JP5841405B2 - Dynamic range expansion device - Google Patents

Description

  The present invention relates to a dynamic range expansion device, and more specifically, when an audio signal that has been flattened by compressing a dynamic range is output from a speaker or the like, the sharpness and presence of the output sound can be expressed. The present invention relates to a dynamic range expansion device capable of improving sound quality.

  In general, the range of volume levels of musical instruments and vocals played in concert halls and recording studios is very large. For this reason, when recording a musical instrument or vocal sound with a very large volume level on the recording device, it is possible that an excessive sound exceeding the dynamic range of the recording device may be input to the recording device. It was.

  Here, the dynamic range means the ratio between the minimum value and the maximum value of the identifiable signal. For example, when the resolution is defined as 16 bits as in a CD, the dynamic range is 96 dB, and when the resolution is defined as 24 bits as in a DVD, the dynamic range is 144 dB.

  If sound exceeding the dynamic range is recorded on the recording device, large distortion occurs in the recorded sound, resulting in deterioration of sound quality. As a means for solving such a problem, a compressor and a limiter are generally applied (for example, see Non-Patent Document 1 and Non-Patent Document 2).

  FIG. 16A shows an input / output operation example showing the relationship between the signal level of the input signal to the compressor and the output level of the output signal, and FIG. 16B shows the signal level and output signal of the input signal to the limiter. An input / output operation example showing the relationship with the output level is shown.

  The compressor functions as a variable gain amplifier. As shown in FIG. 16A, the compressor sets the threshold level of the output signal relative to the signal level of the input signal as compared with the level before the threshold level exceeds a preset threshold level. It has a role of suppressing the signal level of the output signal by lowering the value after exceeding. That is, the compressor constantly monitors the audio signal and automatically controls the gain so that the sound does not become louder than the set threshold level.

  On the other hand, the limiter functions as a compressor having a high compression ratio. The limiter has a role of keeping the output level constant even when the signal level of the input signal greatly exceeds the threshold level.

  Both the compressor and the limiter have a role of automatically suppressing a portion with a high gain (signal level) and keeping it within a predetermined dynamic range. However, the compressor gently suppresses (decreases) the increase of the input signal level exceeding the threshold level, and the limiter cuts the signal exceeding the set threshold level. This is a big difference.

FIG. 17 is a diagram illustrating an example of a signal waveform of an audio signal recorded on a CD. As described above, the CD has a dynamic range of 96 dB, and the signal level of the audio signal whose dynamic range does not fit within 96 dB is suppressed. A portion surrounded by a broken line C in FIG. 17 corresponds to a portion where the signal level is high, and the maximum value of the amplitude within the range of the audio signal is normalized to ± 1 by the compressor and the limiter (clipped). ) The minimum resolution in FIG. 17 is 1 / (2 ^(16-1) ), that is, 3.05 × 10 ⁻⁵ .

  In particular, the portions indicated by the broken lines A and B in FIG. 17 have a rectangular shape in which the amplitude of the audio signal recorded on the CD is suppressed to ± 1. For this reason, the amplitude of the full-scale signal is limited to ± 1 in the rectangular portion, and the dynamic range of the sound source is compressed.

Iwamiya Soichiro, “Basics and Applications of the Newest Sounds to Understand the Illustration”, Hidekazu System, March 2011, p. 242-243 “The Strongest Compressor Manual”, p.5-8, [online], Soundhouse Co., Ltd. [searched on August 10, 2011], <URL: http://www.soundhouse.co.jp/download/sonota /comp.pdf>

  By the way, a CD sound source having such a compressed dynamic range is reproduced (output) from the speaker after the volume is increased by volume setting, or is reproduced (output) from the speaker via a power amplifier that amplifies the CD sound source. There are many cases to do. When the volume of the output sound is increased by the volume setting or the power amplifier in this way, the signal output is already suppressed and flattened in the state recorded on the CD (the dynamic range is compressed, the clip Therefore, there is a problem that the sharpness of the sound is greatly reduced and the sense of reality is also lowered.

  Further, when the audio signal is clipped with an amplitude of ± 1 as shown by a broken line A and a broken line B in FIG. 17, the signal waveform becomes rectangular, and unnecessary harmonics are easily generated. Therefore, there is a problem that sound quality is deteriorated. For this reason, there is an increasing demand for extending the dynamic range of the output sound of the audio signal that has been flattened by compressing the dynamic range.

  On the other hand, when the dynamic range is not compressed, there is a problem that the sound quality may be deteriorated if the dynamic range is expanded.

  Furthermore, in audio distribution services such as Internet radio, the dynamic range may be compressed even if the signal is less than full scale (when the output range of speakers, etc. is sufficient). possible. As a result of dynamic range expansion processing for signals that are the same song but less than full scale and signals that are full scale, the variation shape of the complemented signal differs due to the difference in amplitude scale Care should be taken so that nothing happens.

  The present invention has been made in view of the above problems, and even if the audio signal is less than full scale with respect to the output sound of the audio signal whose dynamic range has been compressed and flattened, the signal is not full. It is an object of the present invention to provide a dynamic range expansion device capable of improving the sound quality by expressing sharpness and presence even if the signal satisfies the scale.

In order to solve the above problem, a dynamic range expansion device according to the present invention includes first envelope detection means for detecting a first envelope based on a maximum value change state of an audio signal output from a sound source, Average value calculating means for calculating an average amplitude value for a predetermined time in the first envelope detected by the first envelope detecting means as an amplitude average value Y _mean , and detected by the first envelope detecting means Further, a maximum value calculating means for calculating the maximum value of the amplitude of the first envelope at the predetermined time as an amplitude maximum value Y _max , an amplitude average value Y _mean calculated by the average value calculating means, and the maximum value calculating means based the calculated maximum amplitude value Y _max, the detection gain determining means for determining a detection gain, as the lower limit of the detection gain which has been determined by the detection gain determination unit, the sound source A second envelope detecting means for detecting a second envelope by obtaining a maximum value change state of the audio signal output from the signal, and decibel transforming the second envelope detected by the second envelope detecting means. A decibel transforming means, a differentiating means for generating a change amount signal indicating a change amount of the second envelope by differentiating the second envelope transformed by the decibel transforming means, and the differentiating means Inversion means for inverting the signal by multiplying the generated change amount signal by −1, and linear conversion for generating a detection signal by converting the change amount signal inverted by the inversion means into a linear signal Means, a complementary signal generating means for generating a complementary signal by subtracting a signal obtained by multiplying the audio signal by the detection signal from the audio signal, and the complementary signal Complementary signal generated by formation means, and an adding means for adding to the audio signal, the detection gain determination means, the value of the ratio of the average amplitude value Y _mean for the amplitude maximum value Y _max is 0. The detection gain is determined to be 0 [dB] when smaller than a first threshold value set to a value smaller than 5, and when the ratio value is the first threshold value, the detection gain is set to 10 log Y _max ² [DB], and when the ratio value is equal to or greater than the first threshold value and equal to or less than a second threshold value set to a value greater than 0.5, the ratio value is equal to the first threshold value. When the detection gain is determined to be a value reduced from 10 logY _max ² [dB] in an inverse proportion to the increase from the threshold value to the second threshold value, and the ratio value is greater than the second threshold value Reduced detection gain The value of the second threshold value in is determined as the detection gain.

  The dynamic range extending apparatus according to the present invention detects the second envelope based on the maximum value change state of the audio signal in the second envelope detecting means, and the change amount indicating the change amount of the second envelope by the differentiating means. Generate a signal.

  For this reason, when the maximum value change in the second envelope is large, the change amount of the change amount signal is large, and when the maximum value change in the second envelope is small, the change amount of the change amount signal is small. . The inversion means inverts the change amount signal indicating the change amount of the second envelope and then multiplies the audio signal, and subtracts the multiplied signal from the audio signal to compress and flatten it. It is possible to generate a complementary signal for extending the dynamic range of the audio signal.

  Therefore, by adding the complementary signal generated by the complementary signal generating means in the adding means to the audio signal, the dynamic range of the portion where the dynamic range is compressed and flattened can be expanded, and the output audio is output. It is possible to express the sharpness and presence of the signal.

  Furthermore, in the dynamic range expansion apparatus according to the present invention, the second envelope is obtained based on the maximum value change state of the audio signal, with the detection gain determined by the detection gain determination means as the lower limit value. That is, the lower limit value of the second envelope detection used for the dynamic range expansion process is determined based on the detection gain.

Here, when the value of the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max is smaller than the first threshold value set to a value smaller than 0.5, the detection gain is determined to be zero. That is, when the value of the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max is relatively small (less than 0.5), the lower limit value of the second envelope detection is determined to be 0 [dB]. become.

