JP2009141471A - Automatic volume correcting apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that an excessive gain is applied to a pulse-like signal in an automatic volume correcting apparatus for calculating a gain for an input signal on the basis of an RMS (root-mean-square) value. <P>SOLUTION: An absolute value converting section 11 converts an input signal into an absolute value, and a hold/release section 12 applies hold/release processing (processing for holding a peak value of the pulse-like signal and gradually attenuating it) to the signal converted into the absolute value. An RMS estimating section 30 obtains an RMS estimation value on the basis of the signal subjected to the hold-release processing. A gain calculating section 34 calculates a gain on the basis of the RMS estimation value, and an attack/release section 36 smoothes the gain. An output signal amplitude adjusting section 22 multiplies an input signal delayed by a predetermined value using an input signal delay section 21 by the smoothed gain, thereby adjusting the amplitude of the input signal. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an automatic volume correction device that calculates a gain (gain) for an input signal based on a root mean square (RMS) value (including an estimated RMS value), and in particular, an excessive gain for a pulsed signal. The present invention relates to an automatic sound volume correction device that can avoid the problem of being applied.

  When watching TV, when the program changes to CM (commercial), the volume may be reduced because the CM is noisy, or when listening to music, the volume difference between songs may be large and the volume may change. . In order to reduce the difficulty of hearing due to the volume difference of the sound source, the discomfort due to the large volume, and the troublesome operation of the volume control, the gain is adjusted according to the volume level of the sound source and the output volume is kept constant. Is used to keep Such processing is called automatic volume correction in this specification, and a device that performs this automatic volume correction processing is called an automatic volume correction device.

  As the volume control at this time is performed according to a criterion close to human audibility, a better volume correction effect can be obtained. Therefore, audibility weighting may be performed on the input signal or the root mean square (RMS) value of the signal may be used. is there. For example, there is a conventional automatic volume correction device (dynamic range compression device) using an RMS value (see Patent Document 1).

  FIG. 6 is a diagram illustrating a configuration of an automatic sound volume correction device using a conventional RMS. The automatic volume correction device 3 shown in FIG. 6 includes an RMS calculation unit 30.

  In the RMS calculation unit 30, the audio input signal is squared by the square calculation unit 31, and the squared signal is passed through the low-pass filter 32 to calculate an estimated value of the mean square value of the input signal. The root mean square estimated value is logarithmically converted by the logarithmic conversion unit 33, the gain is calculated in the logarithmic (dB) region by the gain calculation unit 34, and the obtained gain is converted to the linear region by the inverse logarithmic conversion unit 35. Perform inverse log transformation. The gain calculation unit 34 calculates the gain after obtaining the square root of the input root mean square estimated value. Thereby, the gain with respect to the root mean square (RMS) of the input signal is obtained.

  The output signal amplitude adjustment unit 22 multiplies the input signal delayed by the input signal delay unit (delay) 21 by the smoothed gain calculated by the gain calculation unit 34 to obtain an output signal.

  However, in the above-described conventional dynamic range compression apparatus, when a pulse-like signal having a sharp peak is input, the pulse-like signal is cut by the low-pass filter (LPF) 32 used for the RMS calculation. . For this reason, the level of the input signal is evaluated to be small in the gain calculation of the gain calculation unit 34, and as a result, an excessive gain may be applied to the pulse signal.

As described above, an excessive gain is applied to the pulse-like signal, so that an overflow of the pulse portion is likely to occur. In addition, if the signal is clipped to prevent overflow, the sound quality may be deteriorated due to distortion.
Japanese Patent No. 3497413

  As described above, the method using the root mean square (RMS) value has a problem in that an excessive gain is applied to a pulse-like signal, so that sound quality is deteriorated and a sufficient dynamic range cannot be secured. Here, the dynamic range is reduced because if the head margin is increased in order to prevent overflow when a pulse-like signal having a sharp peak is excessively gained, the signal level is lowered. That is why.

