JP6508551B2 - Decryption apparatus and method, and program - Google Patents

Description

  The present invention relates to a decoding apparatus and method, and a program, and more particularly to a decoding apparatus and method, and a program that can reproduce music signals with higher sound quality by expanding a frequency band.

  BACKGROUND In recent years, music distribution services for distributing music data via the Internet etc. are becoming widespread. In this music distribution service, encoded data obtained by encoding a music signal is distributed as music data. As a coding method for music signals, a coding method for reducing the file capacity of coded data and lowering the bit rate has become mainstream so that it does not take time for downloading.

  Encoding methods for such music signals are roughly classified into encoding methods such as MP3 (MPEG (Moving Picture Experts Group) Audio Layer 3) (international standard ISO / IEC 11172-3), and HE-AAC (High Efficiency). There are coding methods such as MPEG4 AAC (international standard ISO / IEC 14496-3).

  In the encoding method represented by MP3, the signal component of the high frequency band (hereinafter referred to as high frequency band) of about 15 kHz or more which is hard to be perceived by the human ear in the music signal is deleted and the remaining low frequency band The signal component (hereinafter referred to as the low band) is encoded. Hereinafter, such a coding method is referred to as a high frequency band deletion coding method. In this high frequency band deletion encoding method, the file capacity of encoded data can be suppressed. However, since the sound in the high region is slightly perceptible to humans, when the sound is generated and output from the music signal after decoding obtained by decoding the encoded data, the presence of the original sound is lost. In some cases, the sound quality is degraded, such as the sound being muffled.

  On the other hand, in the coding method represented by HE-AAC, characteristic information is extracted from the high frequency band signal component and is encoded together with the low frequency band signal component. Hereinafter, such a coding method is referred to as a high band feature coding method. In this high-frequency feature encoding method, only characteristic information of high-frequency signal components is encoded as information on high-frequency signal components, so it is possible to improve the coding efficiency while suppressing deterioration in sound quality. .

  In the decoding of encoded data encoded by this high-frequency feature coding method, the low-frequency signal component and characteristic information are decoded, and the low-frequency signal component after decoding and the characteristic information are high. Generate the signal component of the As described above, a technique for expanding the frequency band of the low band signal component by generating the high band signal component from the low band signal component is hereinafter referred to as a band expansion technology.

  As one application example of the band expansion technique, there is post-processing after decoding of encoded data by the above-described high frequency band elimination coding technique. In this post-processing, the frequency band of the low frequency band signal component is expanded by generating the high frequency band signal component lost in coding from the low frequency band signal component after decoding (see Patent Document 1). . In addition, the method of frequency band expansion of patent document 1 is hereafter called the band expansion method of patent document 1. FIG.

  In the band expansion method of Patent Document 1, the apparatus uses the low-pass signal component after decoding as an input signal and extracts the power spectrum of the input signal from the power spectrum of the input signal (hereinafter referred to as a high-frequency envelope). Is estimated, and a high-pass signal component having its high-pass frequency envelope is generated from the low-pass signal component.

  FIG. 1 shows an example of a low frequency power spectrum after decoding as an input signal and an estimated high frequency envelope.

  In FIG. 1, the vertical axis indicates power in logarithm, and the horizontal axis indicates frequency.

  The apparatus refers to a band at the low band end of a signal component of a high band (hereinafter referred to as an expansion start band) from information such as the type of encoding method regarding the input signal, sampling rate, bit rate etc. (hereinafter referred to as side information). To determine). Next, the apparatus divides the input signal as the low-pass signal component into a plurality of sub-band signals. The apparatus is configured to group, in the time direction, respective powers of a plurality of divided subband signals, that is, a plurality of subband signals on the lower side (hereinafter simply referred to as the lower side) than the expansion start band. An average of the group powers (hereinafter referred to as group power) is calculated. As shown in FIG. 1, the apparatus starts from a point where power is an average of group powers of signals of a plurality of subbands on the lower band side and a frequency at the lower end of the expansion start band is a frequency. . The apparatus estimates a linear line of a predetermined slope passing through the origin as a frequency envelope on the higher side (hereinafter, simply referred to as the higher side) than the expansion start band. The position of the starting point in the power direction is adjustable by the user. The apparatus generates signals of the plurality of sub-bands on the high band side from the signals of the plurality of sub-bands on the low band side so as to be the estimated frequency envelope on the high band side. The apparatus adds the generated signals of the plurality of sub bands on the high band side to generate a high band signal component, and further adds and outputs a low band signal component. Thereby, the music signal after the expansion of the frequency band becomes closer to the original music signal. Therefore, it is possible to reproduce music signals with higher sound quality.

  The band expansion method of Patent Document 1 described above can expand the frequency band of the music signal after decoding of the encoded data of various high-pass deletion encoding methods and encoded data of various bit rates. It has the feature of

JP, 2008-139844, A

  However, there is room for improvement in the band expansion method of Patent Document 1 in that the estimated high-frequency side frequency envelope is a linear line with a predetermined slope, that is, the shape of the frequency envelope is fixed. There is.

  That is, the power spectrum of the music signal has various shapes, and depending on the type of the music signal, the frequency envelope of the high frequency side estimated by the band expansion method of Patent Document 1 often deviates greatly.

  FIG. 2 shows an example of the original power spectrum of an attacking music signal (attacking music signal) accompanied by a sudden change in time, such as when the drum is hit hard once.

  In FIG. 2, with the signal component on the low band side of the attacking music signal as the input signal by the band expansion method of Patent Document 1, the frequency envelope on the high band side estimated from the input signal is also added. It is shown.

  As shown in FIG. 2, the power spectrum on the high frequency side of the attacking music signal is substantially flat.

  On the other hand, the estimated high-frequency side frequency envelope has a predetermined negative slope, and even if the power source is adjusted to a power close to the original power spectrum, the original power is increased as the frequency increases. The difference with the spectrum is large.

  As described above, in the band expansion method of Patent Document 1, the estimated high-frequency side frequency envelope can not reproduce the original high-frequency side frequency envelope with high accuracy. As a result, when a sound is generated and output from the music signal after the expansion of the frequency band, the audibility may be lost in the clarity of the sound rather than the original sound.

  Further, in the above-described high-frequency feature coding method such as HE-AAC, although the high-frequency side frequency envelope is used as characteristic information of the high-frequency signal component to be coded, the original high frequency is used on the decoding side It is required to reproduce the frequency envelope on the band side with high accuracy.

  The present invention has been made in view of such a situation, and is intended to reproduce music signals with higher sound quality by expanding a frequency band.

A decoding device according to an aspect of the present technology demultiplexes input coded data into low-pass coded data and an index indicating an estimated coefficient used to generate a high-pass signal, and A low band signal obtained by decoding low band encoded data is divided into a plurality of low band sub-bands, and sub-band dividing means for generating a low band sub-band signal for each low band sub-band; and for each low band sub-band , using said low-frequency subband signal, a feature amount calculating means for calculating a low-frequency sub-band power, each of the estimated coefficients indicated by the index, is multiplied by each of the low frequency sub-band power, the High-frequency sub-band power for calculating high-frequency sub-band power of the high-frequency sub-band signal of the high-frequency signal by calculating the sum of the respective low-frequency sub-band powers multiplied by an estimation coefficient High band signal generation means for generating the high band sub-band signal using the output means, the high band sub-band power, and the low band sub-band signal of the mapping source, and the respective low band sub-band signals And combining means for combining and outputting the high band sub-band signals generated by the high band signal generating means.

A decoding method or program according to one aspect of the present technology demultiplexes input coded data into low-pass coded data and an index indicating an estimated coefficient used to generate a high-pass signal. A subband dividing step of dividing the low band signal obtained by decoding the low band encoded data into a plurality of low band sub-bands and generating a low band sub-band signal for each low band sub-band; and the low band sub-band each, using said low-frequency subband signal, a feature amount calculating step of calculating a low frequency sub-band power, each of the estimated coefficients indicated by the index, is multiplied by each of the low frequency sub-band power A high band sub-band power of the high band sub-band signal of the high band signal is obtained by obtaining a sum of the respective low band sub-band powers multiplied by the estimation coefficient. A high band signal generation step of generating the high band sub-band signal using the high band sub-band power calculation step, the high band sub-band power, and the low band sub-band signal of the mapping source; And combining the low-pass sub-band signals and the high-pass sub-band signals generated by the process of the high-pass signal generating step.

In one aspect of the present technology, input coded data is demultiplexed into low band coded data and an index indicating an estimated coefficient used to generate a high band signal, and the low band coded data is decoded. The low band signal is divided into a plurality of low band sub-bands to generate a low band sub-band signal for each low band sub-band, and the low band sub-band signal is used for each low band sub-band, low frequency sub-band power is calculated, the sum of the each of the estimated coefficients indicated by the index is multiplied to each of the low frequency sub-band power, each of the low frequency sub-band power multiplication of the estimated coefficients By calculating the high band sub-band power of the high band sub-band signal of the high band signal, the high band sub-band power and the low band sub-band signal of the mapping source And it is used, the high frequency sub-band signals are generated, and each of the low frequency sub-band signals, each of said high frequency sub-band signals generated are output after being synthesized.