In general, when the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max is small in the audio signal, the possibility that the dynamic range is compressed is low. For this reason, when the value of the ratio of the amplitude average value Y _mean to the maximum amplitude value Y _max is relatively small (less than 0.5), the detection gain determination means is the lower limit value (detection gain) of the second envelope detection. Is determined to be 0 [dB], so that it is possible to prevent the second envelope from being excessively detected and an unnecessary complementary signal being added to the audio signal, thereby reducing the sound quality of the audio signal. Become.

On the other hand, when the value of the ratio of the average amplitude value Y _mean to the maximum amplitude value Y _max is equal to or larger than the first threshold value and smaller than the second threshold value set to a value larger than 0.5, the ratio As a detection gain, a value reduced from 10 logY _max ² [dB] is determined so that the value of is increased in inverse proportion to the increase from the first threshold value to the second threshold value.

In general, when the ratio of the average amplitude value Y _mean to the maximum amplitude value Y _max is large in the audio signal, there is a high possibility that the dynamic range is being compressed. For this reason, as the value of the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max increases from the first threshold value to the second threshold value, a value reduced from 10 log Y _max ² [dB] corresponding to the first threshold value is increased. By determining the lower limit value of the two envelope detection, an appropriate complementary signal can be added to the audio signal whose dynamic range has been compressed, and the sound quality of the audio signal can be improved.

Further, when the ratio value of the average amplitude value Y _{mean with} respect to the maximum amplitude value Y _max is larger than the second threshold value, the detection gain at the second threshold value reduced in inverse proportion to the ratio value is detected as the second envelope. By determining the lower limit value (detection gain), it is possible to prevent the maximum change in the second envelope from being excessively detected and adding an inappropriate complementary signal to the audio signal. .

Further, in the dynamic range expansion device according to the present invention, the dynamic range compression audio is performed in order to determine the degree of compression of the dynamic range based on the value of the ratio of the average amplitude value Y _mean to the maximum amplitude value Y _max . Whether the signal is full scale or less than full scale, the dynamic range can be expanded regardless of the output level. For this reason, it is possible to eliminate the possibility that the dynamic range expansion processing is not performed on a signal that is less than full scale and that has undergone dynamic range compression.

Further, in the above-described dynamic range expansion apparatus, the detection gain determination unit may determine that the ratio value is the first threshold value when the ratio value is equal to or greater than the first threshold value and equal to or less than the second threshold value. The detection gain is reduced from 10 logY _max ² [dB] to −30 + 10 log Y _max ² [dB] in an inverse proportion to the increase from 1 threshold to the second threshold, and the value of the ratio is the second When it is larger than the threshold value, the detection gain may be determined to be −30 + 10 logY _max ² [dB].

  For an audio signal recorded with a dynamic range compressed on a CD, DVD, or the like, the dynamic range is expanded after setting the detection range of the second envelope detection to -30 [dB] to 0 [dB]. Therefore, it is preferable to improve the sound quality effectively.

For this reason, when the value of the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max is equal to or larger than the first threshold value and equal to or smaller than the second threshold value, the ratio value is changed from the first threshold value to the second threshold value. When the detection gain is reduced from 10 log Y _max ² [dB] to −30 + 10 log Y _max ² [dB] in an inverse proportion to the increase to the threshold, and the ratio value is greater than the second threshold, the detection gain is By determining -30 + 10 log Y _max ² [dB], it is possible to perform a dynamic range expansion process suitable for an audio signal recorded on a CD, a DVD, or the like.

  In addition, the above-described dynamic range extending apparatus has gain adjusting means for performing gain adjustment by amplifying or attenuating the gain of the complementary signal generated by the complementary signal generating means, and the adding means is the gain adjusting means. The complementary signal that has been gain-adjusted according to the above may be added to the audio signal.

  Since the complementary signal is generated based on the maximum value change in the envelope, it is possible to expand the gain (amplitude) at the location where the dynamic range is compressed and flattened. However, the gain of the dynamic range that should be expanded differs for each audio signal that has been flattened by compressing the dynamic range, and even if the complementary signal is added to the audio signal as it is, sufficient dynamic range can be supplemented (expanded). There is also a possibility that the sound quality is deteriorated because there is not enough or the amount of complementation (expansion) of the dynamic range is too large.

  Therefore, according to the compression state of the audio signal, the gain adjustment means amplifies or attenuates the gain of the complementary signal in advance, thereby appropriately adjusting the dynamic range of the audio signal to which the complementary signal is added by the adding means. It becomes possible to expand.

  Further, in the above-described dynamic range expansion device, the differentiating unit performs a differentiation process by performing a filtering process of the audio signal using a high-pass filter, and the change time of the amount of change obtained by the filtering process is It may be adjusted by changing the set value of the normalized cutoff frequency in the high-pass filter.

  As described above, the differentiation process in the differentiation unit is realized by performing a filtering process using a high-pass filter on the audio signal. Further, by changing the set value of the normalized cutoff frequency of the high-pass filter, it is possible to adjust the change time of the change amount obtained by the filtering process.

  For example, the state where the dynamic range is compressed and the amplitude is flattened differs depending on the setting in the compressor and limiter, and the compression state of the dynamic range differs between CD and DVD. Therefore, by adjusting the normalized cutoff frequency according to the compression state of the dynamic range in the audio signal, it is possible to adjust the expansion time of the dynamic range suitable for the compression state of the dynamic range.

  In the dynamic range expansion apparatus according to the present invention, the second envelope is obtained based on the maximum value change state of the audio signal with the detection gain determined by the detection gain determination means as the lower limit value. That is, the lower limit value of the second envelope detection used for the dynamic range expansion process is determined based on the detection gain.

Here, when the value of the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max is smaller than the first threshold value set to a value smaller than 0.5, the detection gain is determined to be zero. That is, when the value of the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max is relatively small (less than 0.5), the lower limit value of the second envelope detection is determined to be 0 [dB]. become.

In general, when the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max is small in the audio signal, the possibility that the dynamic range is compressed is low. For this reason, when the value of the ratio of the amplitude average value Y _mean to the maximum amplitude value Y _max is relatively small (less than 0.5), the detection gain determination means is the lower limit value (detection gain) of the second envelope detection. Is determined to be 0 [dB], so that it is possible to prevent the second envelope from being excessively detected and an unnecessary complementary signal being added to the audio signal, thereby reducing the sound quality of the audio signal. Become.

On the other hand, when the value of the ratio of the average amplitude value Y _mean to the maximum amplitude value Y _max is equal to or larger than the first threshold value and smaller than the second threshold value set to a value larger than 0.5, the ratio As a detection gain, a value reduced from 10 logY _max ² [dB] is determined so that the value of is increased in inverse proportion to the increase from the first threshold value to the second threshold value.

In general, when the ratio of the average amplitude value Y _mean to the maximum amplitude value Y _max is large in the audio signal, there is a high possibility that the dynamic range is being compressed. For this reason, as the value of the ratio of the amplitude average value Y _mean to the amplitude maximum value Y _max increases from the first threshold value to the second threshold value, a value reduced from 10 log Y _max ² [dB] corresponding to the first threshold value is increased. By determining the lower limit value of the two envelope detection, an appropriate complementary signal can be added to the audio signal whose dynamic range has been compressed, and the sound quality of the audio signal can be improved.

Further, when the ratio value of the average amplitude value Y _{mean with} respect to the maximum amplitude value Y _max is larger than the second threshold value, the detection gain at the second threshold value reduced in inverse proportion to the ratio value is detected as the second envelope. By determining the lower limit value (detection gain), it is possible to prevent the maximum change in the second envelope from being excessively detected and adding an inappropriate complementary signal to the audio signal. .

Further, in the dynamic range expansion device according to the present invention, the dynamic range compression audio is performed in order to determine the degree of compression of the dynamic range based on the value of the ratio of the average amplitude value Y _mean to the maximum amplitude value Y _max . Whether the signal is full scale or less than full scale, the dynamic range can be expanded regardless of the output level. For this reason, it is possible to suppress an unnecessary dynamic range expansion process from being performed on an audio signal that is less than full scale and has not been subjected to dynamic range compression. .