  The present invention has been made in view of such circumstances, and an object of the present invention is to provide a pulse having a sharp peak in an automatic volume correction apparatus that calculates a gain for an input signal based on a root mean square (RMS) value. It is an object of the present invention to provide an automatic sound volume correction device capable of avoiding deterioration of sound quality due to excessive gain being applied to a signal and ensuring a sufficient dynamic range.

The present invention has been made to solve the above problems, and the automatic volume correction device of the present invention has an RMS calculation unit for calculating an RMS value which is a root mean square value of an input signal, and the RMS value is calculated. An automatic volume correction device that calculates a gain of an input signal based on the input signal, detects a peak point of the input signal, holds the value of the peak point as a hold value, and holds the hold value during a predetermined hold period. A hold / release unit is provided that outputs to the RMS calculation unit as it is, and performs a hold / release process that outputs the hold value to the RMS calculation unit while attenuating the hold value at a predetermined attenuation rate after the hold period has elapsed. It is characterized by that.
The automatic volume correction device of the present invention having the above-described configuration includes an RMS calculation unit that calculates the root mean square (RMS) value of the input signal, and calculates the gain of the input signal based on the RMS value. In this case, the value of the peak point of the input signal is detected and held, and this held value is held and outputted as it is for a predetermined hold period, and after the hold period, the held value is set at a predetermined attenuation rate. Output to the RMS calculator while attenuating.
As a result, in the conventional automatic volume correction device using RMS, it is possible to avoid deterioration of sound quality because an excessive gain is applied to a pulse-like signal having a sharp peak, and an excessive gain is applied. Since this phenomenon is suppressed, a large head margin is not required, so that a sufficient dynamic range can be secured.

Further, in the automatic sound volume correction device of the present invention, the hold / release unit compares the input signal with the hold value signal, holds and outputs the signal having the larger level, and the level of the input signal is When the value is smaller than the holding value (when the input signal is smaller than the holding value), the maximum value holding unit that counts the holding period of a predetermined time and the output of the maximum value holding unit are delayed by a predetermined time and output The delay unit outputs the output signal of the delay unit to the maximum value holding unit as it is as a hold value signal during the hold period, and outputs a predetermined signal to the output signal of the delay unit after the hold period elapses. An attenuation unit that generates a signal of a new holding value multiplied by an attenuation factor and outputs the signal to the maximum value holding unit.
The automatic volume correction device of the present invention having the above-described configuration includes an RMS calculation unit that calculates the root mean square (RMS) value of the input signal, and calculates the gain of the input signal based on the RMS value. The maximum value holding unit compares the input signal and the holding value signal, detects and holds the level of the larger signal, and outputs the held signal as it is. Then, when “input signal <holding value” is satisfied, counting of the hold period is started. Further, the output of the maximum value holding unit is delayed by the delay unit and output. Then, the attenuation unit outputs the output of the delay unit as it is as a hold value signal to the maximum value holding unit during the hold period, and after the hold period has elapsed, a new attenuation factor is applied to the output of the delay unit. A holding value signal is generated and output to the maximum value holding unit. The maximum value holding unit selects and outputs the holding value signal generated by the attenuation unit until an input signal having a level exceeding the holding value signal generated by the attenuation unit is input.
As a result, the peak point of the input signal is detected, and the value of this peak point is output as the hold value as it is during the hold period, and after the hold period, the hold value is output while being attenuated by a predetermined attenuation rate. be able to. For this reason, in the conventional method using RMS, an excessive gain is applied to a pulse-like signal having a sharp peak, so that deterioration in sound quality can be avoided and a sufficient dynamic range can be secured.