  According to one aspect of the present invention, the music signal can be reproduced with higher sound quality by the expansion of the frequency band.

It is a figure which shows an example of the power spectrum of the low region after decoding as an input signal, and the frequency envelope of the estimated high region. It is a figure which shows an example of the original power spectrum of an attack-type music signal with a rapid change in time. It is a block diagram showing an example of functional composition of a frequency band expansion device in a 1st embodiment of the present invention. It is a flowchart explaining the example of the frequency band expansion process by the frequency band expansion apparatus of FIG. It is a figure which shows the arrangement | positioning on the frequency axis of the power spectrum of the signal input into the frequency band expansion apparatus of FIG. 3, and a band pass filter. It is a figure which shows the example of the frequency characteristic of a vocal section, and the power spectrum of the presumed high region. It is a figure which shows the example of the power spectrum of the signal input into the frequency band expansion apparatus of FIG. It is a figure which shows the example of the power spectrum after liftering of the input signal of FIG. It is a block diagram which shows the functional structural example of the coefficient learning apparatus for learning the coefficient used by the high region signal generation circuit of the frequency band expansion apparatus of FIG. It is a flowchart explaining the example of the coefficient learning process by the coefficient learning apparatus of FIG. It is a block diagram showing an example of functional composition of an encoding device in a 2nd embodiment of the present invention. It is a flowchart explaining the example of the encoding process by the encoding apparatus of FIG. It is a block diagram showing an example of functional composition of a decoding device in a 2nd embodiment of the present invention. It is a flowchart explaining the example of the decoding process by the decoding apparatus of FIG. Function of coefficient learning apparatus for learning representative vector used in high band coding circuit of coding apparatus in FIG. 11 and decoded high band sub-band power estimation coefficient used in high band decoding circuit of decoding apparatus in FIG. 13 Block diagram showing an exemplary configuration. It is a flowchart explaining the example of the coefficient learning process by the coefficient learning apparatus of FIG. It is a figure which shows the example of the code string which the encoding apparatus of FIG. 11 outputs. It is a block diagram showing an example of functional composition of an encoding device. It is a flowchart explaining an encoding process. It is a block diagram showing an example of functional composition of a decoding device. It is a flowchart explaining a decoding process. It is a flowchart explaining an encoding process. It is a flowchart explaining a decoding process. It is a flowchart explaining an encoding process. It is a flowchart explaining an encoding process. It is a flowchart explaining an encoding process. It is a flowchart explaining an encoding process. It is a figure showing an example of composition of a coefficient learning device. It is a flowchart explaining a coefficient learning process. It is a block diagram showing an example of composition of hardware of a computer which performs processing to which the present invention is applied by a program.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. The description will be made in the following order.
1. First embodiment (when the present invention is applied to a frequency band expanding device)
2. Second embodiment (when the present invention is applied to a coding apparatus and a decoding apparatus)
3. Third embodiment (when coefficient index is included in high band encoded data)
4. Fourth embodiment (when coefficient index and pseudo high band sub-band power difference are included in high band encoded data)
5. Fifth embodiment (when selecting a coefficient index using an evaluation value)
6. Sixth embodiment (when a part of coefficients is shared)

<1. First embodiment>
In the first embodiment, a process of expanding the frequency band to the low-pass signal component after decoding obtained by decoding the encoded data by the high-pass deletion coding method (hereinafter referred to as frequency band expansion processing Is called).

[Functional Configuration Example of Frequency Band Expansion Device]
FIG. 3 shows a functional configuration example of a frequency band expanding device to which the present invention is applied.

  Frequency band expansion device 10 performs frequency band expansion processing on the input signal using the low-pass signal component after decoding as an input signal, and outputs the signal after frequency band expansion processing obtained as a result as an output signal Do.

  The frequency band expansion device 10 includes a low pass filter 11, a delay circuit 12, a band pass filter 13, a feature amount calculation circuit 14, a high band sub-band power estimation circuit 15, a high band signal generation circuit 16, a high pass filter 17, And a signal adder 18.

  The low pass filter 11 filters the input signal at a predetermined cutoff frequency, and supplies a low pass signal component which is a low pass signal component to the delay circuit 12 as a filtered signal.

  The delay circuit 12 delays the low band signal component by a fixed delay time to synchronize in adding the low band signal component from the low pass filter 11 and the high band signal component described later. The data is supplied to the adder 18.

  The band pass filter 13 includes band pass filters 13-1 to 13-N each having a different pass band. The band pass filter 13-i (1 ≦ i ≦ N) passes a signal in a predetermined pass band of the input signal and, as one of the plurality of sub band signals, the feature amount calculation circuit 14 and the high band The signal generation circuit 16 is supplied.

  The feature quantity calculation circuit 14 calculates one or more feature quantities using at least one of the plurality of subband signals from the band pass filter 13 and the input signal, and a high band subband power estimation circuit Supply to 15. Here, the feature amount is information representing a feature of the input signal as a signal.

  The high frequency sub-band power estimation circuit 15 performs high frequency sub-band estimation of the high frequency sub-band power, which is the power of the high frequency sub-band signal, based on the one or more feature quantities from the feature quantity calculation circuit 14 It is calculated for each band, and these are supplied to the high band signal generation circuit 16.

  The high band signal generation circuit 16 is a high band signal based on the plurality of sub-band signals from the band pass filter 13 and the plurality of high band sub-band power estimation values from the high band sub-band power estimation circuit 15. A high-pass signal component which is a component is generated and supplied to the high pass filter 17.

  The high pass filter 17 filters the high band signal component from the high band signal generation circuit 16 with the cut-off frequency corresponding to the cut-off frequency in the low-pass filter 11 and supplies the high pass filter to the signal adder 18.

  The signal adder 18 adds the low pass signal component from the delay circuit 12 and the high pass signal component from the high pass filter 17 and outputs the result as an output signal.

  In the configuration of FIG. 3, the band pass filter 13 is applied to obtain the sub-band signal. However, the present invention is not limited to this. For example, a band division filter as described in Patent Document 1 may be used. It may be applied.

  Similarly, although the signal adder 18 is applied to combine sub-band signals in the configuration of FIG. 3, the present invention is not limited to this. For example, band combining as described in Patent Document 1 is performed. A filter may be applied.

[Frequency band expansion processing of frequency band expansion device]
Next, with reference to the flowchart of FIG. 4, the frequency band expansion processing by the frequency band expansion device of FIG. 3 will be described.

  In step S1, the low pass filter 11 filters the input signal at a predetermined cutoff frequency, and supplies the low pass signal component as a filtered signal to the delay circuit 12.

  The low pass filter 11 can set an arbitrary frequency as the cutoff frequency, but in the present embodiment, it corresponds to the frequency at the lower end of the enlargement start band as the enlargement start band described later. And the cutoff frequency is set. Therefore, the low pass filter 11 supplies the low pass signal component which is a signal component of the low pass from the expansion start band to the delay circuit 12 as a signal after filtering.

  The low pass filter 11 can also set an optimal frequency as the cutoff frequency according to the high band removal coding method of the input signal and the coding parameters such as the bit rate. As this encoding parameter, for example, side information adopted by the band expansion method of Patent Document 1 can be used.

  In step S 2, the delay circuit 12 delays the low-pass signal component from the low-pass filter 11 by a fixed delay time and supplies the delayed signal component to the signal adder 18.

  In step S3, the band pass filter 13 (the band pass filters 13-1 to 13-N) divides the input signal into a plurality of sub-band signals, and each of the plurality of divided sub-band signals is a feature amount calculation circuit 14 and the high band signal generation circuit 16. The process of dividing the input signal by the band pass filter 13 will be described in detail later.

  In step S4, the feature amount calculation circuit 14 calculates one or more feature amounts using at least one of the plurality of sub-band signals from the band pass filter 13 and the input signal, and the high range sub A band power estimation circuit 15 is supplied. The details of the process of calculating the feature amount by the feature amount calculation circuit 14 will be described later.

  In step S5, the high band sub-band power estimating circuit 15 calculates estimated values of a plurality of high band sub-band powers based on one or more feature quantities from the feature quantity calculating circuit 14, and generates high band signals. The circuit 16 is supplied. The details of the process of calculating the estimated value of the high band sub-band power by the high band sub-band power estimation circuit 15 will be described later.

  In step S 6, high band signal generation circuit 16 is based on the plurality of sub band signals from band pass filter 13 and the plurality of high band sub band power estimation values from high band sub band power estimation circuit 15, A high-pass signal component is generated and supplied to the high pass filter 17. The high frequency band signal component referred to here is a signal component in the high frequency band above the enlargement start band. The details of the process of generating the high frequency band signal component by the high frequency band signal generation circuit 16 will be described later.

  In step S7, the high pass filter 17 filters the high band signal component from the high band signal generation circuit 16 to remove noise such as aliasing to the low band included in the high band signal component, and The high band signal component is supplied to the signal adder 18.

  In step S8, the signal adder 18 adds the low band signal component from the delay circuit 12 and the high band signal component from the high pass filter 17 and outputs the result as an output signal.

  According to the above process, the frequency band can be expanded with respect to the low frequency band signal component after decoding.

  Next, details of each processing of steps S3 to S6 in the flowchart of FIG. 4 will be described.