It is the block diagram which showed schematic structure of the dynamic range expansion apparatus which concerns on embodiment. It is the block diagram which showed schematic structure of the music analysis part which concerns on embodiment. It is the graph which showed the relationship between the lower limit f ( _Ymean , _Ymax ) of envelope detection, and ratio ( _Ymean / _Ymax ) of the amplitude average value with respect to the amplitude maximum value of an envelope. (A) is the figure which showed the amplitude change of the audio signal before dynamic range compression is performed by the limiter process, (b) is the audio signal after the dynamic range compression is performed by the limiter process. It is the figure which showed the amplitude change. The graph in the case where the audio signal in which the dynamic range is greatly compressed and a large number of clips are generated is input to the dynamic range expansion apparatus according to the embodiment is shown in (a), and the amplitude change of the audio signal for 3 seconds is shown. (B) shows the envelope calculated by the first smoothing unit, the average amplitude of the envelope calculated by the average value holding unit, and the maximum amplitude of the envelope calculated by the second maximum value holding unit. (C) is a graph showing the ratio of the average amplitude value to the maximum amplitude value of the envelope (Y _mean / Y _max ). FIG. 5A is a graph in a case where an audio signal whose dynamic range is compressed smaller than that in FIG. 5 and a clip is scattered is input to the dynamic range expansion apparatus according to the embodiment. Is a graph corresponding to. FIG. 6 is a graph when an audio signal that has not been compressed in the dynamic range is input to the dynamic range expansion apparatus according to the embodiment, and corresponds to FIGS. It is the block diagram which showed schematic structure of the 1st fluctuation | variation detection part which concerns on embodiment. It is the block diagram which showed schematic structure of the 1st weighting part which concerns on embodiment. It is the table | surface which showed the operation parameter of each function part set in order to demonstrate the operation example of the dynamic range expansion apparatus which concerns on embodiment. (A) is the figure which showed the amplitude change state when the audio signal by which the dynamic range was compressed and the amplitude was flattened is input into the dynamic range expansion apparatus which concerns on embodiment, (b). The amplitude of the output signal of the dynamic range expansion apparatus when the lower limit value of detection is fixed to −30 [dB] in the second maximum value detection unit of the first fluctuation detection unit and the second fluctuation detection unit according to the embodiment. is a diagram showing a change state, (c), the output of the dynamic range expansion apparatus in the case of varying the lower limit of detection as -30 + 10logY _max ^{2 [dB]} in the second maximum value detection unit according to the embodiment It is the figure which showed the amplitude change state of the signal. (A) is the figure which showed the amplitude change state in case the audio signal in which the dynamic range is not compressed is input into the dynamic range expansion apparatus which concerns on embodiment, (b) is the 2nd maximum value. a view showing the amplitude variation state of the output signal in case of fixing the lower limit of detection in the detecting portion -30 [dB], (c), the variation lower limit value of the detection as -30 + 10logY _max ² [dB] It is the figure which showed the amplitude change state of the output signal in the case of making it. (A) is calculated by the first smoothing unit when the audio signal with the compressed dynamic range shown in FIG. 11 (a) is input to the dynamic range expansion apparatus and the signal shown in FIG. 11 (c) is output. The change of the envelope of the input signal, the average amplitude value obtained by the average value hold unit, the maximum amplitude value obtained by the second maximum value hold unit, and the ratio of the average amplitude value to the maximum amplitude value It is the figure which showed the result of having detected the state change of the value of ( _Ymean / _Ymax ) and the fluctuation state of the lower limit of envelope detection from the head of the music of the input audio signal for 30 seconds, (b) Is the same change as (a) in the case where an audio signal whose dynamic range is not compressed shown in FIG. 12 (a) is input to the dynamic range extender and the signal of FIG. 12 (c) is output. State shows the result of detection 30 seconds. (A) is the figure which showed the amplitude fluctuation state of the full-scale audio signal by which the dynamic range was compressed and the amplitude was flattened, (b) is the dynamic range was compressed and the amplitude was flattened It is the figure which showed the amplitude fluctuation state of the audio signal which is less than full scale, (c) is the audio signal by which the dynamic range of the audio signal shown by (a) was expanded with the dynamic range expansion apparatus which concerns on embodiment (D) is a diagram showing the amplitude variation state of the audio signal in which the dynamic range of the audio signal shown in (b) is expanded by the dynamic range expansion device. (A) shows the first smoothing in the case where the full-scale audio signal in which the dynamic range shown in FIG. 14 (a) is compressed is input to the dynamic range expansion apparatus and the signal of FIG. 14 (c) is output. The input signal envelope calculated by the unit, the amplitude average value obtained by the average value hold unit, the change state of the amplitude maximum value obtained by the second maximum value hold unit, and the amplitude average for the amplitude maximum value (B) shows the result of detecting the state change of the value ratio (Y _mean / Y _max ) and the fluctuation state of the lower limit value of the envelope detection for 30 seconds from the beginning of the music of the input audio signal. FIG. 14B shows a case where an audio signal whose dynamic range is less than full scale and whose dynamic range is compressed is input to the dynamic range expansion apparatus and the signal shown in FIG. 14D is output. Similar variation state (a), showing the results of detection 30 seconds. (A) shows an input / output operation example showing the relationship between the signal level of the input signal to the compressor and the output level of the output signal, and (b) shows the signal level of the input signal to the limiter and the output level of the output signal. It is the figure which showed the input / output operation example which showed this relationship. It is the figure which showed an example of the signal waveform of the audio signal recorded on CD.

  Hereinafter, an example of a dynamic range expansion apparatus according to the present invention will be described in detail with reference to the drawings.

[Dynamic range expansion device]
FIG. 1 is a block diagram showing a schematic configuration of a dynamic range expansion apparatus 1 according to the present embodiment. The dynamic range expansion device 1 has a role of expanding (complementing) the gain (signal level) of an audio signal that has been flattened by compression of the dynamic range.

  The dynamic range expansion apparatus 1 includes an LPF unit (low-pass filter unit) 10, an HPF unit (high-pass filter unit) 11, a first variation detection unit 12, a second variation detection unit 13, a first weighting unit 14, A second weighting unit 15, a synthesis unit (adding unit) 16, and a music analysis unit 17 are included.

  The dynamic range expansion device 1 receives an audio signal with a compressed dynamic range (for example, an audio signal with a dynamic range compressed with recording on a CD or the like). As shown in FIG. 1, the input audio signal is output to the LPF unit 10, the HPF unit 11, the synthesis unit 16, and the music analysis unit 17 of the dynamic range expansion apparatus 1.

[Music analysis section]
The music analysis unit 17 detects an envelope (first envelope) from an input audio signal (an audio signal on which dynamic range compression has been performed), and detects a first variation based on the detected envelope. It has a role of dynamically determining (setting) a lower limit value (detection gain) of detection of an envelope (second envelope) with respect to the unit 12 and the second variation detection unit 13.

  As shown in FIG. 2, the music analysis unit 17 includes a first maximum value detection unit (first envelope detection unit) 21, a first maximum value hold unit (first envelope detection unit) 22, and a first smoothing. Section (first envelope detection means) 23, monaural conversion section 24, average value hold section (average value calculation means) 25, second maximum value hold section (maximum value calculation means) 26, and detection gain calculation section (Detection gain determination means) 27.

  The first maximum value detection unit 21 detects the maximum value within a predetermined time (m samples) after converting the amplitudes of the input audio signals of the L channel and the R channel into absolute values, and detects each channel. It has a role to output every time. Accordingly, the first maximum value detection unit 21 updates the maximum value every predetermined time (m samples) and outputs it to the first maximum value hold unit 22. However, data sampled at a predetermined frequency is used as an input signal to the second maximum value detector 31 (see FIG. 8).

  The first maximum value hold unit 22 receives not only the latest maximum value received when the first maximum value detection unit 21 receives the maximum value of each channel updated every predetermined time (m samples). Among the maximum values (n maximum values) for the past n times including the maximum value, the maximum value indicating the largest value is obtained for each channel. Then, the first maximum value holding unit 22 outputs the obtained maximum value of each channel to the first smoothing unit 23 every predetermined time (m samples).

  That is, the first maximum value hold unit 22 acquires the maximum value of the largest value within a predetermined time (m samples) from the first maximum value detection unit 21 every predetermined time (m samples). A maximum value having the largest value among the maximum values for the past n times including the acquired maximum value (that is, the maximum value having the highest value among n × m samples) is detected, and a predetermined time (m samples) For each output to the first smoothing unit 23. The maximum value of the highest value among n × m samples output from the first maximum value hold unit 22 to the first smoothing unit 23 is held (held) for a predetermined time (m samples). It is output in the state.

  The first smoothing unit 23 has a role of smoothing the fluctuation of the maximum value (smoothing process) by integrating the signal of the maximum value of each channel acquired from the first maximum value holding unit 22. . By smoothing the signal acquired from the first maximum value hold unit 22 in this way, signal detection of the envelope (first envelope) for the audio signal is performed for each channel (L channel and R channel). The detection signal of the envelope (first envelope) has linear signal characteristics (characteristics indicating gain by amplitude). The detection signal of the envelope (first envelope) detected in this way is referred to as a first envelope signal.

  The monaural conversion unit 24 has a role of calculating an average of the first envelope signals of the L channel and the R channel obtained by the first smoothing unit 23. The calculated average envelope signal is output to each of the average value hold unit 25 and the second maximum value hold unit 26.