The automatic volume correction device of the present invention includes an input signal delay unit that outputs an input signal with a predetermined time delay, an absolute value conversion unit that converts the input signal into an absolute value signal, and the absolute value conversion unit. A hold / release section that performs the hold / release process on the output signal of the signal, a square calculation section that squares the signal output from the hold / release section, and a signal squared by the square calculation section A low-pass filter for performing filtering and calculating an estimated value of the mean square value of the amplitude of the input signal, a logarithmic converter that converts the output signal of the low-pass filter into a logarithmic domain signal, and a logarithm by the logarithmic converter A gain calculation unit that performs gain calculation based on the signal converted into the region, and inversely converts the gain in the logarithmic region calculated by the gain calculation unit from the logarithmic region to the gain in the linear region A logarithmic conversion unit, an attack / release unit that smoothes the gain by an attack / release process, and a signal smoothed by the attack / release unit multiplied by a signal output from the input signal delay unit. An output signal amplitude adjusting unit that adjusts and outputs the amplitude.
In the automatic sound volume correction apparatus of the present invention having the above-described configuration, the input signal is converted into an absolute value, and a hold / release process is performed on the signal of the absolute value. Also, the squared root mean square (RMS) value is obtained by squaring the signal subjected to the hold / release process and passing through a low-pass filter. Then, the gain is calculated based on the RMS value, and the amplitude of the input signal delayed by a predetermined amount is adjusted and output based on the calculated gain.
As a result, in the conventional automatic volume correction apparatus using RMS, it is possible to avoid deterioration of sound quality because an excessive gain is applied to a pulse-like signal having a sharp peak, and to ensure a sufficient dynamic range. Become.

The automatic volume correction device of the present invention also includes an L-side input signal delay unit that outputs a left-side channel input signal of a stereo signal with a predetermined time delay, and a stereo-signal right-channel input signal that is delayed by a predetermined time. An R side input signal delay unit that outputs the left channel input signal, an L side absolute value conversion unit that converts the left channel input signal into an absolute value signal, and an R side absolute signal that converts the right channel input signal into an absolute value signal. A value conversion unit, an L-side hold / release unit that performs the hold / release process on the output signal from the L-side absolute value conversion unit, and an output signal from the R-side absolute value conversion unit. R-side hold / release section to be applied, left channel signal subjected to hold / release processing by the L-side hold / release section, and R-side hold / release A maximum value detection unit that takes a maximum value with respect to the signal of the left channel for which hold / release processing has been performed by the unit, a square calculation unit that calculates a square of the signal output from the maximum value detection unit, and the square A low-pass filter that performs filtering of the signal squared by the calculation unit and calculates an estimated value of the mean square value of the amplitude of the input signal; and a gain calculation unit that performs gain calculation based on the output of the low-pass filter; A smoothing unit for smoothing the gain, and an L-side output signal amplitude adjustment for adjusting the amplitude of the signal by multiplying the signal output from the L-side input signal delay unit by the smoothed gain and outputting the signal And an R-side output signal amplitude adjustment unit that adjusts the amplitude of the signal by multiplying the signal output from the R-side input signal delay unit by the smoothed gain and outputs the adjusted signal. Do
In the automatic volume correction apparatus of the present invention having the above-described configuration, each of the left channel input signal and the right channel input signal of the stereo signal is converted into an absolute value, and a hold / release process is performed on each absolute value signal. . Then, the maximum value of the left and right channel signals subjected to the hold / release process is taken. The maximum value signal is squared and passed through a low-pass filter to obtain a root mean square (RMS) value. Then, the gain is calculated based on the RMS value. Then, based on the calculated gain, the amplitude is adjusted and output with respect to the input signals of the left and right channels delayed by a predetermined amount.
As a result, in a conventional automatic volume correction device for stereo using RMS, excessive gain can be applied to a pulse-like signal having a sharp peak, so that deterioration of sound quality can be avoided and a sufficient dynamic range is secured. become able to.

In the automatic sound volume correction apparatus of the present invention, the hold period in the hold / release unit is approximately 5 msec, and the attenuation rate d of the attenuation unit is d = 1−e ^X , where X = −1 / (Tr × FS), where Tr = 20 msec and FS is a sampling rate.
As a result, in the conventional automatic volume correction apparatus using RMS, an excessive gain is applied to the pulse-like signal, so that deterioration of the sound quality is avoided and a sufficient dynamic range can be ensured. Hold that takes into account the forward-looking masking (a phenomenon in which the sound sensation gradually disappears when the sound is stopped, so the sound generated during that time is masked by the preceding sound) / Release processing can be performed.