Details of processing by band pass filter
First, details of processing by the band pass filter 13 in step S3 of the flowchart of FIG. 4 will be described.

  In the following, the number N of band pass filters 13 is set to N = 4 for the convenience of description.

  For example, one of 16 sub-bands obtained by dividing the Nyquist frequency of the input signal into 16 equal parts is set as the expansion start band, and the lower band than the expansion start band of the 16 sub-bands. Let each of four sub-bands be each of the pass band of the band pass filters 13-1 to 13-4.

  FIG. 5 shows the arrangement on the frequency axis of each of the pass bands of the band pass filters 13-1 to 13-4.

  As shown in FIG. 5, the index of the first sub-band from the high frequency side of the frequency band (sub-band) lower than the expansion start band is sb, the index of the second sub-band is sb-1, I Assuming that the index of the second subband is sb- (I-1), each of the band pass filters 13-1 to 13-4 has an index of sb to sb-3 among subbands lower than the expansion start band. Each of the subbands is assigned as a passband.

  In the present embodiment, each of the pass bands of band pass filters 13-1 to 13-4 is a predetermined four of the sixteen subbands obtained by dividing the Nyquist frequency of the input signal into sixteen equal parts. However, the present invention is not limited to this. Each of predetermined four of the 256 sub-bands obtained by equally dividing the Nyquist frequency of the input signal into 256 may be used. . Also, the bandwidths of the band pass filters 13-1 to 13-4 may be different from one another.

[Details of processing by feature amount calculation circuit]
Next, details of the process by the feature value calculation circuit 14 in step S4 of the flowchart of FIG. 4 will be described.

  The feature quantity calculation circuit 14 estimates the high band sub-band power of the high band sub-band power estimation circuit 15 using at least one of the plurality of sub-band signals from the band pass filter 13 and the input signal. To calculate one or more feature quantities used to calculate

  More specifically, the feature quantity calculation circuit 14 calculates the power of subband signals (hereinafter referred to as low band subband power) for each subband from the four subband signals from the band pass filter 13. ) Is calculated as the feature amount and supplied to the high band sub-band power estimation circuit 15.

  That is, the feature quantity calculation circuit 14 calculates the low band sub-band power power (ib, J) in a predetermined time frame J from the four sub-band signals x (ib, n) supplied from the band pass filter 13 Is obtained by the following equation (1). Here, ib represents the index of the sub-band and n represents the index of the discrete time. Here, it is assumed that the number of samples of one frame is FSIZE, and the power is expressed in decibels.

・・・（１）

... (1)

  Thus, the low band sub-band power power (ib, J) obtained by the feature quantity calculation circuit 14 is supplied to the high band sub-band power estimation circuit 15 as a feature quantity.

[Details of processing by high band sub-band power estimation circuit]
Next, details of the processing by the high-frequency sub-band power estimation circuit 15 in step S5 of the flowchart of FIG. 4 will be described.

  The high band sub-band power estimation circuit 15 tries to expand the sub-band (the enlargement start band) whose index is sb + 1 based on the four sub-band powers supplied from the feature quantity calculation circuit 14 The estimated value of the sub-band power (high frequency sub-band power) of the band (frequency expansion band) is calculated.

  That is, assuming that the high-order sub-band power estimation circuit 15 sets the index of the high-order sub-band of the frequency expansion band to eb, (eb−sb) sub-band powers for the sub-bands whose index is sb + 1 to eb presume.

The estimated value power _est (ib, J) of the subband power whose index is ib in the frequency expansion band is calculated using the four subband powers power (ib, j) supplied from the feature quantity calculation circuit 14 For example, it is expressed by the following equation (2).

Here, in the equation (2), the coefficients A _ib (kb) and B _ib are coefficients having different values for each subband ib. The coefficients A _ib (kb) and B _ib are coefficients appropriately set so as to obtain suitable values for various input signals. In addition, the coefficients A _ib (kb) and B _ib are also changed to optimal values by changing the sub-band sb. The derivation of the coefficients A _ib (kb) and B _ib will be described later.

  In the equation (2), the estimated value of the high band sub-band power is calculated by first-order linear combination using the power of each of the plurality of sub-band signals from the band pass filter 13, but not limited to this It may be calculated using a linear combination of a plurality of low band sub-band powers of several frames before and after the time frame J, or may be calculated using a non-linear function.

  Thus, the estimated value of the high band sub-band power calculated by the high band sub-band power estimating circuit 15 is supplied to the high band signal generating circuit 16.

[Details of processing by high band signal generation circuit]
Next, details of the processing by the high frequency band signal generation circuit 16 in step S6 of the flowchart of FIG. 4 will be described.

The high band signal generation circuit 16 generates low band sub-band powers power (ib, J) of the respective sub-bands from the plurality of sub-band signals supplied from the band pass filter 13 based on the above equation (1). calculate. The high frequency band signal generation circuit 16 calculates a plurality of low frequency band sub-band powers (ib, J) and the high frequency sub-band calculated by the high frequency sub-band power estimation circuit 15 based on the above equation (2). The amount of gain G (ib, J) is determined by the following equation (3) using the power estimate value power _est (ib, J).

Here, in equation (3), sb _map (ib) indicates the index of the subband of the mapping source when subband ib is the subband of the mapping destination, and is represented by the following equation (4) .

・・・（４）

... (4)

  In Equation (4), INT (a) is a function for rounding off the decimal part of the value a.

  Next, the high band signal generation circuit 16 performs gain adjustment by multiplying the output of the band pass filter 13 by the gain amount G (ib, J) obtained by the formula (3) using the following formula (5). The subsequent subband signal x2 (ib, n) is calculated.

・・・（５）

... (5)

  Further, the high frequency band signal generation circuit 16 corresponds to the frequency at the upper end of the subband whose index is sb from the frequency corresponding to the frequency at the lower end of the subband whose index is sb-3 according to the following equation (6) By performing cosine modulation to the frequency to be calculated, a cosine-transformed gain-adjusted subband signal x3 (ib, n) is calculated from the gain-adjusted subband signal x2 (ib, n).

・・・（６）

... (6)

  In equation (6), Π represents the circle ratio. This equation (6) means that the sub-band signal x2 (ib, n) after gain adjustment is shifted to a frequency higher by four bands, respectively.

Then, the high band signal generation circuit 16 calculates the high band signal component x _high (n) from the sub-band signal x 3 (ib, n) after gain adjustment shifted to the high band side by the following equation (7) Do.

・・・（７）

... (7)

  In this manner, the four low band sub-band powers calculated by the high band signal generation circuit 16 based on the four sub-band signals from the band pass filter 13 and the high band sub-band power estimation circuit 15 The high band signal component is generated based on the estimated value of the high band sub-band power and is supplied to the high pass filter 17.

  According to the above processing, for the input signal obtained after decoding of the encoded data by the high-pass deletion coding method, the low-pass sub-band power calculated from the plurality of sub-band signals is used as a feature amount. An estimated value of high band sub-band power is calculated based on appropriately set coefficients, and a high band signal component is adaptively generated from the low band sub-band power and the estimated value of high band sub-band power Therefore, the sub-band power of the frequency expansion band can be estimated with high accuracy, and the music signal can be reproduced with higher sound quality.

  In the above, an example has been described in which the feature amount calculation circuit 14 calculates only low band sub-band power calculated from a plurality of sub-band signals as a feature amount. In this case, frequency expansion may be performed depending on the type of input signal. It may not be possible to estimate the subband power of a band with high accuracy.

  Therefore, the high-frequency sub-band power estimation circuit is configured such that the feature-value calculation circuit 14 calculates a feature quantity that strongly correlates with the output of the sub-band power of the frequency expansion band (the shape of the power spectrum of the high band) The estimation of the sub-band power of the frequency extension band at 15 can also be performed with higher accuracy.

[Another Example of Feature Value Calculated by Feature Value Calculation Circuit]
FIG. 6 estimates high band sub-band power by calculating only low band sub-band power as an example of the frequency characteristic of a vocal section in which a vocal occupies most of an input signal and calculating low band sub-band power only And the power spectrum of the high region obtained by

  As shown in FIG. 6, in the frequency characteristic of the vocal zone, the estimated high frequency power spectrum is often positioned above the high frequency power spectrum of the original signal. It is necessary to estimate the high-frequency sub-band power with high accuracy in the vocal section since the sense of incongruity of the human singing voice is easily perceived by the human ear.

  Also, as shown in FIG. 6, in the frequency characteristic of the vocal section, there is often one large recess between 4.9 kHz and 11.025 kHz.

  So, below, the example which applies the degree of the dent in 4.9 kHz to 11.025 kHz in a frequency domain as a feature-value used for presumption of the high region sub-band power of a vocal section is explained. In addition, the feature-value which shows the degree of this dent is hereafter called dip.

  Hereinafter, an example of calculation of the dip dip (J) in the time frame J will be described.

  First, among the input signals, 2048-point FFT (Fast Fourier Transform) is applied to a signal of a 2048 sample section included in the range of several frames before and after the time frame J, and coefficients on the frequency axis are calculated. A power spectrum is obtained by applying db conversion to the calculated absolute value of each coefficient.