  The average value holding unit 25 calculates the average value (amplitude average value) of the first envelope signal acquired from the monaural conversion unit 24 while the audio signal is being input (when the audio signal is a specific music piece). Has a role of calculating and holding (holding / maintaining) until the music ends. That is, in a short time after the audio signal is input, the average value of the amplitude until the short time elapses from the input is calculated and held, and then after a while has passed since the input, The average value of the amplitude until a certain period of time elapses from the input is calculated and held.

  The second maximum value hold unit 26 uses the maximum value (amplitude maximum value) of the first envelope signal acquired from the monaural conversion unit 24 while the input audio signal is being input (audio signal). Has a role of calculating and holding until the music ends (predetermined time). Therefore, in a short time after the audio signal is input, the maximum value of the amplitude until the short time elapses from the input is calculated and held, and then after a while has elapsed from the input, The maximum value of the amplitude until a certain time elapses from the input is calculated and held.

  The average amplitude value of the envelope (first envelope) calculated in the average value hold unit 25 and the maximum amplitude value of the envelope (first envelope) calculated in the second maximum value hold unit 26 are detected. It is output to the gain calculation unit 27.

  The detection gain calculation unit 27 includes the average amplitude value of the envelope (first envelope) calculated by the average value hold unit 25 and the envelope (first envelope) calculated by the second maximum value hold unit 26. Role of dynamically calculating (determining) the lower limit [dB] of envelope detection (detection of the second envelope) in the first variation detector 12 and the second variation detector 13 using the maximum amplitude value. have.

・・・・式１
で求められる。 The average amplitude value of the envelope (first envelope) calculated by the average value hold unit 25 is Y _mean, and the maximum amplitude value of the envelope (first envelope) calculated by the second maximum value hold unit 26 is When _{Y max, f (Y mean,} Y max) which indicates the lower limit value [dB] of the envelope detection (detection of the second envelope) is

.... Formula 1
Is required.

When Y _mean / Y _max is a value smaller than 0.45 (first threshold value), the lower limit f (Y _mean , Y _max ) of envelope detection (detection of the second envelope) is 0 [dB]. When Y _mean / Y _max is a value greater than 0.60 (second threshold), the lower limit f (Y _mean , Y _max ) of envelope detection (second envelope detection) is -30 + 10 log Y _max ² [dB], and when Y _mean / Y _max is 0.45 or more and 0.60 or less, the lower limit f (Y of the envelope detection (detection of the second envelope)) _mean , Y _max ) is determined to be −200 · Y _mean / Y _max + 90 + 10 log Y _max ² [dB]. Note that 10 logY _max ² indicates a conversion formula (conversion relation) for converting the maximum value of the amplitude of the envelope (first envelope) into a decibel value.

FIG. 3 shows the lower limit value f (Y _mean , Y _max ) of envelope detection (detection of the second envelope) and the amplitude relative to the maximum amplitude value of the envelope (first envelope) based on the above-described equation 1. It is the graph which showed the relationship with ratio ( _Ymean / _Ymax ) of an average value. As shown in FIG. 3, when the ratio value (Y _mean / Y _max ) is smaller than 0.45, the lower limit f (Y _mean , Y _max ) of envelope detection (second envelope detection) is 0. [DB] is maintained (determined / set), but when the ratio value (Y _mean / Y _max ) becomes 0.45, the lower limit f (Y _mean , envelope detection (detection of the second envelope)) Y _max) is Y value obtained by substituting a 0.45 to _mean / Y _max) [dB] in 10logY _max ² values _{(-200 · Y mean / Y max} + 90 + 10logY max 2. Thereafter, the lower limit f (Y _mean , Y _max ) of envelope detection (detection of the second envelope) is inversely proportional to the ratio value (Y _mean / Y _max ) increasing from 0.45 to 0.60. ) Decreases and the ratio value (Y _mean / Y _max ) becomes 0.60 or more, the lower limit f (Y _mean , Y _max ) of envelope detection is a value of −30 + 10 log Y _max ² (− 200 · Y _mean / Y _max + 90 + 10 log Y _max ² is a value obtained by substituting 0.60 into Y _mean / Y _max ) [dB] (determined / set).

Thus, the ratio values (Y _mean / Y _max ) of 0.45 and 0.60 become threshold values (first threshold value: 0.45, second threshold value: 0.60), and the ratio value (Y _mean / Y _max ) is between 0.00 and 0.45, the lower limit f (Y _mean , Y _max ) of envelope detection (detection of the second envelope) is maintained at 0 [dB] (decision / setting) When the ratio value (Y _mean / Y _max ) is between 0.45 and 0.60, the lower limit f (Y _mean , Y _max ) of envelope detection (detection of the second envelope) is the ratio value. When it is reduced in inverse proportion to (Y _mean / Y _max ) and the ratio value (Y _mean / Y _max ) is 0.60 or more, the lower limit f of envelope detection (detection of the second envelope) (Y _mean, Y _max) is maintained at ^{_{-30 + 10logY max 2 [dB]}} .

  FIG. 4A is a diagram showing an amplitude change of an audio signal before the dynamic range is compressed by the limiter process, and FIG. 4B is an audio after the dynamic range is compressed by the limiter process. It is the figure which showed the amplitude change of the signal.

  As shown in FIG. 4 (a), in the audio signal before the dynamic range compression is performed, the amplitude fluctuation is large, so that the difference between the maximum amplitude value and the average amplitude value is large. However, as shown in FIG. 4B, in the audio signal after the dynamic range compression is performed by the limiter process, the maximum amplitude value is limited by the limiter process and a clip is generated. Therefore, the maximum amplitude value is low. It can be suppressed.

Further, since the maximum amplitude value is suppressed, the average amplitude value shows a relatively high value, and the difference between the maximum amplitude value and the average amplitude value becomes small. Therefore, the ratio of the average amplitude value to the maximum amplitude value (the maximum amplitude value of the first envelope) between the audio signal before the dynamic range compression and the audio signal after the dynamic range compression are performed. Compared to the ratio of the amplitude average value to Y (corresponding to the value of Y _mean / Y _max ), the ratio value is smaller before the dynamic range compression is performed, and after the dynamic range compression is performed. There is a tendency that the ratio value becomes larger.

  Accordingly, when the ratio of the average amplitude value to the maximum amplitude value is large, it can be determined that the audio signal is highly likely to be compressed in the dynamic range. When the value of the ratio is small, it can be determined that the audio signal is less likely to be compressed in the dynamic range.

  FIG. 5 is a graph when an audio signal in which a large dynamic range is compressed and a large number of clips are generated is input to the music analysis unit 17. FIG. 6 is a graph in the case where an audio signal in which a dynamic range is compressed smaller than that in FIG. 5 but a clip is scattered is input to the music analysis unit 17. FIG. 7 is a graph when an audio signal not subjected to dynamic range compression is input to the music analysis unit 17.

Specifically, in FIGS. 5 to 7, (a) is a graph showing an amplitude change of an audio signal for 3 seconds. (B) is calculated by the first smoothing unit 23, the average amplitude value of the envelope (first envelope) calculated by the average value holding unit 25, and the second maximum value holding unit 26. It is the graph which showed the change of the amplitude maximum value of the envelope (1st envelope) to be performed. (C) is the graph which showed the value of the ratio ( _Ymean / _Ymax ) of the amplitude average value with respect to the amplitude maximum value of an envelope (1st envelope).

Since the audio signal shown in FIG. 5A has a large dynamic range compression, the amplitude is almost continuously limited (maintained) to ± 1, and clips are continuously generated. For this reason, as shown in FIG. 5B, the amplitude value and the maximum amplitude value of the envelope (first envelope) maintain a value of about 1 (amplitude), and the average amplitude value is also the maximum amplitude value. It shows a value close to. Accordingly, as shown in FIG. 5C, the ratio (Y _mean / Y _max ) of the amplitude average value to the amplitude maximum value of the envelope (first envelope) is high (about 0.9). The value is shown.

On the other hand, since the audio signal shown in FIG. 6A has a small dynamic range compression, only a part of the amplitude is limited (maintained) to ± 1, and a clip is scattered. . For this reason, as shown in FIG.6 (b), the amplitude value of an envelope (1st envelope) is in the state fluctuate | varied in the location where the clip does not generate | occur | produce. Accordingly, as shown in FIG. 6C, the ratio (Y _mean / Y _max ) of the average amplitude value to the maximum amplitude value of the envelope (first envelope) (Y _mean / Y _max ) is shown in FIG. Compared to this, the value is about 0.7.