  In the automatic sound volume correction apparatus of the present invention, the root mean square (RMS) value is calculated based on a signal obtained by subjecting an input signal to hold / release processing (processing to hold and gradually attenuate the peak value of a pulsed signal). Since the estimated value is obtained and the gain is calculated based on the RMS value, it is possible to avoid applying an excessive gain to the pulse signal having a sharp peak. As a result, the risk of signal overflow is reduced, sound quality deterioration due to clipping can be avoided, and a reduction in dynamic range can be suppressed because a large head margin is not required.

  Embodiments of the present invention will be described below with reference to the accompanying drawings.

[First Embodiment]
FIG. 1 is a diagram showing a configuration of an automatic sound volume correction apparatus according to the first embodiment of the present invention.
The automatic volume correction apparatus 1 shown in FIG. 1 differs from the automatic volume correction apparatus (dynamic range compression apparatus) 3 shown in FIG. 6 in terms of configuration in the automatic volume correction apparatus 3 shown in FIG. The conversion unit 11 and the hold / release unit 12 are newly added, and other configurations are the same as those of the automatic volume correction apparatus 3 shown in FIG. For this reason, the same code | symbol is attached | subjected to the same component. Note that the input signal and the output signal to the automatic sound volume correction apparatus shown in FIG. 1 are digital signals, and each processing unit is configured to process the digital signals, but has a circuit configuration using analog signals. You can also.

  In FIG. 1, an audio signal is input as an input signal In, and the input signal In is input to an absolute value conversion unit 11 and an input signal delay unit (delay) 21. Here, as will be described later, in the gain calculation process, there is a process of taking a time average of the input signal, and the calculated gain is delayed with respect to the input signal. The input signal delay unit 21 compensates for this gain calculation delay by delaying the input signal. If you multiply the gain without delaying the input signal, the gain will not follow when the level of the input signal changes suddenly, and a large amplitude signal will be output due to excessive gain, or the gain will be small despite the small input signal. However, by providing the input signal delay unit 21, such a problem is solved. In order to simplify the explanation, only a single channel is shown in FIG. 1, but in the case of 2 channels, 4 channels, and 5.1 channels, the automatic operation of the present invention is performed corresponding to each channel. A volume correction device is provided.

  In FIG. 1, an input signal In input to the absolute value converter 11 is converted into an absolute value signal. The absolute value signal output from the absolute value conversion unit 11 is input to the hold / release unit 12.

  The hold / release unit 12 compares the input signal In with the signal of the hold value held in the hold / release unit 12 and holds and outputs the signal having the higher signal level. The hold value signal is generated by multiplying the output signal of the hold / release unit 12 by a predetermined attenuation factor d. The hold / release unit 12 outputs the held signal to the root mean square estimation unit 30 while attenuating the held signal after the lapse of a predetermined hold period (details of the hold / release unit 12 will be described later).

  A signal output from the hold / release unit 12 is input to the root mean square estimation unit 30. In the mean square estimation unit 30, the signal output from the hold / release unit 12 is squared by the square calculation unit 31. The signal squared by the square calculator 31 is input to a low-pass filter (LPF) 32. The low-pass filter 32 outputs only the low frequency component of the input signal, but the process is equivalent to estimating the average value of the input signal. If the time average of the input signal is calculated as it is, the amount of calculation increases, so an alternative method of estimating the average value is used. In this way, the estimated value of the mean square value of the amplitude of the input signal is calculated by the square calculation unit 31 and the low-pass filter 32.

  The root mean square estimated value is output to the logarithmic conversion unit 33, converted to a logarithm (dB) with 10 as the base by the logarithmic conversion unit 33, and output to the gain calculation unit 34 as a dB output value.

  The gain calculation unit 34 performs gain calculation based on the root mean square estimated value of the logarithmic (dB) region input from the logarithmic conversion unit 33. At this time, the gain calculation unit 34 uses the following gain table in order to perform gain calculation based on the root mean square (RMS) of the input signal In. That is, since the input to the gain calculation unit 34 is an estimated value of “root mean square”, a pseudo “square root” calculation is performed using the gain table. In this gain table, for example, as a gain of an input value “−10 dB” to the gain calculation unit 34, a gain value for “−5 dB” which is the square root thereof is stored. A large gain is stored for a small input value, and a small gain is stored for a large input value. Then, the gain value in the logarithmic region obtained by the gain calculating unit 34 is inversely transformed from the logarithmic region to the linear region by the inverse logarithmic transformation unit 35.