  FIG. 7 shows an example of the power spectrum obtained as described above. Here, in order to remove a fine component of the power spectrum, for example, a lifting process is performed to remove a component of 1.3 kHz or less. According to the refilling processing, it is possible to smooth fine components of the spectrum peak by performing filtering processing by applying low-pass filtering to each dimension of the power spectrum as a time series.

  FIG. 8 shows an example of the power spectrum of the input signal after liftering. In the power spectrum after lifter shown in FIG. 8, the difference between the minimum value and the maximum value of the power spectrum included in the range corresponding to 4.9 kHz to 11.025 kHz is taken as dip dip (J).

  In this way, feature quantities that are strongly correlated with the sub-band power of the frequency expansion band are calculated. The calculation example of the dip dip (J) is not limited to the method described above, and may be another method.

  Next, another example of calculation of a feature quantity strongly correlated with the sub-band power of the frequency expansion band will be described.

[Another Example of Feature Amount Calculated by Feature Amount Calculation Circuit]
In the frequency characteristics of the attack section, which is a section including an attacking music signal in a certain input signal, the power spectrum on the high frequency side is often substantially flat as described with reference to FIG. In the method of calculating only the low band sub-band power as the feature amount, since the sub-band power of the frequency expansion band is estimated without using the feature amount representing the time variation peculiar to the input signal including the attack period, it can be seen in the attack period It is difficult to accurately estimate the sub-band power of the substantially flat frequency expansion band.

  So, below, the example which applies the time change of low-pass sub-band power as a feature-value used for presumption of the high-pass sub-band power of an attack section is explained.

The time variation power _d (J) of the low band sub-band power in a certain time frame J is obtained, for example, by the following equation (8).

・・・（８）

... (8)

According to the equation (8), the time variation power _d (J) of the low band sub-band power is the sum of the four low band sub-band powers in time frame J and the time frame one frame before time frame J It represents the ratio to the sum of the four low-frequency sub-band powers in J-1), and the larger this value, the greater the time variation of power between frames, ie, the signal contained in time frame J is an attack It is thought that sex is strong.

  In addition, comparing the statistically average power spectrum shown in FIG. 1 with the power spectrum of the attack section (attack music signal) shown in FIG. It is rising. In the attack section, such frequency characteristics are often shown.

  So, below, the example which applies the inclination in the middle region as a feature-value used for estimation of the high region sub-band power of an attack section is explained.

  The slope slope (J) of the middle region in a certain time frame J is determined, for example, by the following equation (9).

・・・（９）

... (9)

  In equation (9), the coefficient w (ib) is a weighting coefficient adjusted to weight the high band sub-band power. According to equation (9), slope (J) represents the ratio of the sum of the four low band sub-band powers weighted in the high band to the sum of the four low band sub-band powers. For example, if four low band sub-band powers are for the middle band sub-bands, slope (J) is large when the middle band power spectrum is rising to the right, and it is falling when falling to the right. Take a small value.

In addition, since the slope of the midrange often fluctuates largely before and after the attack interval, the time variation of slope _d (J) expressed by the following equation (10) can be used to estimate the high frequency sub-band power of the attack interval. It may be used as a feature amount used for

・・・（１０）

... (10)

Similarly, the time variation dip _d (J) of the above-mentioned dip dip (J) expressed by the following equation (11) is made to be the feature amount used for estimation of the high band sub-band power of the attack section. May be

・・・（１１）

... (11)

  According to the above-described method, since the feature amount having a strong correlation with the sub-band power of the frequency expansion band is calculated, estimation of the sub-band power of the frequency expansion band in the high band sub-band power estimation circuit 15 is performed using these Can be done with higher accuracy.

  In the above, an example of calculating a feature quantity that is strongly correlated with the sub-band power of the frequency expansion band has been described, but in the following, the high band sub-band power is estimated using the feature quantity calculated in this way An example will be described.

[Details of processing by high band sub-band power estimation circuit]
Here, an example in which the high-frequency sub-band power is estimated using the dip described with reference to FIG. 8 and the low-frequency sub-band power as feature amounts will be described.

  That is, in step S4 of the flowchart of FIG. 4, the feature quantity calculation circuit 14 uses the low band sub-band power and the dip as feature quantities from the four sub-band signals from the band pass filter 13 for each sub-band. It is calculated and supplied to the high band sub-band power estimation circuit 15.

  Then, in step S5, the high band sub-band power estimating circuit 15 calculates an estimated value of the high band sub-band power based on the four low band sub-band powers and dips from the feature quantity calculating circuit 14.

  Here, since the range (scale) of possible values is different between the sub-band power and the dip, the high-frequency sub-band power estimation circuit 15 performs, for example, the following conversion on the value of the dip.

The high band sub-band power estimation circuit 15 previously calculates the sub-band power of the highest band among the four low band sub-band powers and the dip value for a large number of input signals, and the average value is calculated for each of them. And standard deviation. Here, the average value of subband power is power _ave , the standard deviation of subband power is power _std , the average value of dip is dip _ave , and the standard deviation of dip is dip _std .

The high band sub-band power estimation circuit 15 uses these values to convert the dip value dip (J) into the following equation (12) to obtain the converted dip dip _s (J).

・・・（１２）

... (12)

By performing the conversion represented by the equation (12), the high band sub-band power estimating circuit 15 statistically changes the dip value dip (J) into a variable (dip) equal to the average and the variance of the low band sub-band powers. It is possible to convert to dip _s (J), and it is possible to make the range of possible values of dip approximately the same as the range of possible values of subband power.

The estimated value power _est (ib, J) of the subband power whose index is ib in the frequency expansion band is the four lower subband powers power (ib, J) from the feature value calculation circuit 14 and the formula ( Using the linear combination with the dip dip _s (J) shown in 12), for example, it is represented by the following formula (13).

Here, in the equation (13), the coefficients C _ib (kb), D _ib and E _ib are coefficients having different values for each subband ib. The coefficients C _ib (kb), D _ib and E _ib are appropriately set so as to obtain suitable values for various input signals. Also, the coefficients C _ib (kb), D _ib and E _ib are also changed to optimal values by changing the sub-band sb. The derivation of the coefficients C _ib (kb), D _ib and E _ib will be described later.

  In the equation (13), the estimated value of the high band sub-band power is calculated by first-order linear combination, but not limited to this, for example, linear combination of a plurality of feature quantities of several frames before and after the time frame J It may be calculated using or may be calculated using a non-linear function.

  According to the above process, by using the value of dip specific to the vocal section as the feature quantity to estimate the high-frequency sub-band power, compared to the case where only the low-frequency sub-band power is used as the feature quantity, The estimation accuracy of the high band sub-band power is improved, and a method that uses only the low band sub-band power as a feature amount, and is caused by the high band power spectrum being estimated larger than the high band power spectrum of the original signal Since the sense of incongruity which is easily perceived by the user's ear is reduced, it is possible to reproduce the music signal with higher sound quality.

  By the way, with regard to the dip (degree of depression in the frequency characteristic of the vocal zone) calculated as the feature amount in the method described above, when the number of subbands is 16 divisions, the frequency resolution is low, so only the low band subband power So, we can not express the degree of this dent.

  Therefore, the number of divisions of subbands is increased (for example, 16 times by 256 divisions), the number of band divisions by the band pass filter 13 is increased (for example, 64 times of 16 times), and the low band subbands calculated by the feature amount calculation circuit 14 By increasing the number of powers (e.g. 64 of 16 times), it is possible to increase the frequency resolution and express the degree of denting only with the low band sub-band power.

  Thus, it is considered possible to estimate the high-frequency sub-band power with almost the same accuracy as the estimation of the high-frequency sub-band power using the above-described dip as the feature amount with only the low-frequency sub-band power. .

  However, the amount of calculation increases by increasing the number of subband divisions, the number of band divisions, and the number of lower band subband powers. Considering that the high band sub-band power can be estimated with the same accuracy by any of the methods, the number of sub-band divisions is not increased, and the method of estimating the high band sub-band power using dip as a feature amount is more It is considered to be efficient in terms of quantity.

  In the above, the method of estimating the high band sub-band power using the dip and the low band sub-band power has been described, but the feature quantity used for the high band sub-band power estimation is not limited to this combination. Alternatively, one or more of the feature quantities described above (low-frequency subband power, dip, low-frequency subband power temporal variation, slope, slope temporal variation, and dip temporal variation) may be used. Good. Thus, the accuracy can be further improved in the estimation of the high band sub-band power.

  In addition, as described above, in the input signal, by using a parameter specific to a section where high band sub-band power estimation is difficult as a feature quantity used for high band sub-band power estimation, the section estimation Accuracy can be improved. For example, time variation of low frequency sub-band power, time variation of slope, time variation of slope, and time variation of dip are parameters specific to the attack section, and using these parameters as feature quantities, the high value in the attack section The estimation accuracy of the region subband power can be improved.

  Note that high-frequency subband power is estimated using feature quantities other than low-frequency subband power and dip, that is, temporal fluctuation of low-frequency subband power, inclination, temporal fluctuation of inclination, and temporal fluctuation of dip. The high band sub-band power can be estimated by the same method as the method described above.

  In addition, each calculation method of the feature-value shown here may be made to use not only the method demonstrated above but another method.