Further, since the audio signal shown in FIG. 7A is not compressed in the dynamic range, the maximum value of the amplitude is limited to a range of ± 1, and no clip is generated. For this reason, as shown in FIG. 7 (b), the amplitude value of the envelope (first envelope) is largely changed compared to FIGS. 5 (b) and 6 (b), and the amplitude average is maintained. The value is also lower than those in FIGS. 5B and 6B. Therefore, as shown in FIG. 7C, the ratio (Y _mean / Y _max ) of the average amplitude value to the maximum amplitude value of the envelope (first envelope) (Y _mean / Y _max ) is as shown in FIG. Compared to FIG. 6C, the value is reduced to about 0.3.

  As shown in FIGS. 5 to 7, the larger the compression of the dynamic range, the larger the ratio of the average amplitude value to the maximum amplitude value, and the smaller the dynamic range compression, the smaller the amplitude relative to the maximum amplitude value. When the ratio of the average values is reduced and the dynamic range is not compressed, the ratio of the average amplitude value to the maximum amplitude value is low.

For this reason, the music analysis unit 17 sets the lower limit of envelope detection (detection of the second envelope) within a certain range according to the value of the ratio (Y _mean / Y _max ) (in the above-described equation 1, 0 to −30 + 10 log Y _max ² [dB]). For example, when the value of the ratio (Y _mean / Y _max ) is high and there is a high possibility that compression of the dynamic range is being performed, the lower limit value of the envelope (second envelope) for expanding the dynamic range is set. It is set (determined) to a low value so that the dynamic range of the audio signal where clipping can occur is effectively expanded.

On the other hand, when the value of the ratio (Y _mean / Y _max ) is low and there is a high possibility that the dynamic range is not compressed, the lower limit value of the envelope (second envelope) for expanding the dynamic range By setting (determining) to a high value, it is possible to prevent the dynamic range of an audio signal in which clipping has not occurred from being excessively expanded.

Thereafter, the value of the lower limit f (Y _mean , Y _max ) of the envelope detection (detection of the second envelope) calculated by the detection gain calculation unit 27 of the music analysis unit 17 is the first variation detection unit 12 and the first variation detection unit 12. The output is output in real time (continuously) to each second maximum value detection unit 31 (see FIG. 8) of the two fluctuation detection unit 13.

[LPF part and HPF part]
The LPF unit 10 has a role of extracting a low frequency component of the input audio signal by performing a low-pass filtering process on the input audio signal (an audio signal subjected to dynamic range compression). ing. On the other hand, the HPF unit 11 has a role of extracting a high frequency component of the input audio signal by performing a high-pass filtering process on the input audio signal.

[First variation detection unit and second variation detection unit]
The first fluctuation detection unit 12 obtains an envelope of the low-frequency audio signal input from the LPF unit 10, and detects the amount by which the obtained envelope (second envelope) changes from a small signal to a large signal. And output to the first weighting unit 14. Similarly, the second fluctuation detection unit 13 obtains an envelope (second envelope) of the high frequency audio signal input from the HPF unit 11, and a signal with a small obtained envelope (second envelope). The amount of change from 1 to a large signal is detected as a detection signal and output to the second weighting unit 15.

  FIG. 8 is a block diagram showing a schematic configuration of the first fluctuation detecting unit 12. As shown in FIG. 8, the first fluctuation detecting unit 12 includes a second maximum value detecting unit (second envelope detecting unit) 31, a third maximum value holding unit (second envelope detecting unit) 32, 2 smoothing unit (second envelope detection unit) 33, decibel conversion unit (decibel conversion unit) 34, differentiation unit (differentiation unit) 35, inversion unit (inversion unit) 36, level limiting unit 37, linear And a conversion unit (linear conversion means) 38.

  The second maximum value detection unit 31 serves to detect and output the maximum value within a predetermined time (m samples) after the amplitude of the audio signal from which the low frequency component has been extracted by the LPF unit 10 is converted to an absolute value. Have. The second maximum value detection unit 31 performs the same process as the first maximum value detection unit 21. Accordingly, in the second maximum value detection unit 31, as in the first maximum value detection unit 21, the maximum value is updated every predetermined time (m samples) and output to the third maximum value hold unit 32. . However, data sampled at a predetermined frequency is used as an input signal to the second maximum value detector 31.

  In the second maximum value detection unit 31 or the third maximum value hold unit 32, a detection range in maximum value detection is set in advance, and detection of a signal level below the set detection range (below a predetermined value) is performed. Not done. Here, the lower limit value (detection gain) of envelope detection (detection of the second envelope) is input to the second maximum value detection unit 31 from the detection gain calculation unit 27 of the music analysis unit 17 in real time. Based on the lower limit value of the envelope detection (detection of the second envelope) acquired from the detection gain calculation unit 27, the second maximum value detection unit 31 sets the lower limit value of the set detection range in real time (dynamically). ) Change the maximum value.

  When the second maximum value detecting unit 31 receives the maximum value updated every predetermined time (m samples), the third maximum value holding unit 32 not only receives the latest maximum value but also the maximum value. Among the maximum values (n maximum values) for the past n times including the maximum value indicating the largest value. Then, the third maximum value holding unit 32 outputs the obtained maximum value to the second smoothing unit 33 every predetermined time (m samples).

  Here, the third maximum value hold unit 32 performs the same processing as the first maximum value hold unit 22. For this reason, the third maximum value hold unit 32 also, like the first maximum value hold unit 22, within a predetermined time (m samples) every predetermined time (m samples) from the second maximum value detection unit 31. The maximum value of the largest value is acquired, and further, the maximum value having the largest value among the past n maximum values including the acquired maximum value (that is, the highest value among the n × m samples). Is output to the second smoothing unit 33 every predetermined time (m samples). Note that the maximum value of the highest value among n × m samples output from the third maximum value hold unit 32 to the second smoothing unit 33 is held (held) for a predetermined time (m samples). It is output in the state.

  The second smoothing unit 33 has a role of smoothing the fluctuation of the maximum value (smoothing process) by integrating the signal of the maximum value acquired from the third maximum value holding unit 32. The second smoothing unit 33 also performs the same process as the first smoothing unit 23. Since the second smoothing unit 33 smoothes the signal acquired from the third maximum value holding unit 32, it is possible to detect the envelope (second envelope) signal for the low-frequency audio signal.

  The decibel conversion unit 34 has an envelope signal (second envelope: the second envelope signal obtained by performing the smoothing process in the second smoothing unit 33). Has a role of converting into a decibel envelope signal (envelope signal having a signal characteristic whose gain is expressed in decibels). The signal converted into the decibel envelope signal by the decibel converter 34 is output to the differentiator 35.

  The differentiating unit 35 has a role of performing change amount detection by differentiating the decibel envelope signal (second envelope signal) acquired from the decibel transform unit 34. A signal obtained by differentiating the envelope signal (second envelope signal) by the differentiating unit 35 shows a high value when the amount of change in gain (decibel value) is large, and the amount of change is small. In some cases, it will show a low value. The signal subjected to differentiation processing by the differentiating unit 35 is output to the inverting unit 36.

  The differentiation process in the differentiation unit 35 is realized by performing a filtering process using a high-pass filter on the audio signal from which the low-frequency component has been extracted by the LPF unit 10. Further, it is possible to adjust the change time of the change amount obtained by the filtering process by changing the set value of the normalized cutoff frequency in the high-pass filter.

  For example, the state in which the dynamic range is compressed and the amplitude is flattened differs depending on the settings in the compressor and limiter, and as described above, the compression state of the dynamic range differs between CD and DVD. . Therefore, by adjusting the normalized cutoff frequency according to the compression state of the dynamic range in the audio signal, it is possible to adjust the expansion time of the dynamic range suitable for the compression state of the dynamic range.

  The inverting unit 36 has a role of inverting the signal by multiplying the signal (change amount signal) acquired from the differentiating unit 35 by -1. Further, the level limiting unit 37 limits the minus side change amount of the signal inverted by the inversion unit 36, detects only the plus side change amount, and outputs the detected change amount to the linear conversion unit 38. The linear conversion unit 38 converts the signal detected only by the plus side change amount in the level limiting unit 37 into a linear signal and outputs the signal to the first weighting unit 14 as a detection signal.

  As described above, the first fluctuation detecting unit 12 detects the amount of change from a small signal to a large signal based on the envelope (second envelope) of the low frequency audio signal extracted by the LPF unit 10. Low-frequency detection signal), and the detected amount of detection signal increases as the detected signal changes greatly.

  Moreover, the 2nd fluctuation | variation detection part 13 is also provided with the function part similar to the 1st fluctuation | variation detection part 12 shown in FIG. The second fluctuation detection unit 13 detects the amount of change from a small signal to a large signal based on the envelope (second envelope) of the high frequency audio signal extracted by the HPF unit 11 as a detection signal (for high frequency). Detection signal), and the detected amount of detection signal increases as the detected signal changes greatly. Since the configuration and the role of the functional unit in the second variation detection unit 13 are the same as those of the first variation detection unit 12 shown in FIG.