  Next, the obtained gain value is output to the attack / release unit 36. In this attack / release unit 36, when the gain decreases (when the input signal In increases), an attack process for suppressing the gain decreasing speed; When the gain increases (the input signal In decreases), release processing is performed to suppress the gain increase speed. In this way, the attack / release unit 36 smoothes the gain change, so that when the input signal level changes abruptly and the gain also changes abruptly, distortion that gives a sense of incongruity to the output signal occurs. To prevent it.

  Next, the output signal amplitude adjustment unit 22 multiplies the input signal delayed by a predetermined amount in the input signal delay unit (delay) 21 by the smoothed gain calculated by the attack / release unit 36 to adjust the amplitude. Output signal Out is output.

FIG. 2 is a diagram illustrating a circuit configuration of the hold / release unit 12.
As shown in FIG. 2, the hold / release unit 12 compares the input signal In with the signal of the hold value Hold, and holds a larger signal as a maximum value, and a maximum value holding unit 41 of one sampling time. The delay unit (Z ⁻¹ ) 42 and an attenuation unit 43 having an attenuation rate d (d ≦ 1) are included.

  In FIG. 2, a maximum value holding unit 41 holds a comparator 41A that compares the level of the input signal In and the level of the hold value Hold, and holds and outputs the larger one of the input signal In and the hold signal. It has a hold unit 41B and a hold counter 41C that counts a signal hold time T in the hold unit 41B.

  In the above configuration, if the input signal In is greater than the signal of the hold value Hold, the level of the input signal In is held as a hold value in the hold unit 41B. However, if a larger signal is input during the hold, the hold value of the hold unit 41B is replaced and the hold is started again.

  The output signal Out of the maximum value holding unit 41 is input to the delay unit 42 of one sampling time and output to the attenuation unit 43. The signal input to the attenuating unit 43 is multiplied by the attenuation rate d, and is output toward the maximum value holding unit 41 as a new holding value Hold. The maximum value holding unit 41 compares the new holding value Hold with the input signal In of the next sample, and holds the larger signal in the holding unit 41B.

  Then, after the new hold value is held in the hold unit 41B, when the hold value Hold becomes larger than the input signal In, the hold period 41C is counted by the hold counter 41C. During this hold period, the attenuation rate in the attenuation unit 43 is set to 1, so that the held value in the maximum value holding unit 41 is not attenuated. As a result, the maximum value holding unit 41 can hold the new hold value as it is during the hold period.

  That is, when “In ≧ Hold”, the hold time (hold period T) is set in the hold counter 41C. When “In <Hold”, the count value of the hold counter 41C is decremented. When the count value of the hold counter 41C is 1 or more, the attenuation rate in the attenuation unit 43 is set to 1, and the signal is not attenuated (hold state).

  The hold period T in the hold / release unit 12 is, for example, 5 msec, and the attenuation rate d is

d = 1−e ^X.
However, X = −1 / (Tr × FS).
(Tr = 20 msec: release time, FS is the sampling rate)

  FIG. 3 is a diagram for explaining the detailed operation of the hold / release process in the hold / release unit 12. The operation will be described below with reference to FIG.

  In FIG. 3, the interval from time t0 to t1 is “holding value Hold <input signal In”, and the output value (level of the output signal Out) from the maximum value holding unit 41 is the input value (the level of the input signal In). Amplitude level).

  After that, the level of the input signal In decreases, and during the interval from time t1 to t2, it is a hold period in the maximum value holding unit 41. The maximum value holding unit 41 has a peak value (for a certain time T (hold period)). The signal indicated by a in the figure is held and output as it is. Then, after a certain time T (hold period T) has elapsed, the release period starts from time t2.