[How to find coefficients C _ib (kb), D _ib , E _ib ]
Next, how to obtain the coefficients C _ib (kb), D _ib and E _{ib in} the above-described equation (13) will be described.

Coefficient _{C ib (kb), D ib} , as Determination of E _ib, the coefficient _{C ib (kb), D ib} , E ib is for various input signals in estimating the sub-band power of the frequency extension band In order to obtain suitable values, learning is performed in advance using a wide-band teacher signal (hereinafter, referred to as a wide-band teacher signal), and a method of determining based on the learning result is applied.

When learning the coefficients C _ib (kb), D _ib and E _ib , the same pass as the band pass filters 13-1 to 13-4 described with reference to FIG. A coefficient learning apparatus in which a band pass filter having a bandwidth is arranged is applied. The coefficient learning device performs learning when the broadband teacher signal is input.

[Functional Configuration Example of Coefficient Learning Device]
FIG. 9 shows an example of a functional configuration of a coefficient learning device that performs learning of the coefficients C _ib (kb), D _ib and E _ib .

  The signal component lower than the expansion start band of the wide band teacher signal input to the coefficient learning device 20 of FIG. 9 is encoded with the band-limited input signal input to the frequency band expansion device 10 of FIG. It is preferable that the signal is encoded in the same manner as the encoding performed in the above.

  The coefficient learning apparatus 20 includes a band pass filter 21, a high band sub-band power calculation circuit 22, a feature quantity calculation circuit 23, and a coefficient estimation circuit 24.

  The band pass filter 21 includes band pass filters 21-1 to 21- (K + N) each having a different pass band. The band pass filter 21-i (1 ≦ i ≦ K + N) passes a signal in a predetermined pass band of the input signal and outputs the high band sub-band power calculation circuit 22 as one of a plurality of sub-band signals. Alternatively, it is supplied to the feature value calculation circuit 23. The band-pass filters 21-1 to 21-K among the band-pass filters 21-1 to 21- (K + N) pass signals in the higher band than the enlargement start band.

  The high band sub-band power calculating circuit 22 calculates high band sub-band power for each sub-band for each predetermined time frame with respect to a plurality of high-band sub-band signals from the band-pass filter 21. The signal is supplied to the estimation circuit 24.

  The feature quantity calculation circuit 23 calculates the feature quantity calculation circuit 14 of the frequency band expansion device 10 of FIG. 3 every time frame that is the same as a predetermined time frame for which the high band sub-band power is calculated by the high band sub-band power calculation circuit 22. The same feature amount as the feature amount calculated by is calculated. That is, the feature amount calculation circuit 23 calculates one or more feature amounts using at least one of the plurality of subband signals from the band pass filter 21 and the wide band teacher signal, and the coefficient estimation circuit 24. Supply to

  The coefficient estimation circuit 24 expands the frequency band of FIG. 3 based on the high band sub-band power from the high band sub-band power calculation circuit 22 and the feature quantity from the feature quantity calculation circuit 23 for each fixed time frame. The coefficient (coefficient data) used in the high band sub-band power estimation circuit 15 of the apparatus 10 is estimated.

[Coefficient learning process of coefficient learning device]
Next, coefficient learning processing by the coefficient learning device in FIG. 9 will be described with reference to the flowchart in FIG.

  In step S11, the band pass filter 21 divides the input signal (wideband teacher signal) into (K + N) sub-band signals. The band pass filters 21-1 to 21-K supply the high band sub-band power calculation circuit 22 with a plurality of sub-band signals higher than the enlargement start band. In addition, the band pass filters 21-(K + 1) to 21-(K + N) supply a plurality of sub-band signals lower than the expansion start band to the feature amount calculation circuit 23.

  In step S12, the high-frequency sub-band power calculation circuit 22 sets the high-frequency plurality of sub-band signals from the band-pass filter 21 (the band-pass filters 21-1 to 21-K) to a predetermined time frame. Then, the high band sub-band power power (ib, J) for each sub-band is calculated. The high band sub-band power power (ib, J) is obtained by the above-mentioned equation (1). The high frequency sub-band power calculation circuit 22 supplies the calculated high frequency sub-band power to the coefficient estimation circuit 24.

  In step S13, the feature quantity calculation circuit 23 calculates a feature quantity every time frame that is the same as a predetermined time frame for which the high band sub-band power is calculated by the high band sub-band power calculation circuit 22.

  In the following, assuming that the four sub-band powers and dips in the low band are calculated as feature amounts in the feature amount calculation circuit 14 of the frequency band expansion device 10 of FIG. Also in the feature amount calculation circuit 23, it is assumed that four low band sub-band powers and dips are similarly calculated.

That is, the feature quantity calculation circuit 23 receives four of the feature quantity calculation circuits 14 of the frequency band expansion device 10 from the band pass filters 21 (band pass filters 21-(K + 1) to 21-(K + 4)). Four low band sub-band powers are calculated using four sub-band signals in the same band as the sub-band signal. Also, the feature value calculation circuit 23 calculates a dip from the wide band teacher signal, and calculates a dip dip _s (J) based on the above-mentioned equation (12). The feature quantity calculation circuit 23 supplies the calculated four low band sub-band powers and the dip dip _s (J) to the coefficient estimation circuit 24 as a feature quantity.

In step S14, the coefficient estimation circuit 24 supplies (eb-sb) high-frequency sub-band powers and feature amounts (4) supplied from the high-frequency sub-band power calculation circuit 22 and the feature amount calculation circuit 23 in the same time frame. number of based on a number of combinations of the low frequency sub-band power and the dip dip _s (J)), the coefficient C _ib (kb), performs D _ib, estimation of E _ib. For example, the coefficient estimation circuit 24, for one of the sub-bands of a high band, five feature amounts (four low frequency sub-band power and the dip dip _s (J)) as explanatory variables, the high frequency sub-band power The coefficients C _ib (kb), D _ib , and E _ib in the equation (13) are determined by performing regression analysis using the least squares method with the power (ib, J) of L as the dependent variable.

As a matter of course, the estimation method of the coefficients C _ib (kb), D _ib , and E _ib is not limited to the above-described method, and various general parameter identification methods may be applied.

  According to the above processing, learning of the coefficients used for estimation of the high band sub-band power is performed in advance using the wide band teacher signal, and therefore, various input signals input to the frequency band expanding apparatus 10 are obtained. It is possible to obtain a suitable output result, which in turn makes it possible to reproduce the music signal with higher sound quality.

The coefficients A _ib (kb) and B _ib in the above equation (2) can also be determined by the above-described coefficient learning method.

  In the above, in the high band sub-band power estimation circuit 15 of the frequency band expansion device 10, each of the high band sub-band power estimation values is calculated by linear combination of four low band sub-band powers and dips. The coefficient learning processing on the premise has been described. However, the method of estimating the high band sub-band power in the high band sub-band power estimation circuit 15 is not limited to the above-described example, and for example, the feature quantity calculation circuit 14 may use feature quantities other than dip (low band sub-band power The high band sub-band power may be calculated by calculating one or more of time fluctuation, inclination, time fluctuation of inclination, and time fluctuation of dip), and a plurality of plural frames before and after time frame J may be calculated. A linear combination of feature quantities of or a non-linear function may be used. That is, in the coefficient learning process, the coefficient estimation circuit 24 uses the feature amount, the time frame, and the function used when the high frequency sub-band power is calculated by the high frequency sub-band power estimation circuit 15 of the frequency band expansion device 10 The coefficient may be calculated (learned) under the same condition as the condition of.

<2. Second embodiment>
In the second embodiment, the encoding device and the decoding device perform the encoding process and the decoding process in the high band feature encoding method.

[Example of Functional Configuration of Encoding Device]
FIG. 11 shows an example of a functional configuration of a coding device to which the present invention is applied.

  The encoding device 30 includes a low pass filter 31, a low pass encoding circuit 32, a subband dividing circuit 33, a feature quantity calculating circuit 34, a pseudo high band sub-band power calculating circuit 35, and a pseudo high band sub-band power difference calculating circuit 36, a high band encoding circuit 37, a multiplexing circuit 38, and a low band decoding circuit 39.

  The low pass filter 31 filters the input signal at a predetermined cut-off frequency, and as a filtered signal, a low-pass signal below the cut-off frequency (hereinafter referred to as a low-pass signal) is The signal is supplied to the band division circuit 33 and the feature quantity calculation circuit 34.

  The low band encoding circuit 32 encodes the low band signal from the low pass filter 31, and supplies the resulting low band encoded data to the multiplexing circuit 38 and the low band decoding circuit 39.

  The sub-band dividing circuit 33 equally divides the input signal and the low-pass signal from the low-pass filter 31 into a plurality of sub-band signals having a predetermined bandwidth, and the feature quantity calculation circuit 34 or the pseudo high band sub-band power The difference calculation circuit 36 is supplied. More specifically, the sub-band dividing circuit 33 supplies a plurality of sub-band signals (hereinafter referred to as low-pass sub-band signals) obtained with the low-pass signal as an input to the feature quantity calculation circuit 34. Also, among the plurality of sub-band signals obtained with the input signal as the input, the sub-band dividing circuit 33 is a sub-band signal having a higher frequency than the cutoff frequency set by the low pass filter 31 (hereinafter referred to as high frequency sub-band The signal is supplied to the pseudo high band sub-band power difference calculation circuit 36.

