JP2007221216A - Mix-down method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a mix-down method and a mix-down apparatus capable of executing mix-down with an excellent sound volume balance. <P>SOLUTION: The mix-down apparatus multiplies a preset static gain value with a scale factor of a received acoustic signal to output a value of the multiplication result corresponding to each acoustic signal, and calculates a dynamic gain value corresponding to each acoustic signal by normalizing the respective values of the multiplication results, then multiplies the dynamic gain value by each normalized acoustic signal with a plurality of received acoustic signals and summates a plurality of multiplication results to calculate an output signal of an output system. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a mixdown method and a mixdown apparatus that dynamically determine the gain of each channel for mixdown used in digital audio.

  In recent years, in the audio reproduction in the AV (audio / visual) field, a multi-channel reproduction method in which a center channel and a surround channel are added to a conventional L / R channel stereo signal has been realized. When reproducing the original sound field in this multi-channel reproduction system, it is necessary to arrange at least one speaker behind or on the side of the viewer.

  However, in order to respond to the viewer's desire to reproduce a multi-channel playback system sound signal with two L / R channel speakers, it is necessary to convert the multi-channel sound signal into two channels.

  Therefore, it is necessary to mix down the multi-channel input sound source by converting the acoustic signal into a smaller number of channels than the number of channels on the input side and outputting it.

  As an example of this mixdown technique, Patent Document 1 reports a multichannel stereo downmixing apparatus.

According to Patent Document 1, whether the multi-channel acoustic signal to be mixed down is music or voice is added based on the added L / Rch signal by adding the input L / Rch signal. Judging. Based on the determination result, the gain coefficient of the volume adjustment unit is changed to a value that enables realism in the case of music playback and a value that can ensure clarity in the case of audio. Therefore, even in 2-channel speaker reproduction, there is an advantage that it is possible to realize optimum sound reproduction for each program source.
JP-A-6-165079

  As described above, according to Patent Document 1, it is determined whether music is sound or sound based on the input L / Rch signal, and the gain coefficient for volume adjustment is changed, so that music reproduction and sound can be performed. Appropriate sound field reproduction is realized.

  However, in Patent Document 1, since two gain coefficients for mixdown are prepared in advance and switched between music playback and voice playback, it is optimal for various acoustic signals. There was a problem that could not be adapted to. As a result, the switching time of the sound field reproduction became clear, and the volume balance between channels was poor, causing a sense of incongruity and fatigue.

  The present invention has been made in view of the above, and an object thereof is to provide a mixdown method and a mixdown apparatus capable of performing a mixdown with a good volume balance.

  In order to solve the above-mentioned problem, the invention according to claim 1 is a mixdown device that mixes a plurality of input sound signals and converts them into a number of output signals smaller than the number of sound signals. The gain determination means for determining the dynamic gain value of the sound signal based on the scale factor of the sound signal, and the sound signal and the dynamic gain value determined for the sound signal are multiplied to support a plurality of sound signals. And a mixdown means for outputting the sum of the multiplication result values as an output signal.

  According to a third aspect of the present invention, there is provided a mixdown method for mixing a plurality of input sound signals and converting them into a number of output signals smaller than the number of the sound signals in order to solve the above-mentioned problem. The gain determination step for determining the dynamic gain value of the acoustic signal based on the scale factor of the acoustic signal, and the acoustic signal multiplied by the dynamic gain value determined for the acoustic signal, to support a plurality of acoustic signals And a mixdown step of outputting the sum of the multiplication result values to be output as an output signal.

  According to the mixdown method and the mixdown apparatus of the present invention, the dynamic gain value is determined based on the scale factor of the input acoustic signal, so that the mixdown with a good volume balance can be performed.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(First embodiment)
MPEG-2 / 4 AAC is a multi-channel audio encoding method that achieves high sound quality and high compression rate by eliminating compatibility with MPEG-1 audio. The sampling frequency of the input corresponds to a very wide range from 8kHz to 96kHz. In addition, it can transmit up to 48 channels of audio signals, 15 LFE (Low Frequency Enhancement) channels, coupling channels, and general-purpose data streams.

  The audio decoding device is opposed to the audio encoding device, and is input via an encoded bit stream bs output from the audio encoding device, a bit stream bs stored in a DVD, for example, or the Internet. The decoded bit stream bs is decoded and reproduced as an audio signal.