  In this release period, a hold value Hold is generated by multiplying the signal obtained by delaying the output signal Out from the maximum value holding unit 41 by the delay unit 42 by the attenuation unit 43 by the attenuation rate d. The maximum value holding unit 41 compares the input signal In with the holding value Hold output from the attenuation unit 43, and holds and outputs the larger signal (the holding value Hold that gradually attenuates in this release period). . Therefore, during the release period, a signal that gradually attenuates is output from the maximum value holding unit 41 as indicated by the symbol b.

  Thereafter, at time t3, when “holding value Hold <input signal In”, the output value from the maximum value holding unit 41 (the level of the output signal Out) is the same as the input value (the amplitude level of the input signal In). Become.

  Thereafter, at time t4, the level of the input signal In decreases, and during the period from time t4 to t5, the hold period in the maximum value holding unit 41 is reached, and the maximum value holding unit 41 has a fixed time T (hold period T). During this time, the peak value (signal indicated by c in the figure) is held and output. Then, after a certain time T (hold period T) has elapsed, the release period starts from time t5. In this release section, as indicated by reference sign d, a signal that gradually attenuates is output from the maximum value holding unit 41. Thereafter, the same operation as described above is repeated.

  Thus, the automatic volume correction device of the present invention is configured to perform hold / release processing on the absolute value of the input signal In. As a result, the level gradually decreases when the level of the input signal decreases, and the pulse-like signal gradually attenuates due to the hold / release effect. This hold / release process refers to forward masking, which is a characteristic of human hearing. Prospective masking is a phenomenon in which when a sound is stopped, the sensation of the sound gradually disappears, so that the sound generated during that time is masked by the preceding sound.

  FIG. 4 is a diagram for explaining the prospective masking (cited reference; E Zwicker (author), Yukiko Yamada (translation), “Psychoacoustics”, “Nishimura Shoten”, pp 133 (1992)).

  FIG. 4 shows a graph relating to the time effect in masking (the test sound level SLγ (above the silent minimum audible value) related to the time Δt after the start of masking noise or the time tv after the stop). As shown in FIG. 4, there is a forward masking phenomenon in human hearing, and this phenomenon can also be used in hold / release processing.

  As described above, in the automatic sound volume correction apparatus of the present invention, hold / release processing is performed on the absolute value of the input signal In, and the mean square estimated value is calculated by the square calculator 31 and the low-pass filter (LPF). It is configured as follows.

  Since this hold / release process increases the energy of the pulse-like signal, the root mean square estimated value is also calculated to be larger. The energy increase due to the hold / release process is small in the originally high energy section. As described above, since the root mean square estimated value of the pulse signal is increased, the calculated gain value is decreased, and the problem that an excessive gain is applied to the pulse signal can be avoided.

[Second Embodiment]
In the first embodiment, an example in the case of a single channel has been described in order to explain the basic configuration of the automatic sound volume correction apparatus of the present invention. In the second embodiment of the present invention, An example of a stereo two-channel automatic volume correction apparatus will be described.

FIG. 5 is a diagram showing a configuration of an automatic sound volume correction apparatus according to the second embodiment of the present invention.
In the automatic sound volume correction apparatus 2 shown in FIG. 5, the input signal Lin of the left channel of the stereo signal and the input signal Rin of the right channel are input.

  The input signal Lin of the left channel is input to the L side input signal delay unit (delay) 21L and the L side absolute value conversion unit 11L. Further, the input signal Rin of the right channel is input to the R side input signal delay unit (delay) 21R and the R side absolute value conversion unit 11R, respectively.

  The input signal Lin of the left channel is detected by the L-side absolute value converter 11L, and the hold / release process described above is performed by the L-side hold / release unit 12L. Similarly, the R channel absolute value converter 11R detects the absolute value of the right channel input signal Rin, and the R side hold / release unit 12R performs hold / release processing.

  Then, the signal of the left channel that has been held / released by the L-side hold / release unit 12L by the maximum value detector 13 and the signal of the left channel that has been held / released by the R-side hold / release unit 12R. Take the maximum value with the signal.

  Subsequently, the output signal from the maximum value detection unit 13 is input to the root mean square estimation unit 30. In the mean square estimation unit 30, the output signal from the maximum value detection unit 13 is squared by the square calculation unit 14, and the squared signal is input to a low pass filter (LPF) 15. The square calculation unit 14 and the low-pass filter 15 calculate an estimated value of the mean square value of the amplitude of the input signal.

