noise correction by the averaging function with low attack time and high release time as described above is a control method with less sense of incongruity for hearing in comparison with other methods. . The attack time and release time, time constant or averaging method and its parameters are determined by design according to the purpose. Also, the attack time and release time in the low, mid, and high ranges are not necessarily the same.

FIG. 5 is an explanatory diagram of the tendency of the relationship between the magnitude of noise and the correction amount for each noise duration.
(A) is the low frequency range, (b) is the mid range, and (c) is the high range.
The horizontal axis E_NOISE is the maximum noise value, the vertical axis COMPENSATION is the correction amount,
BGN is the maximum background noise.
BASS, MID, and TREBLE indicate that the graph is the control amount for each duration corresponding to the maximum noise value in the low, mid, and high ranges, respectively.
2sec, 5sec, 10sec and 20sec are noise durations.
In either case, the shorter the noise duration, the smaller the correction amount, and naturally, the smaller the noise, the smaller the correction amount.

FIG.
+,-Positive polarity and Negative polarity
INPUT signal input
Source Playback signal
SAMP Playback signal amplification function
SP speaker
MUSIC (source) Playback sound reaching the hybrid sound microphone
en Noise source
NOISE (en) Noise reaching the hybrid microphone
MR hybrid sound microphone
Amplification function of electromotive force of MRAMP hybrid sound microphone
EN + SOURCE hybrid sound signal
GAIN gain adjustment function
CTRLGAIN GAIN control signal
BPF-, BPF + Band filter for playback signal and mixed sound signal
SOURCE, EN + SOURCE BPF-, BPF + output
BPF (SOURCE), BPF (EN + SOURCE) Bandpass filter output
PEAK-, PEAK + maximum value detection function
PK (BPF (S)), PK (BPF (EN, S) Maximum value detection function output
ADD addition function
PK (BPF (EN)) Maximum noise level
CONDITIONER System-wide control functions
CTRLGAIN Playback signal maximum gain control signal
CTRLBPF Band filter control signal
CTRLPK Maximum value detection function control signal
NOISE_PATTERN Spectral pattern of detected noise

FIG.
COMP (B), COMP (M), COMP (T) Volume correction function for low, mid, and high frequencies
SUM_INP addition function
BASS-, BASS + Bass band filter
MID-, MID + Mid-range filter
TREBLE-, TREBLE + high frequency band filter
GAINB, GAINM, GAINT Gain adjustment function
SOURCEB, SOURCEM, SOURCET GAINB, GAINM, GAINT output
CTRLGAIN GAINB, GAINM, GAINT gain control signal
BASS (SOURCEB), BASS (EN + SOURCE) Output of low frequency band filter
MID (SOURCE), MID (EN + SOURCE) Mid-range filter output
TREBLE (SOURCE), TREBLE (EN + SOURCE) Output of high-frequency band filter
PEAKB-, PEAKB + Maximum value detection function for bass range
PEAKM-, PEAKM + Maximum midrange detection function
PEAKT-, PEAKT + Maximum treble detection function
CTRLPK Maximum value detection function control signal
PK (B (S)), PK (B (EN, S)) Maximum detection value in low range
PK (M (S)), PK (M (EN, S)) Maximum midrange detection value
PK (T (S)), PK (T (EN, S)) Maximum detection value in high range
ADDB, ADDM, ADDT addition function
PK (B (EN)), PK (M (EN)), PK (T (EN))
S / Q_B, S / Q_M, S / Q_T Averaging function with low attack time and high release time
CONDITIONER System-wide control functions
CTRL Volume sound quality control signal

FIG.
MN source noise microphone
MNAMP amplification function
EN its output
FEEDBACK ITRS output
NOISE (en) Noise reaching the hybrid microphone
SAMP Amplification function of harmonized signal of reproduced sound and noise cancellation signal
Transfer constant until the TRS playback signal reaches the hybrid microphone
TRN Transfer constants from the noise source to the hybrid microphone
SOURCE The output signal
ITRS TRN inverse constant
ADDSS addition function
ERR The output of the system residual component
GAINB, GAINM, GAINT Gain adjustment function
CTRLGAIN its control signal
SOURCEB, SOURCEM, SOURCET GAINB, GAINM, GAINT output
BASS-, MID-, TREBLE- Band-splitting filter
BASS +, MID +, TREBLE + Band-splitting filter