  Note that the lower limit value (detection gain) of envelope detection is also input in real time from the detection gain calculation unit 27 of the music analysis unit 17 to the second maximum value detection unit 31 of the second variation detection unit 13. For this reason, the second maximum value detection unit 31 of the second fluctuation detection unit 13 also sets the lower limit value of the set detection range based on the lower limit value (detection gain) of the envelope detection acquired from the detection gain calculation unit 27. The maximum value is detected in real time (dynamically).

[First weighting unit and second weighting unit]
FIG. 9 is a block diagram illustrating a schematic configuration of the first weighting unit 14. As shown in FIG. 9, the first weighting unit 14 includes a multiplication unit (complementary signal generation unit) 41, a subtraction unit (complementary signal generation unit) 42, and a gain unit (gain adjustment unit) 43.

  The multiplication unit 41 has a role of performing multiplication processing of the low frequency audio signal input from the LPF unit 10 and the low frequency detection signal input from the first fluctuation detection unit 12. By the multiplication process of the multiplication unit 41, the signal level of the portion where the small signal changes to the large signal in the low frequency audio signal is amplified corresponding to the detection amount of the detection signal.

  The subtracting unit 42 has a role of subtracting a low-frequency audio signal that has been subjected to amplification processing corresponding to the detection amount by the multiplying unit 41 from the low-frequency audio signal input from the LPF unit 10. In this way, by performing the subtraction process in the subtracting unit 42, the signal level difference between the audio signal before amplification and the audio signal after amplification at the portion where the low signal changes from the small signal to the large signal is weighted. It can be obtained as a quantity. A signal relating to the weighting amount thus obtained (low-frequency weighting signal: low-frequency complementary signal before gain adjustment by the gain adjusting means) is output from the subtracting unit 42 to the gain unit 43.

  The gain unit 43 has a role of adjusting the weighting amount by amplifying and attenuating the gain for the low-frequency weighting signal and generating a low-frequency complementary signal (low-frequency complementary signal). . The generated low band complementary signal is output to the synthesis unit 16.

  Since the low-frequency weighting signal is generated based on the maximum value change in the envelope signal, it is possible to expand the gain (amplitude) at a location where the dynamic range is compressed and flattened. However, the gain of the dynamic range to be expanded differs for each audio signal that has been flattened by compressing the dynamic range, and even if the weighting signal for low frequency is added to the audio signal as it is, sufficient complementation of the dynamic range (expansion) ) Cannot be performed, or the amount of complementation (expansion) of the dynamic range is too large, leading to a decrease in sound quality.

  Therefore, in accordance with the compression state of the audio signal, the gain unit 43 performs gain amplification and attenuation on the low-frequency weighting signal in advance to generate a complementary signal in which the weighting amount is adjusted. It is possible to appropriately expand and adjust the dynamic range of the signal obtained by adding the complementary signal to the audio signal in the synthesis unit 16.

  As described above, in the first weighting unit 14, the subtracting unit 42 subtracts the low frequency audio signal subjected to the amplification process corresponding to the detection amount from the low frequency audio signal input from the LPF unit 10. Thus, the weighting amount in the low frequency audio signal is obtained, and the gain unit 43 adjusts the obtained weighting amount to generate the low frequency complementary signal.

  The second weighting unit 15 also includes a functional unit similar to the first weighting unit 14 shown in FIG. In the second weighting unit 15, the subtracting unit subtracts the high frequency audio signal subjected to the amplification process corresponding to the detection amount from the high frequency audio signal input from the HPF unit 11, thereby obtaining the high frequency audio signal. A high-frequency complementary signal is generated by obtaining a weighting amount in the audio signal and adjusting the obtained weighting amount. The configuration and the role of the functional unit in the second weighting unit 15 are the same as those of the first weighting unit 14 shown in FIG.

[Combining section]
The synthesizing unit 16 performs the low-frequency complementary signal obtained by the first weighting unit 14 on the audio signal input to the dynamic range extending apparatus 1 (the audio signal that has been flattened by compression of the dynamic range). And a process of adding the high-frequency complementary signal obtained by the second weighting unit 15. By performing the addition process in this manner, the low frequency range in which gain compensation is performed on the audio signal of the sound source flattened by compression of the dynamic range in accordance with the amount of change in the envelope (second envelope) of the signal Since the audio signal of the component / high frequency component is added, the dynamic range can be expanded, and the sharpness, presence and sound quality can be improved in terms of hearing.

  Note that the dynamic range expansion apparatus 1 according to the present embodiment does not perform dynamic range expansion processing on the mid-range component. The mid-range component generally has high auditory sensitivity and includes many voice components such as vocals. Therefore, expanding the dynamic range of the mid-range component gives the listener a sense of variation in audibility. Because there is a fear. However, the dynamic range extending apparatus 1 according to the present embodiment only shows the case where the dynamic range is expanded only for the low frequency component and the high frequency component as an example. It is also possible to adopt a configuration for performing range expansion processing. In addition, it is possible to extend the dynamic range by various combinations such as only the low frequency component, only the mid frequency component, only the high frequency component, only the low frequency component and the mid frequency component, and only the high frequency component and the mid frequency component. It is.

[Explanation of operation example]
Next, an operation example of the dynamic range expansion apparatus 1 described above will be described. FIG. 10 is a table showing operation parameters of each functional unit set for explaining an operation example in the dynamic range expansion apparatus 1. It is assumed that an audio signal recorded on a CD is input to the dynamic range expansion apparatus 1.

  As shown in FIG. 10, the LPF unit 10 and the HPF unit 11 use second-order Butterworth filters. The cutoff frequency of the LPF unit 10 is set to 200 Hz, and the cutoff frequency of the HPF unit 11 is set to 4 kHz. ing. In addition, since the audio signal input to the first variation detection unit 12, the second variation detection unit 13, and the music analysis unit 17 is an audio signal recorded on a CD, it is based on a sound source with a sampling frequency of 44.1 kHz. It becomes. For such an audio signal, the maximum value detection intervals of the first variation detection unit 12, the second variation detection unit 13, and the music analysis unit 17 are set to 64 samples.

Further, the detection range of the maximum value detection by the second maximum value detection unit 31 of the first variation detection unit 12 and the second variation detection unit 13 is dynamically between −30 + 10 log Y _max ² [dB] to 0 [dB]. The setting is variable. The lower limit value of the detection range is determined by the lower limit value (detection gain) of envelope detection (second envelope detection) set by the detection gain calculation unit 27. However, since the detection range in the maximum value detection is between −30 + 10 log Y _max ² [dB] and 0 [dB], an audio signal that is −30 + 10 log Y _max ² [dB] or less is set to −30 + 10 log Y _max ² [dB]. Is output. Therefore, the low-frequency complementary signal and the high-frequency complementary signal are not generated for the audio signal of −30 + 10 log Y _max ² [dB] or less. Further, the sampling frequency of maximum value detection and the interval of maximum value detection in the first maximum value detection unit 21 of the music analysis unit 17 are the same as those of the second maximum value detection unit 31.

  Also, in an audio signal with a sampling frequency of 44.1 kHz, if the maximum value detection interval is set to 64 samples, sampling of the audio signal that is held by the first to third maximum value hold units 22, 26, and 32 is performed. The frequency is 689 Hz (44,110 Hz ÷ 64 samples).

  Further, a first Butterworth low-pass filter is used for the first smoothing unit 23 of the music analysis unit 17 and the second smoothing unit 33 of the first variation detection unit 12 and the second variation detection unit 13, and the differentiation unit 35. A first-order Butterworth high-pass filter is used. In addition, the normalization cutoff frequency of the 1st smoothing part 23 of the music analysis part 17 and the 2nd smoothing part 33 in the 1st fluctuation | variation detection part 12 and the 2nd fluctuation | variation detection part 13 assumes that 689Hz is set to 1.0. Is set to a frequency (68.9 Hz) corresponding to 0.1.

  Further, the normalized cutoff frequency of the differentiating unit 35 in the first fluctuation detecting unit 12 is set to a frequency corresponding to 0.0171 when 689 Hz is 1.0, and the normalization cutoff frequency of the differentiating unit in the second fluctuation detecting unit 13 is set. The normalized cut-off frequency is set to a frequency corresponding to 0.0051 when 689 Hz is 1.0.