  FIG. 1 is a block diagram showing a configuration for explaining a speech decoding apparatus applicable to the mixdown apparatus according to the first embodiment of the present invention. FIG. 2 is a diagram illustrating a mutual connection relationship between the gain determination unit 19, the time-frequency inverse conversion unit 21, and the mixdown unit 23. Note that the mixdown device referred to here is a device that mixes a plurality of input acoustic signals and converts them into a smaller number of output signals than the number of acoustic signals.

The speech decoding apparatus 11 includes a syntax decoding unit 13, a Huffman code decoder unit 15, an inverse quantization unit 17, a gain determination unit 19, a time-frequency inverse transform unit 21, and a mixdown unit 23. Further, each unit is a DSP ( Digital Signal Processor) and software processing.
The syntax decoding unit 13 separates a high-efficiency encoded audio stream and a band signal, normalization information, quantization accuracy information, and the like from the input encoded data string bs (bit stream) according to a prescribed grammar.

Note that the band signal is encoded after the original data is divided into a plurality of frequency bands and weighted according to human hearing for each band in the speech encoding apparatus. Further, the normalization information is information for aligning coefficient positions used in an inverse quantization process described later. The quantization accuracy information is information on the quantization accuracy (level) of the hierarchized DCT coefficient used in the inverse quantization process described later.
The Huffman code decoder unit 15 performs Huffman decoding on the input high-efficiency encoded audio stream h according to the encoding format obtained from the sub information, and outputs quantized spectra q1 to q6 and a scale factor.

  The inverse quantization unit 17 reproduces the original frequency signals f1 to f6 based on the decoded quantized spectra q1 to q6 and the scale factor.

The scale factor indicates the quantization width in the quantization. For example, in MPEG-2 / 4 AAC, a quantization spectrum q is given to a certain frequency signal f.
q = INT ((f ^ 3/4) / 2 ^ (sf / 4) + 0.4054) (Equation 1)
The integer sf at this time is called a scale factor. The scale factor is commonly used for a plurality of continuous frequency signals. Sf1 (1), sf1 (2). . . It shall be written as sf1 (N). In the present embodiment, description will be made using the lowest scale factor sfi (1).

  The scale factors sf1 (1) to sf6 (1) have independent values for each channel. In the case of encoding with a constant bit rate, the scale factor generally indicates a small value for a small volume, and the scale factor indicates a large value for a large volume. For example, sf1 (1) indicates a scale factor of 1 channel and sf6 (1) indicates a scale factor of 6 channels. Here, the channel number represents the position of the sound image. For example, 1 is the center channel, 2 is the left front channel, 3 is the right front channel, 4 is the left rear channel, 5 is the right rear channel, and 6 is the bass channel.

  The gain determination unit 19 determines dynamic gains g11 to g61 and g12 to g62 based on the scale factors sf1 (1) to sf6 (1).

  The time-frequency inverse transform unit 21 performs time-frequency inverse transform such as inverse modified discrete cosine transform (IMDCT). That is, the frequency signals f1 to f6 in each band are inversely transformed to synthesize the bands to obtain time-series acoustic signals i1 to i6.

  The mixdown unit 23 calculates the output signals o1 and o2 based on the acoustic signals i1 to i6 and the dynamic gains g11 to g61 and g12 to g62.

  FIG. 3 is a diagram illustrating a configuration of the gain determination unit 19. In FIG. 3, the gain determination unit 19 includes a weighter 31 and normalizers 33 and 35.

  The scale factors sf1 (1) to sf6 (1) output from the Huffman code decoder unit 15 are input to the multipliers MP11 to MP61 and the multipliers MP12 to MP62, respectively. The static gains s11 to s61 and s12 to s62 are stored in a memory in advance and fixedly given, and are input to the multipliers MP11 to MP61 and MP12 to MP62, respectively. Multipliers g'11 obtained by multiplying scale factors sf1 (1) to sf6 (1) and static gains s11 to s61 and s12 to s62 by multipliers MP11 to MP61 and MP12 to MP62, respectively. To 61 and g′12 to 62 are output to the normalizers 33 and 35, respectively.

  The normalizers 33 and 35 receive the multiplication values g′11 to 61 and g′12 to 62 output from the multipliers MP11 to MP61 and MP12 to MP62, normalize them, and perform dynamic gains g11 to 61, g12 to 62 is calculated and output to the mixdown unit 23.

  FIG. 4 is a diagram illustrating a configuration of the mixdown unit 23.

  The dynamic gains g11 to 61 output from the gain determining unit 19 are respectively input to the multipliers IMP11 to 61, and the acoustic signals i1 to i6 are input to these multipliers. The respective multiplication result values multiplied by the acoustic signals i1 to i6 are input in parallel to the adders ADD11 to ADD15, and the adders ADD11 to 15 connected in series sequentially add to calculate the output signal o1.