FIG.
TIME time axis
E_NOISE Noise maximum value axis
COMPENSATION Volume sound quality correction amount axis
t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10, t11 Noise change points
P0, P1, P2, P3a, P4, P5a, P6, P7, P8a, P9, P10a, P11 Control point change point

FIG.
E_NOISE Noise maximum value axis
COMPENSATION Playback sound correction amount axis
BGN background noise maximum
BASS, MID, and TREBLE control amounts for the low, mid, and high frequencies, respectively
2sec, 5sec, 10sec, 20sec Noise duration

Claims (4)

The maximum value is the maximum amplitude of the signal in that time zone during a specific time interval.
The specific means of obtaining the maximum value shall be determined by design,
The signal that is the source of music and broadcast sound playback is the playback signal,
The sound that plays the playback signal is used as the playback sound.
Separate from the playback sound, the sound that reaches the hearing that interferes with listening is defined as noise.
The mixed sound is the sound that is mixed with the playback sound and noise.
The microphone that detects the hybrid sound is a hybrid microphone,
A signal obtained by amplifying the signal from the hybrid microphone is used as a hybrid signal.
Band division means a frequency band that is meaningful for human hearing, and that band is divided into one or more bands.
The band division filter is a band division filter,
Each band divided by the band division filter is defined as a divided band.
Both the band division filter on the playback signal side and the band division filter on the hybrid sound signal side are composed of one or more sets of a pair of filters of the same characteristic band,
A function that adjusts the size of the hybrid signal so that the maximum value of the playback signal and the hybrid sound signal match when the noise is small enough to be ignored compared to the playback sound, or when the playback sound is sufficiently louder than the noise. It is the first to have
Of the output of each band splitting filter,
The value obtained by subtracting the maximum value on the playback side from the maximum value on the hybrid side is the detected noise maximum value.
The second is to have a function to obtain the maximum detected noise value of each divided band.
The set of detected noise maximum values for all the divided bands is defined as the noise spectrum distribution,
The third is to obtain the noise spectrum distribution.
A noise spectrum distribution detecting method for detecting a maximum value of noise for each divided band from a mixed sound signal and a reproduced signal, comprising first, second and third. The terms described in claim 1 shall be cited ,
A noise canceller is a function that reduces noise that reaches the hearing.
The residual signal is a mixed sound signal described in claim 1 when the noise canceller is functioning and the cancel function is not perfect and a noise component remains ,
The fourth is to use the residual signal as a hybrid sound signal .
The spectrum distribution of the maximum value of the noise is detected from the mixed sound signal and the reproduced signal which are the residual signals of the noise canceller consisting of the first, second, third and fourth described in claim 1 Noise spectrum distribution detection method. The terms described in claim 1 and claim 2 shall be cited ,
The attack time is a parameter that corresponds to the time constant to follow the increasing signal of the function of detecting the average value,
The release time is a parameter corresponding to the time constant to follow a signal that becomes smaller in the function of detecting the average value,
The averaging function that has a low attack time and a high release time is called the noise maximum value averaging function.
The above attack time and release time are determined by design,
The maximum noise level averaging function takes the detected maximum noise level as input, the output is the volume sound quality correction value, and the maximum noise level averaging function functions for each divided band.
The fifth is to have a noise maximum value averaging function,
The sixth feature is to have a function of correcting the volume for each divided band of the reproduction signal with the volume sound quality correction value for each divided band;
The relationship between the volume sound quality correction value and the volume correction amount for each divided band is determined by design,
1st, 2nd, 3rd and the above described in claim 1, 5th and 6th or 1st, 2nd, 3rd, 4th, and 5th and 6th described in claim 2 However, the anti-noise volume sound quality correction method for detecting the maximum value of noise and correcting the volume of each band for each divided band with the value. In the playback system in the actual use state of the earphone,
The function that predicts the electromotive force generated at the earphone terminal from the playback signal is called the earphone electromotive force prediction function,
The output of the earphone electromotive force prediction function when there is no noise is the predicted earphone signal,
The seventh function is to obtain a predicted earphone signal.
The measured value of the electromotive force at the earphone terminal is used as an actual earphone signal,
According to an eighth aspect, a signal obtained by subtracting the predicted earphone signal from the measured earphone signal in the actual use state of the earphone in the presence of noise is defined as the mixed sound signal described in claim 1;
The first, the second, the third, the seventh, and the eighth described in claim 1
A noise level sound quality correction method for detecting a maximum value of noise from a mixed sound signal and a reproduction signal, and correcting the volume of each band for each divided band based on the detected value.