  By adjusting the cut-off frequency of the high-pass filter in the differentiating unit 35, it is possible to control the response time in the low-frequency detection signal and the high-frequency detection signal. For example, the higher the cutoff frequency, the longer the response time of the detection signal (the detection amount fluctuation time (reflection time) in the detection signal), and the generation time of the low-frequency complementary signal and the high-frequency complementary signal in the expansion of the dynamic range ( The increase reflection time of the audio signal increased by the complementary signal can be lengthened.

  Further, in the gain unit 43 of the first weighting unit 14, a gain amplification amount of 6 dB is set for the input low-frequency weighting signal in order to adjust the weighting amount. In the gain section of the second weighting section 15, a gain amplification amount of 0 dB is set for the input high-frequency weighting signal, and is set so as not to increase or decrease substantially.

  FIG. 11 to FIG. 15 are diagrams showing an operation example when the parameters of each functional unit are set as shown in FIG.

FIG. 11A is a diagram illustrating an amplitude change state when an audio signal whose dynamic range is compressed and whose amplitude is flattened is input to the dynamic range expansion apparatus 1. On the other hand, FIG. 11B shows the dynamic range when the lower limit value of detection is fixed to −30 [dB] in the second maximum value detector 31 of the first fluctuation detector 12 and the second fluctuation detector 13. The amplitude change state of the output signal of the expansion device 1 is shown. FIG. 11C shows the amplitude change state of the output signal of the dynamic range expansion apparatus 1 when the lower limit value of detection in the second maximum value detection unit 31 is changed as −30 + 10 logY _max ² [dB].

  In FIG. 11 (a), the audio signal is flattened, whereas in (b) and (c), the dynamic range is expanded and the flattened amplitude portion (clip) disappears, and the sense of incongruity is lost. It is possible to get no audio output. As described above, in the dynamic range expansion apparatus 1 according to the present embodiment, the low-frequency complementary signal and the high-frequency complementary signal are the low-frequency audio signal envelope (second envelope) and the high-frequency audio signal envelope. Since the signal waveform of the audio signal whose dynamic range is compressed approaches a rectangular shape and the amount of change increases, it is generated by differentiating each line (second envelope). The level of the complementary signal increases. For this reason, as the dynamic range compression is larger (the flattened portion), signal interpolation is actively performed to reduce the flat portion, and a complementary signal corresponding to the degree of compression is synthesized with the audio signal. It will be. Therefore, in any dynamic range compression state, it is possible to bring the dynamic range closer to the source sound before the dynamic range compression processing is performed.

  In addition, the dynamic range expansion processing in the dynamic range expansion apparatus 1 is performed using a low-frequency complementary signal and a high-frequency complementary signal obtained based on the envelope, so that the volume setting of the audio signal whose dynamic range is compressed can be increased or decreased. Regardless, it becomes possible to carry out effectively. Therefore, if the power amplifier or the speaker has sufficient reproduction capability, an audio signal whose dynamic range is compressed due to the limitation of the dynamic range in a recording medium such as a CD or DVD is increased after the volume setting volume is increased. Even if playback (output) from the speaker or playback (output) from the speaker via a power amplifier, etc., the dynamic range of the output signal is expanded, so that you can experience the sharpness and presence of the output sound. It becomes possible to suppress the generation of unnecessary harmonics.

  Note that in both audio signals of FIGS. 11B and 11C, the dynamic range is similarly expanded, and no harmonics are generated in the output sound.

On the other hand, FIG. 12A is a diagram illustrating an amplitude change state when an audio signal whose dynamic range is not compressed is input to the dynamic range expansion apparatus 1. FIG. 12B shows the amplitude change state of the output signal when the lower limit value of detection of the envelope (second envelope) in the second maximum value detector 31 is fixed to −30 [dB]. ) Shows the amplitude change state of the output signal when the lower limit of detection is changed as −30 + 10 logY _max ² [dB].

  In FIG. 12A, since the dynamic range of the audio signal is not compressed, the signal is not flattened. If such an audio signal is subjected to dynamic range expansion processing in a state where the lower limit of detection is fixed to −30 [dB], as shown in FIG. There is a possibility that the dynamic range without noise will be expanded and the sound quality will be reduced.

On the other hand, if the dynamic range expansion process is performed in a state where the lower limit of detection is changed as −30 + 10 logY _max ² [dB], the dynamic range is unnecessarily expanded as shown in FIG. Can be suppressed, and deterioration of sound quality can be suppressed. This is because, in the music analysis unit 17, the lower limit value of the envelope detection is changed according to the value of the ratio of the average amplitude value to the maximum amplitude value (Y _mean / Y _max ).

Specifically, when the value of the ratio (Y _mean / Y _max ) is high and there is a high possibility that the dynamic range is being compressed, an envelope for expanding the dynamic range (second envelope) By actively setting (determining) the lower limit value of the audio signal to a low value so as to effectively extend the dynamic range of the audio signal that can be clipped, the ratio (Y _mean / Y _max ) is reduced. If there is a high possibility that the dynamic range is not compressed, the music analysis unit 17 positively increases the lower limit value of the envelope (second envelope) for expanding the dynamic range to a high value. By setting (determining), it is possible to prevent the dynamic range of an audio signal that is not clipped from being excessively expanded.

Therefore, the dynamic range is compressed by changing the lower limit value of envelope detection (second envelope detection) according to the ratio of the average amplitude value to the maximum amplitude value (Y _mean / Y _max ). It is possible to effectively suppress the dynamic range of the audio signal not being unnecessarily expanded.

FIG. 13 is a diagram for explaining a variation state of the lower limit value of envelope detection set in the music analysis unit 17. 13A is calculated by the first smoothing unit 23 when the compressed audio signal shown in FIG. 11A is input to the dynamic range expansion apparatus 1 and the signal of FIG. 11C is output. The change state of the envelope of the input signal (first envelope), the amplitude average value obtained by the average value hold unit 25, and the amplitude maximum value obtained by the second maximum value hold unit 26 ((a ), The state change of the ratio of the amplitude average value to the maximum amplitude value (Y _mean / Y _max ) (see the middle graph of (a)), and envelope detection (detection of the second envelope) ) Of the lower limit value (see the lower graph of (a)) shows the result of detecting for 30 seconds from the beginning of the music of the input audio signal. FIG. 13B shows the same fluctuation state when the uncompressed audio signal shown in FIG. 12A is input to the dynamic range expansion apparatus 1 and the signal of FIG. (B) (see upper, middle, and lower graphs) for 30 seconds.

In the middle graph of FIG. 13A, as the ratio of the average amplitude value to the maximum amplitude value (Y _mean / Y _max ) exceeds 0.45 (after 20 seconds), the lower graph of FIG. FIG. 3 shows a state where the lower limit value of envelope detection is reduced. The fluctuation of the lower limit value of the envelope detection with respect to the value of this ratio (Y _mean / Y _max ) corresponds to the graph shown in FIG. On the other hand, in the middle graph of FIG. 13B, the value of the ratio of average amplitude values (Y _mean / Y _max ) is less than 0.45, so the lower graph of FIG. The lower limit of line detection is not shown.

  FIG. 14A shows the amplitude fluctuation state of a full-scale audio signal whose dynamic range is compressed and the amplitude is flattened, and FIG. 14B is a full scale whose dynamic range is compressed and the amplitude is flattened. The amplitude fluctuation state of the audio signal that is less than (less than full scale) is shown. FIG. 14C shows the amplitude variation state of the audio signal obtained by extending the dynamic range of the full-scale audio signal shown in FIG. The dynamic range expansion apparatus 1 expands the dynamic range of the audio signal of less than full scale shown in FIG.

  Specifically, FIG. 14A shows the amplitude fluctuation state of a full-scale audio signal recorded on a CD. FIG. 14B shows a signal that is less than full scale and that is obtained by multiplying the amplitude of the audio signal in FIG. Therefore, in the audio signal in FIG. 14A, the flattened amplitude limit value is ± 1, whereas in the audio signal in FIG. 14B, the flattened amplitude limit is set. The value is ± 0.01, which is one hundredth of the amplitude shown in FIG. It should be noted that the audio signal in FIG. 14A and the audio signal in FIG. 14B have different amplitude scales, but have a common amplitude change state and are limited in amplitude (amplitude ± 1 or amplitude ± 0.01) is partially flattened.

15 (a) is similar to FIG. 13 (a), the compressed full-scale audio signal shown in FIG. 14 (a) is input to the dynamic range expansion apparatus 1 and the signal shown in FIG. 14 (c). Is output, the envelope of the input signal calculated by the first smoothing unit 23, the average amplitude value obtained by the average value holding unit 25, and the maximum amplitude value obtained by the second maximum value holding unit 26. Change state (see the upper graph of (a)), change in the value of the ratio of the average amplitude value to the maximum amplitude value (Y _mean / Y _max ) (see the middle graph of (a)), and the envelope The fluctuation state of the lower limit of detection (see the lower graph of (a)) shows the result of detecting for 30 seconds from the beginning of the music of the input audio signal. 15B is the same as that in the case where the compressed audio signal of less than full scale shown in FIG. 14B is input to the dynamic range expansion apparatus 1 and the signal of FIG. 14D is output. The fluctuation state (see the upper, middle and lower graphs in (b)) is shown for 30 seconds.