  Similarly, the dynamic gains g12 to 62 output from the gain determination unit 19 are respectively input to the multipliers IMP12 to IMP62, and the acoustic signals i1 to i6 are input to these multipliers. ˜62 and the acoustic signals i1 to i6 are respectively multiplied and input in parallel to the adders ADD12 to 52, and the adders ADD12 to 52 connected in series sequentially add to calculate the output signal o2. To do.

  Next, the operation of the speech decoding apparatus 11 according to the first embodiment will be described with reference to FIGS. FIG. 5 is a flowchart for explaining the operation of the normalizers 33 and 35.

  The syntax decoding unit 13 separates the high-efficiency encoded acoustic stream h, the quantization accuracy information, and the normalized information from the input encoded data string bs according to a specified grammar.

  Next, the Huffman code decoder unit 15 performs Huffman decoding on the high-efficiency encoded acoustic stream h from the syntax decoding 13 according to the encoding format obtained from the sub information, and a plurality of quantized spectra q1 to q6 and the quantized spectrum Scale factors sf1 (1) to sf6 (1) indicating the quantization width for each are output.

  Next, in inverse quantization 17, based on the plurality of quantized spectra q1 to q6 from the Huffman code decoder 15 and the plurality of scale factors sf1 (1) to sf6 (1), the original frequency signals f1 to f6 for each band. Play each one.

  Next, the gain determination unit determines a plurality of dynamic gain values g11 to 61 and g12 to 62 for each output system based on the plurality of scale factors sf1 (1) to sf6 (1) from the Huffman code decoder unit 15. .

  Here, in the gain determination unit 19, as shown in FIG. 3, a plurality of static gain values s11 to s61 preset in the plurality of scale factors sf1 (1) to sf6 (1) from the Huffman code decoder unit 15. , S12 to s62 are multiplied by multipliers MP11 to MP61 and MP12 to MP62, respectively, and a plurality of multiplication values g′11 to 61 and g′12 to 62 representing gains weighted for each scale factor are output.

  Next, the gain determination unit 19 normalizes the plurality of multiplication values g′11 to 61 and g′12 to 62 to calculate a plurality of dynamic gain values g11 to 61 and g12 to 62 for each output system, and normalizes them. To the devices 33 and 35.

  Specifically, as shown in FIG. 5, in step S10, the weighting unit 31 uses the lowest scale factors sf1 (1) to sf6 (1) to set the scale factors sf1 to the static gains s11 to s61 and s12 to s62. (1) to sf6 (1) Multiply and weight. The weighted gain g′ij can be described by multiplying the static gain sij by sfi (1) as shown in Equation 2 shown in Step S10.

  Here, an example of the set value of the static gain sij will be described. P and q are static gains of individual channels, and o1 and o2 are stereo output signals of L / R channels, respectively.

p = (8-2 * sqrt (2)) / 14 = 0.3694 (Equation 3)
q = (4 * sqrt (2)-2) / 14 = 0.2612 (Equation 4)
As
o1 = p * i1 + p * i2 + q * i4 (Equation 5)
o2 = p * i1 + p * i3 + q * i5 (Equation 6)
It becomes. The sqrt (x) function means the non-negative value of the square root of x.
When the static gain sij is described by a determinant,

It becomes.

  Next, in step S20, the normalizers 33 and 35 normalize the weighted gain g ′ to obtain dynamic gains g11 to 61 and g12 to 62. As shown in step S20, first, the denominator value is obtained from the total value of the gains g′1j to g′6j, and the dynamic gain gij is obtained by dividing the gain g′ij by this total value. The conversion is performed for each output channel i. This normalization equation can be written as shown in Equation 8. Needless to say, the total value of the dynamic gains g11 to g61 obtained in step S20 is “1”.

  Next, the time-frequency inverse transform unit 21 inversely transforms the frequency signals f1 to f6 for each band from the inverse quantization unit 17 and performs band synthesis to obtain a plurality of original time series acoustic signals i1 to i6.

  Next, based on the plurality of acoustic signals i1 to i6 from the time frequency inverse conversion unit 21 and the plurality of dynamic gains g11 to 61 and g12 to 62 for each output system from the gain determination unit 19, Output signals o1 and o2 are calculated.

  Here, in the mixdown unit 23, a plurality of dynamic gains g 11 to 61 of one output system from the normalizer 33 are respectively multiplied to the plurality of acoustic signals i 1 to i 6 from the time-frequency inverse transform unit 21 by multipliers IMP 11 to IMP 61. Is multiplied and weighted, and a plurality of multiplication values from the multipliers IMP11 to IMP61 are added by the ADD11 to ADD51 to calculate the output signal o1 of the output system.