  Subsequently, the release unit 16 receives the mean square estimated value output from the low-pass filter 15 and performs a release process for gradually decreasing the input value when the input value decreases.

  Then, the gain calculation unit 17 calculates the gain for the input signal. The gain calculation unit 17 performs logarithmic conversion on the input signal, performs gain calculation in the logarithmic domain, and performs inverse logarithmic conversion on the obtained gain and outputs it. Here, in gain calculation, gain calculation based on the root mean square (RMS) of the input signal In is performed using the same gain table as in the first embodiment. Subsequently, the attack unit 18 performs gain smoothing.

  The L-side output signal amplitude adjusting unit 22L multiplies the gain smoothed by the attack unit 18 by the input signal delayed by a predetermined amount by the L-side input signal delay unit (delay) 21L, and outputs the output signal. Output as Lout.

  Similarly, the R-side output signal amplitude adjusting unit 22R multiplies the gain smoothed by the attack unit 18 by the input signal after being delayed by a predetermined amount by the R-side input signal delay unit (delay) 21R, Output as an output signal Rout.

  Compared with the automatic volume correction apparatus of the first embodiment shown in FIG. 1, the automatic volume correction apparatus of the second embodiment shown in FIG. The difference is that the gain is calculated after taking the value and the gain calculation unit 17 includes logarithmic transformation and inverse logarithmic transformation (however, the gain is in the logarithmic region as in the first embodiment). Also, the attack / release block is divided and release processing is performed before gain calculation.

  However, the concept of performing the hold / release processing on the signal that takes the approximate amount of RMS is the same as that of the automatic volume correction apparatus of the first embodiment.

  Therefore, even in the second embodiment, the hold / release process increases the energy of the pulse-like signal, so that the RMS value is also calculated larger. Thus, since the RMS value of the pulse signal is increased, the calculated gain value is decreased, and the problem that an excessive gain is applied to the pulse-like signal can be avoided.

  Although the embodiments of the present invention have been described above, the automatic sound volume correction apparatus of the present invention is not limited to the above-described illustrated examples, and various modifications can be made without departing from the scope of the present invention. Of course.

  For example, a square root calculation unit that directly calculates the square root from the output signal of the low-pass filter 32 may be provided, and the gain table of the gain calculation unit 34 may be configured not to consider the square root. In this case, the square calculation unit 31, the low-pass filter 32, and the square root calculation unit constitute an RMS calculation unit. In the case of the first and second embodiments, it can be considered that the RMS calculation unit is configured by the square calculation unit 31, the low-pass filter 32, and (a part of) the gain calculation unit 34. .

It is a figure which shows the structure of the automatic sound volume correction apparatus concerning the 1st Embodiment of this invention. It is a figure which shows the structural example of a hold / release part. It is a figure for demonstrating the detailed operation | movement of the hold / release process in a hold / release part. It is a figure for demonstrating forward masking. It is a figure which shows the structure of the automatic sound volume correction apparatus concerning the 2nd Embodiment of this invention. It is a figure which shows the structure of the conventional automatic volume correction apparatus.

Explanation of symbols

1, 2, 3, ... automatic volume correction device, 11 ... absolute value conversion unit, 11L ... L-side absolute value conversion unit, 11R ... R-side absolute value conversion unit, 12 ... hold / Release part, 12L ... L side hold / release part, 12R ... R side hold / release part, 13 ... Maximum value detection part, 14, 31 ... Square calculation part, 15, 32 ... Low pass filter (LPF), 16... Release section, 17... Gain calculation section, 18... Attack section, 21... Input signal delay section, 21 L. ..R side input signal delay unit, 22... Output signal amplitude adjustment unit, 22L... L side output signal amplitude adjustment unit, 22R... R side output signal amplitude adjustment unit, 30. Estimator, 33 ... Logarithmic converter, 34 ... Gain calculator, 35 ... Reverse pair Conversion unit, 36 ... attack / release unit, 41A ... comparator, 41B ... hold section, 41C ... Hold canthaxanthin, 42 ... delay unit, 43 ... damping unit