When the compressed full-scale audio signal shown in FIG. 14A is input to the dynamic range extending apparatus 1 according to the present embodiment, the dynamic range is extended as shown in FIG. 14C. Thus, the amplitude flattened to ± 1 is corrected to ± 1 or more, and the dynamic range is expanded. At this time, in the middle graph of FIG. 15A, when the ratio of the amplitude average value to the maximum amplitude value (Y _mean / Y _max ) exceeds 0.45 (after 20 seconds), FIG. ) As shown in the lower graph, the lower limit value of the envelope detection is reduced. The fluctuation of the lower limit value of the envelope detection with respect to the value of this ratio (Y _mean / Y _max ) corresponds to the graph shown in FIG. 3 and Equation 1, but the audio signal input in FIG. The amplitude maximum value of 1 is 1 as shown in the upper part of FIGS. 14 (a) and 15 (a). Therefore, when the ratio value (Y _mean / Y _max ) exceeds 0.45, envelope detection is performed. When the lower limit value f (Y _mean , Y _max ) is about −10 [dB] (−200 · Y _mean / Y _max + 90 + 10 log Y _max ²⁾ , 0.45 is substituted for Y _mean / Y _max and Y _max (Value substituted with 1)).

On the other hand, when an audio signal less than the compressed full scale shown in FIG. 14B is input to the dynamic range extending apparatus 1 according to the present embodiment, as shown in FIG. Thus, the amplitude flattened to ± 0.01 is corrected to ± 0.01 or more, and the dynamic range is extended. At this time, in the middle graph of FIG. 15B, when the ratio of the average amplitude value to the maximum amplitude value (Y _mean / Y _max ) exceeds 0.45, the lower graph of FIG. As shown, the lower limit value of envelope detection tends to be gradually reduced after being reduced to −40 [dB] at once. The fluctuation of the lower limit value of the envelope detection with respect to the value of this ratio (Y _mean / Y _max ) corresponds to the graph shown in FIG. 3 and Equation 1, but the audio signal input in FIG. 14 is 0.01 as shown in the upper amplitude maximum values in FIGS. 14B and 15B, and the ratio value (Y _mean / Y _max ) is 0.45. The envelope detection lower limit value f (Y _mean , Y _max ) is about −40 [dB] (−200 · Y _mean / Y _max + 90 + 10 log Y _max ² , Y _mean / Y _{max is set} to 0.45) Substituted and reduced (varied) to a value obtained by substituting 0.01 for Y _max .

Thus, in the dynamic range expansion apparatus 1 according to the present embodiment, the lower limit value of envelope detection (second envelope detection) is varied according to the maximum amplitude value (according to the value of log Y _max ² ). Therefore, even when the input audio signal is less than full scale and the restricted amplitude value is low, the input audio signal is full scale and the restricted amplitude value is high. Even in this case, dynamic range expansion processing is appropriately performed according to the limited amplitude value (maximum amplitude value).

  For this reason, it is possible to eliminate the possibility that the dynamic range expansion processing is not performed on a signal that is less than full scale and that has undergone dynamic range compression.

  The dynamic range extending apparatus according to the present invention has been described with reference to an example in the embodiment. However, the dynamic range extending apparatus according to the present invention is not limited to the example shown in the above-described embodiment. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

  For example, in the dynamic range extending apparatus 1 according to the present embodiment, when the LPF unit 10 and the HPF unit 11 extract the low frequency component and the high frequency component from the input audio signal, respectively, the dynamic range is extended. However, the expansion of the dynamic range is not necessarily limited to the case where the low frequency component and the high frequency component are divided. In addition to dividing into high frequency components and low frequency components, it may be configured to divide into three or more bands and expand the dynamic range, respectively, or to divide the bands completely. Alternatively, the dynamic range may be extended.

DESCRIPTION OF SYMBOLS 1 ... Dynamic range expansion apparatus 10 ... LPF part 11 ... HPF part 12 ... 1st fluctuation | variation detection part 13 ... 2nd fluctuation | variation detection part 14 ... 1st weighting part 15 ... 2nd weighting part 16 ... Synthesis | combination part (adding means)
17 ... music analysis unit 21 ... first maximum value detection unit (first envelope detection means)
22 ... 1st maximum value hold part (1st envelope detection means)
23 ... 1st smoothing part (1st envelope detection means)
24 ... monaural conversion unit 25 ... average value hold unit (average value calculation means)
26 ... 2nd maximum value hold part (maximum value calculation means)
27: Detection gain calculation unit (detection gain determination means)
31 ... 2nd maximum value detection part (2nd envelope detection means)
32 ... 3rd maximum value hold part (2nd envelope detection means)
33 ... 2nd smoothing part (2nd envelope detection means)
34. Decibel conversion unit (decibel conversion means)
35 ... Differential part (differentiating means)
36 ... reversing part (reversing means)
37 ... level limiter 38 ... linear converter (linear converter)
41 ... Multiplication unit (complementary signal generating means)
42 Subtracting unit (complementary signal generating means)
43 ... Gain section (gain adjusting means)

Claims (4)

First envelope detecting means for detecting the first envelope based on the maximum value change state of the audio signal output from the sound source;
Average value calculating means for calculating an average amplitude value for a predetermined time in the first envelope detected by the first envelope detecting means as an amplitude average value Y _mean ;
Maximum value calculating means for calculating the maximum value of the amplitude of the first envelope detected by the first envelope detecting means at the predetermined time as an amplitude maximum value Y _max ;
Wherein on the basis of the amplitude maximum value Y _max is calculated by the amplitude average value Y _mean and the maximum value calculation means calculated by the average value calculating means, and the detection gain determining means for determining a detection gain,
A second envelope for detecting a second envelope by obtaining a maximum value change state of an audio signal satisfying a value higher than the detection gain determined by the detection gain determination means among the audio signals output from the sound source. Line detection means;
A decibel transforming means for decibel transforming the second envelope detected by the second envelope detecting means;
Differentiating means for generating a change amount signal indicating a change amount of the second envelope by differentiating the second envelope converted by the decibel conversion means;
Inversion means for inverting the signal by multiplying the change amount signal generated by the differentiating means by -1.
Linear conversion means for generating a detection signal by converting the change amount signal inverted by the inversion means into a linear signal;
Complementary signal generation means for generating a complementary signal by subtracting a signal obtained by multiplying the audio signal by the detection signal from the audio signal;
Adding means for adding the complementary signal generated by the complementary signal generating means to the audio signal;
The detection gain determining means includes
When the value of the ratio of the amplitude average value Y _{mean to} the amplitude maximum value Y _max is smaller than a first threshold set to a value smaller than 0.5, the detection gain is determined to be 0 [dB];
When the value of the ratio is the first threshold, the detection gain is determined to be 10 logY _max ² [dB],
When the ratio value is equal to or greater than the first threshold value and equal to or less than a second threshold value set to a value greater than 0.5, the ratio value ranges from the first threshold value to the second threshold value. A value reduced from the 10 log Y _max ² [dB] is determined as the detection gain so as to be inversely proportional to the increase,
When the value of the ratio is larger than the second threshold value, the ratio value is decreased in proportion to the increase from the first threshold value to the second threshold value, so that the value is less than 10 logY _max ² [dB]. A dynamic range expansion device, wherein the lowest value among the values is determined as the detection gain. The detection gain determining means includes
When the value of the ratio is greater than or equal to the first threshold and less than or equal to the second threshold, the ratio is increased in inverse proportion to increasing from the first threshold to the second threshold. Reducing the detection gain from 10 log Y _max ² [dB] to −30 + 10 log Y _max ² [dB],
2. The dynamic range expansion device according to claim 1, wherein when the value of the ratio is larger than the second threshold, the detection gain is determined to be −30 + 10 logY _max ² [dB]. Gain adjusting means for performing gain adjustment by amplifying or attenuating the gain of the complementary signal generated by the complementary signal generating means;
The dynamic range extending apparatus according to claim 1, wherein the adding unit adds the complementary signal whose gain is adjusted by the gain adjusting unit to the audio signal. The differentiating means performs a differentiation process by performing a filtering process of the second envelope transformed by the decibel transform means by using the high-pass filter,
4. The change time of the change amount obtained by the filtering process is adjusted by changing a set value of a normalized cutoff frequency in the high-pass filter. 5. The dynamic range expansion device as described in the paragraph.

Images