  The feature quantity calculation circuit 34 uses at least one of a plurality of subband signals of the lowband subband signals from the subband division circuit 33 and the lowband signal from the lowpass filter 31. One or more feature quantities are calculated and supplied to the pseudo high frequency sub-band power calculation circuit 35.

  The pseudo high frequency sub-band power calculation circuit 35 generates pseudo high frequency sub-band power based on the one or more feature amounts from the feature amount calculation circuit 34, and the pseudo high frequency sub-band power difference calculation circuit 36 Supply.

  The pseudo high frequency sub-band power difference calculation circuit 36 will be described later based on the high frequency sub-band signal from the sub-band division circuit 33 and the pseudo high frequency sub-band power from the pseudo high frequency sub-band power calculation circuit 35. The pseudo high band sub-band power difference is calculated and supplied to the high band encoding circuit 37.

  The high band encoding circuit 37 encodes the pseudo high band sub-band power difference from the pseudo high band sub-band power difference calculating circuit 36, and supplies the high band encoded data obtained as a result to the multiplexing circuit 38.

  The multiplexing circuit 38 multiplexes the low-pass encoded data from the low-pass encoding circuit 32 and the high-pass encoded data from the high-pass encoding circuit 37 and outputs the result as an output code string.

  The low band decoding circuit 39 appropriately decodes the low band encoded data from the low band encoding circuit 32, and supplies the decoded data obtained as a result to the sub-band dividing circuit 33 and the feature quantity calculation circuit 34.

[Coding Process of Coding Device]
Next, the encoding process by the encoding device 30 of FIG. 11 will be described with reference to the flowchart of FIG.

  In step S111, the low pass filter 31 filters the input signal at a predetermined cutoff frequency, and the low band signal as the filtered signal is calculated by the low band encoding circuit 32, the subband dividing circuit 33, and the feature amount calculation. It supplies the circuit 34.

  In step S112, the low band encoding circuit 32 encodes the low band signal from the low pass filter 31 and supplies the resulting low band encoded data to the multiplexing circuit 38.

  In addition, regarding the encoding of the low-pass signal in step S112, an appropriate encoding method may be selected according to the encoding efficiency and the required circuit scale, and the present invention does not depend on this encoding method. .

  In step S113, the sub-band division circuit 33 equally divides the input signal and the low band signal into a plurality of sub-band signals having a predetermined bandwidth. The sub-band dividing circuit 33 supplies a low-frequency sub-band signal obtained with the low-frequency signal as an input to the feature quantity calculation circuit 34. The sub-band dividing circuit 33 sets a high-frequency sub-band signal of a band higher than the band-limiting frequency set by the low-pass filter 31 among the plurality of sub-band signals obtained with the input signal as an input. The signal is supplied to a pseudo high frequency sub-band power difference calculation circuit 36.

  In step S114, the feature amount calculation circuit 34 performs at least one of a plurality of subband signals of the lowband subband signals from the subband division circuit 33 and the lowband signal from the lowpass filter 31. Is used to calculate one or more feature quantities, which are supplied to the pseudo high band sub-band power calculation circuit 35. The feature quantity calculation circuit 34 of FIG. 11 basically has the same configuration and function as the feature quantity calculation circuit 14 of FIG. 3, and the process in step S114 is the process in step S4 of the flowchart in FIG. And the detailed description thereof will be omitted.

  In step S115, the pseudo high band sub-band power calculation circuit 35 generates pseudo high band sub-band power based on one or more feature quantities from the feature quantity calculation circuit 34, and the pseudo high band sub-band power difference The signal is supplied to the calculation circuit 36. The pseudo high frequency sub-band power calculation circuit 35 of FIG. 11 basically has the same configuration and function as the high frequency sub-band power estimation circuit 15 of FIG. 3, and the processing in step S115 is the same as FIG. The process is basically the same as the process in step S5 of the flowchart of FIG.

  In step S116, the pseudo high band sub-band power difference calculating circuit 36 is based on the high band sub-band signal from the sub-band dividing circuit 33 and the pseudo high band sub-band power from the pseudo high band sub-band power calculating circuit 35. Then, the pseudo high band sub-band power difference is calculated and supplied to the high band encoding circuit 37.

  More specifically, the pseudo high band sub-band power difference calculation circuit 36 calculates (high band) sub-band power power (ib, in a certain time frame J) for the high band sub-band signal from the sub-band division circuit 33. Calculate J). In the present embodiment, it is assumed that all subbands of the lowband subband signal and the subbands of the highband subband signal are identified using the index ib. As a method of calculating the subband power, the same method as that of the first embodiment, that is, the method using the equation (1) can be applied.

Next, the pseudo high band sub-band power difference calculating circuit 36 calculates the high band sub-band power power (ib, J) and the pseudo high band sub-band power power from the pseudo high band sub-band power calculating circuit 35 in the time frame J. _Find the difference (pseudo high band sub-band power difference) power _diff (ib, J) with _lh (ib, J). The pseudo high frequency sub-band power difference power _diff (ib, J) is obtained by the following equation (14).

・・・（１４）

... (14)

  In equation (14), the index sb + 1 represents the index of the lowest subband in the highband subband signal. Also, the index eb represents the index of the highest band sub-band to be encoded in the high band sub-band signal.

  Thus, the pseudo high band sub-band power difference calculated by the pseudo high band sub-band power difference calculating circuit 36 is supplied to the high band encoding circuit 37.

  In step S117, the high band encoding circuit 37 encodes the pseudo high band sub-band power difference from the pseudo high band sub-band power difference calculation circuit 36, and the high band encoded data obtained as a result is transmitted to the multiplexing circuit 38. Supply.

More specifically, the high band encoding circuit 37 vectorizes the pseudo high band sub-band power difference from the pseudo high band sub-band power difference calculation circuit 36 (hereinafter referred to as a pseudo high band sub-band power difference vector ) Determines which of the plurality of clusters in the feature space of the preset pseudo high band sub-band power difference belongs. Here, the pseudo high band sub-band power difference vector in a certain time frame J has the value of the pseudo high band sub-band power difference power _diff (ib, J) for each index i b as each element of the vector, (eb-sb ) Shows a vector of dimensions. Also, the feature space of the pseudo high band sub-band power difference is also a (eb-sb) -dimensional space.

  Then, the high band encoding circuit 37 measures, in the feature space of the pseudo high band sub-band power difference, the distance between each representative vector of a plurality of clusters set in advance and the pseudo high band sub-band power difference vector, The index of the cluster with the shortest distance (hereinafter, referred to as pseudo high band sub-band power difference ID) is determined, and this is supplied to the multiplexing circuit 38 as high band encoded data.

  In step S118, the multiplexing circuit 38 multiplexes the low band encoded data output from the low band encoding circuit 32 and the high band encoded data output from the high band encoding circuit 37, and outputs an output code string. Output

  By the way, as a coding apparatus in the high band feature coding method, a pseudo high band sub-band signal is generated from the low band sub-band signal in Japanese Patent Application Laid-Open No. 2007-17908, and a pseudo high band sub-band signal The power of the subband signal is compared for each subband, and the gain of the power for each subband is calculated in order to match the power of the pseudo high band subband signal with the power of the high band subband signal, and this is used as the high band feature Discloses a technique for including in a code string as information of.

  On the other hand, according to the above processing, it is only necessary to include only the pseudo high band sub-band power difference ID in the output code string as information for estimating the high band sub-band power at the time of decoding. That is, for example, when the number of clusters set in advance is 64, it is only necessary to add 6 bits of information per time frame to the code string as information for recovering the high frequency signal in the decoding apparatus, As compared with the method disclosed in Japanese Patent Application Laid-Open No. 2007-17908, the amount of information included in the code string can be reduced, so that the coding efficiency can be further improved, and consequently, the music signal has a higher sound quality. It is possible to play

  Further, in the above processing, if there is room in the amount of calculation, the low band decoding circuit 39 divides the low band signal obtained by decoding the low band encoded data from the low band encoding circuit 32 into subbands. The circuit 33 and the feature value calculation circuit 34 may be input. In the decoding process by the decoding apparatus, a feature quantity is calculated from the low band signal obtained by decoding the low band encoded data, and the power of the high band sub-band is estimated based on the feature quantity. Therefore, also in the encoding process, it is more accurate in the decoding process by the decoding apparatus to include the pseudo high band sub-band power difference ID calculated based on the feature quantity calculated from the decoded low band signal in the code string. High band sub-band power can be estimated. Therefore, the music signal can be reproduced with higher sound quality.

[Example of Functional Configuration of Decoding Device]
Next, with reference to FIG. 13, a functional configuration example of the decoding device corresponding to the encoding device 30 of FIG. 11 will be described.

  The decoding apparatus 40 includes a demultiplexing circuit 41, a low band decoding circuit 42, a subband dividing circuit 43, a feature quantity calculating circuit 44, a high band decoding circuit 45, a decoded high band sub-band power calculating circuit 46, and a decoded high band signal generation A circuit 47 and a synthesis circuit 48 are provided.