  Similarly, a plurality of dynamic gains g12 to 62 of one output system from the normalizer 35 are respectively multiplied by the plurality of acoustic signals i1 to i6 from the time-frequency inverse transform unit 21 by the multipliers IMP12 to IMP62 and weighted. Then, a plurality of multiplied values from the multipliers IMP12 to IMP62 are added by the ADD12 to ADD52 to calculate the output signal o2 of the output system.

  As a result, the mixdown unit 23 multiplies the acoustic signals i1 to i6 by dynamic gains g11 to g62, and then adds the multiplication results in two systems to add the L / R channel output signals o1 and o2. The input acoustic signals i1 to i6 can be linearly combined.

Here, when a method of obtaining the output signals o1 and o2 from the acoustic signals i1 to i6 and the gains g11 to g62 is expressed by a determinant, Expression 8 is obtained.

  According to the present embodiment, since the gain for performing the mixdown is dynamically determined by the gain determination unit, the dynamic gain is static when the scale factors of the respective channels (1 to 6) are substantially equal. Converges to dynamic gain. On the other hand, when the scale factor of each channel is biased, a large gain is assigned to a channel with a large volume. As a result, a mixdown with a good volume balance can be performed.

  Further, in the process of determining the dynamic gain, the parameters used for decoding the digital audio signal are diverted, so that the processing load on the DSP, for example, can be extremely reduced.

(Second Embodiment)
FIG. 6 is a block diagram showing a configuration for explaining a speech decoding apparatus applicable to the mixdown apparatus according to the second embodiment of the present invention. FIG. 7 is a diagram illustrating a mutual connection relationship of the gain determination unit 19, the mixdown unit 53, and the time frequency inverse conversion unit 55.

  The speech decoding device 51 includes a syntax decoding unit 13, a Huffman code decoder unit 15, an inverse quantization unit 17, a gain determination unit 19, a mixdown unit 53, and a time-frequency inverse conversion unit 55. Note that in the speech decoding device 51 of the second embodiment, the same reference numerals are given to the same blocks as the constituent elements used in the speech decoding device 11 of the first embodiment, and the description thereof is omitted.

  The feature of the speech decoding apparatus 51 of the second embodiment is that it includes a mixdown unit 53 and a time-frequency inverse transform unit 55. This is because the time-frequency inverse transform process has linearity, and therefore the same result as in the first embodiment can be obtained even if the inverse transformation is performed after the frequency signal is mixed down.

  Based on the frequency signals f1 to f6 for each band from the inverse quantization unit 17 and the plurality of dynamic gains g11 to g62 for each output system from the gain determination 19, the mixdown unit 53 uses the frequency signal F1 for each output system. , F2 is calculated.

  The time-frequency inverse transform unit 55 inversely transforms the frequency signals F1 and F2 from the mixdown unit 53 and performs band synthesis to obtain time-series acoustic signals o1 and o2.

  FIG. 8 is a diagram illustrating a configuration of the mixdown unit 53.

  In the mixdown unit 53, the dynamic gains g11 to 61 output from the gain determination unit 19 are respectively input to the multipliers IMP11 to 61, and the frequency signals f1 to f6 are input to the respective multipliers. Multiplicative gains g11-61 and frequency signals f1-f6 are multiplied by multipliers IMP11-61 and weighted multiplication result values are respectively input in parallel to adders ADD11-15, and adders ADD11-51 connected in series. Are sequentially added to calculate the frequency signal F1.

  Similarly, the dynamic gains g12 to 62 output from the gain determining unit 19 are respectively input to the multipliers IMP12 to IMP62, and the frequency signals f1 to f6 are input to the respective multipliers, and the dynamic gain g12 is input. ~ 62, frequency signals f1 to f6 and multipliers IMP12 to 62 are multiplied and weighted multiplication result values are respectively input in parallel to adders ADD12 to 52, and serially connected adders ADD12 to 52 are sequentially added. The frequency signal F2 is calculated.

  Next, the operation of the speech decoding apparatus 51 according to the second embodiment will be described with reference to FIGS. Note that the processing contents of the syntax decoding unit 13, the Huffman code decoder unit 15, the inverse quantization unit 17, and the gain determination unit are the same as those of the speech decoding apparatus 11 of the first embodiment, and thus description thereof is omitted.