Claims (5)

An automatic volume correction apparatus having an RMS calculation unit for calculating an RMS value which is a root mean square value of an input signal, and calculating a gain of the input signal based on the RMS value,
The peak point of the input signal is detected, the value of the peak point is held as a hold value, the hold value is output as it is to the RMS calculation unit during a predetermined hold period, and after the hold period has elapsed, A hold / release unit that performs a hold / release process to output to the RMS calculation unit while attenuating the hold value at a predetermined attenuation rate;
An automatic sound volume correction device comprising: The hold / release part is
When the input signal and the hold value signal are compared and the signal with the higher level is held and output, and the level of the input signal becomes smaller than the hold value (input signal <hold value) A maximum value holding unit that counts a hold period of a predetermined time,
A delay unit that outputs the maximum value holding unit with a predetermined time delay;
During the hold period, the output signal of the delay unit is output as it is to the maximum value holding unit as a hold value signal, and a predetermined attenuation factor is applied to the output signal of the delay unit after the hold period has elapsed. An attenuation unit that generates a signal of a new holding value and outputs the signal to the maximum value holding unit;
The automatic sound volume correction device according to claim 1, further comprising: An input signal delay unit that outputs the input signal with a predetermined time delay;
An absolute value converter for converting an input signal into an absolute value signal;
A hold / release unit that performs the hold / release process on the output signal from the absolute value conversion unit;
A squaring unit that squares the signal output from the hold / release unit;
A low-pass filter for performing filtering of the signal squared by the square calculation unit and calculating an estimated value of the mean square value of the amplitude of the input signal;
A logarithmic converter that converts the output signal of the low-pass filter into a logarithmic domain signal;
A gain calculation unit for performing gain calculation based on the signal converted into the logarithmic domain by the logarithmic conversion unit;
An inverse logarithm conversion unit that inversely converts the gain of the logarithmic region calculated by the gain calculation unit from a logarithmic region to a gain of a linear region;
An attack / release section for smoothing the gain by an attack / release process;
An output signal amplitude adjusting unit that adjusts and outputs the amplitude of the signal by multiplying the signal output from the input signal delay unit by the gain smoothed by the attack / release unit;
The automatic sound volume correction device according to claim 1 or 2, further comprising: An L-side input signal delay unit that outputs the input signal of the left channel of the stereo signal with a predetermined time delay;
An R-side input signal delay unit for delaying and outputting an input signal of the right channel of the stereo signal for a predetermined time;
An L-side absolute value converter that converts the input signal of the left channel into an absolute value signal;
An R-side absolute value converter that converts the input signal of the right channel into an absolute value signal;
An L-side hold / release unit that performs the hold / release process on the output signal from the L-side absolute value conversion unit;
An R-side hold / release unit that performs the hold / release process on an output signal from the R-side absolute value conversion unit;
The maximum value that takes the maximum value of the left channel signal that has been held / released by the L side hold / release unit and the left channel signal that has been held / released by the R side hold / release unit A detection unit;
A square computing unit that squares the signal output from the maximum value detecting unit;
A low-pass filter for performing filtering of the signal squared by the square calculation unit and calculating an estimated value of the mean square value of the amplitude of the input signal;
A gain calculation unit for performing gain calculation based on the output of the low-pass filter;
A smoothing unit for smoothing the gain;
An L-side output signal amplitude adjusting unit that adjusts the amplitude of the signal by multiplying the signal output from the L-side input signal delay unit by the smoothed gain, and an output;
An R-side output signal amplitude adjusting unit that adjusts the amplitude of the signal by multiplying the signal output from the R-side input signal delay unit by the smoothed gain;
The automatic sound volume correction device according to claim 1 or 2, further comprising: The hold period in the hold / release part is approximately 5 msec,
The attenuation rate d of the attenuation unit is
d = 1−e ^X where X = −1 / (Tr × FS),
Here, Tr = 20 msec, FS is the sampling rate,
The automatic sound volume correction device according to claim 2, wherein

Images