  The demultiplexing circuit 41 demultiplexes the input code sequence into high band encoded data and low band encoded data, supplies low band encoded data to the low band decoding circuit 42, and high band encoded data in the high band. The signal is supplied to the decoding circuit 45.

  The low band decoding circuit 42 decodes the low band encoded data from the demultiplexing circuit 41. The low band decoding circuit 42 supplies a low band signal (hereinafter referred to as a decoded low band signal) obtained as a result of decoding to the sub-band dividing circuit 43, the feature quantity calculating circuit 44, and the combining circuit 48.

  The subband division circuit 43 equally divides the decoded low band signal from the low band decoding circuit 42 into a plurality of subband signals having a predetermined bandwidth, and obtains the obtained subband signal (decoded low band subband signal) Are supplied to the feature value calculation circuit 44 and the decoded high band signal generation circuit 47.

  The feature value calculation circuit 44 uses at least one of a plurality of subband signals among the decoded lowband subband signals from the subband division circuit 43 and a decoded lowband signal from the lowband decoding circuit 42. Then, one or more feature quantities are calculated and supplied to the decoded high band sub-band power calculation circuit 46.

  The high band decoding circuit 45 decodes the high band encoded data from the demultiplexing circuit 41, and is prepared in advance for each ID (index) using the pseudo high band sub-band power difference ID obtained as a result. A coefficient for estimating the power of the high band sub-band (hereinafter referred to as a decoded high band sub-band power estimation coefficient) is supplied to the decoded high band sub-band power calculation circuit 46.

  The decoded high band sub-band power calculation circuit 46 decodes the high band based on the one or more feature quantities from the feature quantity calculation circuit 44 and the decoded high band sub-band power estimation coefficient from the high band decoding circuit 45. The sub-band power is calculated and supplied to the decoded high band signal generation circuit 47.

  The decoded high band signal generation circuit 47 is a decoded high band signal based on the decoded low band sub band signal from the sub band dividing circuit 43 and the decoded high band sub band power from the decoded high band sub band power calculation circuit 46. Are generated and supplied to the synthesis circuit 48.

  The synthesis circuit 48 synthesizes the decoded low band signal from the low band decoding circuit 42 and the decoded high band signal from the decoded high band signal generation circuit 47 and outputs it as an output signal.

[Decryption process of decryption device]
Next, the decoding processing by the decoding device in FIG. 13 will be described with reference to the flowchart in FIG.

  In step S131, the demultiplexing circuit 41 demultiplexes the input code string into the high band encoded data and the low band encoded data, supplies the low band encoded data to the low band decoding circuit 42, and performs the high band encoding. The data is supplied to the high band decoding circuit 45.

  In step S132, the low band decoding circuit 42 decodes the low band encoded data from the demultiplexing circuit 41, and the decoded low band signal obtained as a result is subjected to the sub-band dividing circuit 43 and the feature amount calculating circuit 44. , And the synthesis circuit 48.

  In step S133, the sub-band dividing circuit 43 equally divides the decoded low band signal from the low band decoding circuit 42 into a plurality of sub-band signals having a predetermined bandwidth, and obtains the decoded low band sub-band signal. , And supplied to the feature quantity calculation circuit 44 and the decoded high band signal generation circuit 47.

  In step S 134, the feature calculation circuit 44 performs at least one of a plurality of subband signals among the decoded lowband subband signals from the subband division circuit 43 and a decoded lowband signal from the lowband decoding circuit 42. From one or the other, one or more feature quantities are calculated and supplied to the decoded high band sub-band power calculation circuit 46. The feature quantity calculation circuit 44 in FIG. 13 basically has the same configuration and function as the feature quantity calculation circuit 14 in FIG. 3, and the process in step S134 is the process in step S4 of the flowchart in FIG. And the detailed description thereof will be omitted.

  In step S135, the high band decoding circuit 45 decodes the high band encoded data from the demultiplexing circuit 41, and uses the pseudo high band sub-band power difference ID obtained as a result, to pre-ID (index) in advance. The decoded high band sub-band power estimation coefficient prepared in step b. Is supplied to the decoded high band sub-band power calculating circuit 46.

  In step S 136, the decoded high band sub-band power calculation circuit 46 is based on the one or more feature quantities from the feature quantity calculation circuit 44 and the decoded high band sub-band power estimation coefficient from the high band decoding circuit 45. The decoded high band sub-band power is calculated and supplied to the decoded high band signal generation circuit 47. Note that the decoded high band sub-band power calculation circuit 46 of FIG. 13 has basically the same configuration and function as the high band sub-band power estimation circuit 15 of FIG. 3, and the processing in step S136 is the same as FIG. The process is basically the same as the process in step S5 of the flowchart of FIG.

  In step S 137, the decoded high band signal generation circuit 47 is based on the decoded low band sub-band signal from the sub-band division circuit 43 and the decoded high band sub-band power from the decoded high band sub-band power calculation circuit 46 Output the decoded high band signal. The decoded high frequency band signal generation circuit 47 of FIG. 13 basically has the same configuration and function as the high frequency band signal generation circuit 16 of FIG. 3, and the processing in step S137 is the step of the flowchart of FIG. Since the process is basically the same as the process in S6, the detailed description thereof is omitted.

  In step S138, the synthesis circuit 48 synthesizes the decoded low band signal from the low band decoding circuit 42 and the decoded high band signal from the decoded high band signal generation circuit 47 and outputs it as an output signal.

  According to the above process, high frequency sub-band power estimation during decoding according to the feature of the difference between the pseudo high frequency sub-band power calculated in advance during encoding and the actual high frequency sub-band power By using the coefficients, the estimation accuracy of the high band sub-band power at the time of decoding can be improved, and as a result, it is possible to reproduce the music signal with higher sound quality.

  Further, according to the above process, the information for high frequency band signal generation included in the code string is as small as only the pseudo high frequency sub-band power difference ID, so that the decoding process can be efficiently performed.

  Although the encoding process and the decoding process to which the present invention is applied have been described above, the pseudo high band sub-bands preset in the high band encoding circuit 37 of the encoding device 30 of FIG. A method of calculating a representative vector of each of a plurality of clusters in the power difference feature space and a decoded high band sub-band power estimation coefficient output by the high band decoding circuit 45 of the decoding device 40 of FIG. 13 will be described.

[Method of calculating representative vectors of a plurality of clusters in the feature space of pseudo high band sub-band power difference and decoded high band sub-band power estimation coefficients corresponding to each cluster]
As a method of obtaining representative vectors of a plurality of clusters and a decoded high band sub-band power estimation coefficient of each cluster, a high band sub in decoding is performed according to a pseudo high band sub-band power difference vector calculated in coding. It is necessary to prepare a coefficient so that the band power can be accurately estimated. Therefore, learning is performed in advance using a broadband teacher signal, and a method of determining these based on the learning result is applied.

[Functional Configuration Example of Coefficient Learning Device]
FIG. 15 illustrates a functional configuration example of a coefficient learning device that performs learning of representative vectors of a plurality of clusters and decoded high band sub-band power estimation coefficients of each cluster.

  The signal component below the cutoff frequency set by the low pass filter 31 of the encoding device 30 of the wideband teacher signal input to the coefficient learning device 50 of FIG. 15 is the low pass of the input signal to the encoding device 30 It is preferable that the decoded low-pass signal passes through the filter 31, is encoded by the low-pass encoding circuit 32, and is further decoded by the low-pass decoding circuit 42 of the decoding apparatus 40.

  The coefficient learning device 50 includes a low pass filter 51, a subband dividing circuit 52, a feature amount calculating circuit 53, a pseudo high band sub-band power calculating circuit 54, a pseudo high band sub-band power difference calculating circuit 55, a pseudo high band sub band It comprises a power differential clustering circuit 56 and a coefficient estimation circuit 57.

  The low pass filter 51, the subband dividing circuit 52, the feature quantity calculating circuit 53, and the pseudo high band sub-band power calculating circuit 54 in the coefficient learning device 50 of FIG. 15 are the same as those in the coding device 30 of FIG. The configuration and functions are basically the same as those of the low pass filter 31, the subband dividing circuit 33, the feature value calculating circuit 34, and the pseudo high band sub-band power calculating circuit 35, and thus the description thereof will be appropriately omitted. .

  That is, although the pseudo high band sub-band power difference calculation circuit 55 has the same configuration and function as the pseudo high band sub-band power difference calculation circuit 36 of FIG. 11, the calculated pseudo high band sub-band power difference The high frequency sub-band power calculated at the time of calculating the pseudo high frequency sub-band power difference is supplied to the coefficient estimation circuit 57 while being supplied to the low frequency sub-band power differential clustering circuit 56.

  The pseudo high band sub-band power difference clustering circuit 56 clusters pseudo high band sub-band power difference vectors obtained from the pseudo high band sub-band power differences from the pseudo high band sub-band power difference calculation circuit 55 and Calculate a representative vector.

  The coefficient estimation circuit 57 calculates the pseudo high band sub-band power difference based on the high band sub-band power from the pseudo high band sub-band power difference calculation circuit 55 and one or more feature quantities from the feature quantity calculation circuit 53. A high band sub-band power estimation coefficient for each cluster clustered by the clustering circuit 56 is calculated.