  In the mixdown unit 53, the frequency signal F1 for each output system is based on the frequency signals f1 to f6 for each band from the inverse quantization unit 17 and the plurality of dynamic gains g11 to g62 for each output system from the gain determination 19. , F2 is calculated.

  The time-frequency inverse transform unit 55 inversely transforms the frequency signals F1 and F2 from the mixdown unit 53 and performs band synthesis to obtain time-series acoustic signals o1 and o2.

  As a result, the mixdown unit 23 multiplies the frequency signals f1 to f6 by the dynamic gains g11 to g62, and then adds the multiplication results to two systems to add the L / R channel frequency signals F1 and F2. The input frequency signals f1 to f6 can be linearly combined.

Here, when a method of obtaining the frequency signals F1 and F2 from the frequency signals f1 to f6 and the gains g11 to g62 is expressed by a determinant, Equation 9 is obtained.

  According to the present embodiment, since the gain for performing the mixdown is dynamically determined by the gain determination unit, the dynamic gain is static when the scale factors of the respective channels (1 to 6) are substantially equal. Converges to dynamic gain. On the other hand, when the scale factor of each channel is biased, a large gain is assigned to a channel with a large volume. As a result, a mixdown with a good volume balance can be performed.

  Further, in the process of determining the dynamic gain, the parameters used for decoding the digital audio signal are used, so that the processing load on, for example, a DSP (Digital Signal Processor) can be extremely reduced.

  In the embodiment of the present invention, an example in which a 5.1ch signal is converted into a stereo signal is used. However, the present invention can be similarly applied even when a mixdown is performed on a plurality of channels n (n ≧ 2). become. In the embodiment of the present invention, the lowest scale factor is used to determine the dynamic gain. However, a scale factor of another band may be used, or a plurality of scale factors may be used. good.

  In the present embodiment, the mixdown apparatus has been described in conformity with the audio decoding apparatus used for MPEG-2 / 4 AAC. However, the mixdown apparatus of the present invention is limited to such MPEG-2 / 4 AAC. However, the present invention can be applied to other multi-channel playback systems.

It is a block diagram which shows the structure for demonstrating the audio | voice decoding apparatus applicable to the mixdown apparatus of 1st Embodiment which concerns on this invention. It is a figure which shows the mutual connection relation of the gain determination part 19, the time frequency reverse transformation part 21, and the mixdown part 23. FIG. 3 is a diagram illustrating a configuration of a gain determination unit 19. FIG. 3 is a diagram illustrating a configuration of a mixdown unit 23. FIG. 4 is a flowchart for explaining the operation of normalizers 33 and 35; It is a block diagram which shows the structure for demonstrating the audio | voice decoding apparatus applicable to the mixdown apparatus of 2nd Embodiment which concerns on this invention. It is a figure which shows the mutual connection relation of the gain determination part 19, the mixdown part 53, and the time frequency reverse conversion part 55. FIG. 3 is a diagram illustrating a configuration of a mixdown unit 53. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11,51 ... Speech decoding apparatus, 13 ... Syntax decoding part, 15 ... Huffman code decoder part, 17 ... Dequantization part, 19 ... Gain determination part, 21 ... Time frequency inverse transformation part, 23 ... Mixdown part 23, 31 ... Weighting unit, 33, 35 ... Normalizer, IMP11-62 ... Multiplier, ADD11-ADD52 ... Adder, 53 ... Mixdown unit, 55 ... Time-frequency inverse conversion unit

Claims (3)

In a mixdown device that mixes a plurality of input acoustic signals and converts them into a smaller number of output signals than the number of acoustic signals,
Gain determining means for determining a dynamic gain value of the acoustic signal based on the scale factor of each acoustic signal;
Mixing means for multiplying the acoustic signal by a dynamic gain value determined for the acoustic signal, and outputting a sum of multiplication result values corresponding to the plurality of acoustic signals as an output signal. Features a mixdown device. The gain determining means includes
Weighting means for multiplying a preset static gain value corresponding to each scale factor and outputting a multiplication result value for each acoustic signal;
The mixdown device according to claim 1, further comprising: a normalizing unit that normalizes a plurality of multiplication result values from the weighting unit and calculates a dynamic gain value corresponding to each acoustic signal. In a mixdown method of mixing a plurality of input acoustic signals and converting them to a number of output signals smaller than the number of acoustic signals,
A gain determining step of determining a dynamic gain value of the acoustic signal based on a scale factor of each acoustic signal;
A mixdown step of multiplying the acoustic signal by a dynamic gain value determined for the acoustic signal and outputting a sum of multiplication result values corresponding to the plurality of acoustic signals as an output signal. How to mix down.

Images