[Coefficient learning process of coefficient learning device]
Next, coefficient learning processing by the coefficient learning device 50 of FIG. 15 will be described with reference to the flowchart of FIG.

  The processes of steps S151 to S155 in the flowchart of FIG. 16 are the same as the processes of steps S111 and S113 to S116 in the flowchart of FIG. 12 except that the signal input to the coefficient learning device 50 is a broadband teacher signal. Therefore, the explanation is omitted.

  That is, in step S156, the pseudo high band sub-band power differential clustering circuit 56 generates a large number (a large amount of time frames) of pseudo obtained from the pseudo high band sub-band power difference from the pseudo high band sub-band power difference calculation circuit 55. The high band sub-band power difference vector is clustered into, for example, 64 clusters, and a representative vector of each cluster is calculated. As an example of the clustering method, for example, k-means clustering can be applied. The pseudo high frequency sub-band power differential clustering circuit 56 sets the barycentric vector of each cluster obtained as a result of performing clustering according to the k-means method as a representative vector of each cluster. The method of clustering and the number of clusters are not limited to those described above, and other methods may be applied.

  Also, the pseudo high band sub-band power difference clustering circuit 56 generates a pseudo high band sub-band power difference vector obtained from the pseudo high band sub-band power difference from the pseudo high band sub-band power difference calculation circuit 55 in time frame J. The distance to the 64 representative vectors is measured, and the index CID (J) of the cluster to which the representative vector having the shortest distance belongs is determined. The index CID (J) takes an integer value from 1 to the number of clusters (64 in this example). The pseudo high frequency sub-band power differential clustering circuit 56 outputs the representative vector in this manner, and supplies the index CID (J) to the coefficient estimation circuit 57.

  In step S 157, the coefficient estimation circuit 57 supplies (eb−sb) high-frequency sub-band powers and feature amounts supplied to the same time frame from the pseudo high-frequency sub-band power difference calculation circuit 55 and the feature amount calculation circuit 53. Among the many combinations, for each set (belonging to the same cluster) having the same index CID (J), calculation of the decoded high band sub-band power estimation coefficient in each cluster is performed. Although the method of calculating the coefficient by the coefficient estimation circuit 57 is the same as the method by the coefficient estimation circuit 24 in the coefficient learning device 20 of FIG. 9, it is needless to say that other methods may be used.

  According to the above processing, each of a plurality of clusters in the feature space of the pseudo high band sub-band power difference previously set in the high band coding circuit 37 of the coding device 30 of FIG. The learning of the representative vector of the signal and the decoded high band sub-band power estimation coefficient output by the high band decoding circuit 45 of the decoding device 40 of FIG. 13 is performed, so various input signals input to the coding device 30 And, it becomes possible to obtain suitable output results for various input code strings input to the decoding device 40, and it is thus possible to reproduce music signals with higher sound quality.

  Further, with regard to signal encoding and decoding, coefficient data for calculating high band sub-band power in pseudo high band sub-band power calculating circuit 35 of encoding apparatus 30 and decoded high band sub-band power calculating circuit 46 of decoding apparatus 40 It is also possible to handle as follows. That is, it is also possible to record the coefficient at the beginning of the code string by using different coefficient data depending on the type of input signal.

  For example, the coding efficiency can be improved by changing the coefficient data in accordance with a signal such as speech or jazz.

  FIG. 17 shows a code string obtained in this manner.

  The code string A in FIG. 17 is obtained by coding speech, and coefficient data α optimal for speech is recorded in the header.

  On the other hand, the code string B in FIG. 17 is obtained by encoding jazz, and coefficient data β optimal for jazz is recorded in the header.

  A plurality of such coefficient data may be prepared in advance by learning with the same music signal, and the encoding device 30 may select the coefficient data according to genre information as recorded in the header of the input signal. Alternatively, the genre may be determined by performing waveform analysis of the signal, and the coefficient data may be selected. That is, such a genre analysis method of a signal is not specifically limited.

  Further, if the calculation time is allowed, the above-mentioned learning device is built in the encoding device 30, processing is performed using a coefficient dedicated to the signal, and as shown in the code string C of FIG. It is also possible to record in the header.

  The advantages of using this approach are described below.

  The shape of the high band sub-band power has many similar parts in one input signal. By using this feature of many input signals and learning the coefficients for high-frequency sub-band power estimation separately for each input signal, the redundancy due to the presence of similar high-frequency sub-band power can be obtained. It is possible to reduce the coding efficiency and to improve the coding efficiency. Further, high-frequency sub-band power estimation can be performed with higher accuracy than learning of coefficients for high-frequency sub-band power estimation statistically with a plurality of signals.

  As described above, it is also possible to adopt a form in which coefficient data learned from an input signal at the time of encoding is inserted into several frames once.

<3. Third embodiment>
[Example of Functional Configuration of Encoding Device]
Although it has been described above that the pseudo high band sub-band power difference ID is output from the encoding device 30 to the decoding device 40 as high band encoded data, in order to obtain the decoded high band sub-band power estimation coefficient The coefficient index of {circumflex over (f)} may be used as high-pass encoded data.

  In such a case, the encoding device 30 is configured, for example, as shown in FIG. In FIG. 18, parts corresponding to the case in FIG. 11 are assigned the same reference numerals, and the description thereof will be omitted as appropriate.

  The encoding device 30 of FIG. 18 is different in that the encoding device 30 of FIG. 11 and the low-pass decoding circuit 39 are not provided, and is otherwise the same.

  In the encoding device 30 of FIG. 18, the feature quantity calculation circuit 34 calculates low band sub-band power as a feature quantity using the low band sub-band signal supplied from the sub-band division circuit 33, and the pseudo high band sub A band power calculation circuit 35 is supplied.

  Also, the pseudo high band sub-band power calculation circuit 35 includes a plurality of decoded high band sub-band power estimation coefficients obtained in advance by regression analysis, and a coefficient index for specifying those decoded high band sub-band power estimation coefficients. Are associated and recorded.

Specifically, a plurality of sets of the coefficients A _ib (kb) and the coefficients B _ib of each subband used in the calculation of the above-described equation (2) are prepared in advance as the decoded high band sub-band power estimation coefficients . For example, these coefficients A _ib (kb) and B _ib are obtained in advance by regression analysis using the least squares method, using the low band sub-band power as an explanatory variable and the high band sub-band power as an explained variable It is done. In regression analysis, an input signal composed of a low band sub-band signal and a high band sub-band signal is used as a wide band teacher signal.

  The pseudo high band sub-band power calculation circuit 35 uses the decoded high band sub-band power estimation coefficient and the feature quantity from the feature quantity calculation circuit 34 for each of the recorded high band sub-band power estimation coefficients. The pseudo high band sub-band powers of the high band side sub-bands are calculated and supplied to the pseudo high band sub-band power difference calculating circuit 36.

  The pseudo high frequency sub-band power difference calculating circuit 36 calculates high frequency sub-band power obtained from the high frequency sub-band signal supplied from the sub-band dividing circuit 33 and the pseudo high frequency sub-band from the pseudo high frequency sub-band power calculating circuit 35 Compare with band power.

  Then, the pseudo high band sub-band power difference calculation circuit 36 decodes the pseudo high band sub-band power closest to the high band sub-band power among the plurality of decoded high band sub-band power estimation coefficients as a result of comparison. The coefficient index of the high band sub-band power estimation coefficient is supplied to the high band encoding circuit 37. In other words, the coefficient index of the decoded high band sub-band power estimation coefficient is selected, which yields the high band signal of the input signal to be reproduced at the time of decoding, ie the decoded high band signal closest to the true value.

[Coding Process of Coding Device]
Next, the encoding process performed by the encoding device 30 of FIG. 18 will be described with reference to the flowchart of FIG. In addition, since the process of step S181 thru | or step S183 is the same as the process of step S111 thru | or step S113 of FIG. 12, the description is abbreviate | omitted.

  In step S 184, the feature amount calculation circuit 34 calculates a feature amount using the low band sub-band signal from the sub-band division circuit 33, and supplies it to the pseudo high band sub-band power calculation circuit 35.

  Specifically, the feature quantity calculation circuit 34 performs the calculation of the above-mentioned equation (1), and the frame J (where, sb−3 ≦ ib ≦ sb) is obtained for each lower band ib (where sb−3 ≦ ib ≦ sb). The low band sub-band power power (ib, J) of 0 ≦ J) is calculated as the feature value. That is, the low band sub-band power power (ib, J) is calculated by converting the root mean square value of the sample value of each sample of the low band sub-band signal constituting the frame J.

  In step S185, the pseudo high frequency sub-band power calculating circuit 35 calculates the pseudo high frequency sub-band power based on the feature amount supplied from the feature amount calculating circuit 34, and the pseudo high frequency sub-band power difference calculating circuit 36 Supply to

For example, the pseudo high frequency sub-band power calculation circuit 35 calculates the low frequency sub-band power power (kb, J) with the coefficient A _ib (kb) and the coefficient B _ib recorded in advance as the decoded high frequency sub-band power estimation coefficient. (However, sb−3 ≦ kb ≦ sb) is used to calculate the above equation (2) to calculate the pseudo high band sub-band power power _est (ib, J).

That is, the low band obtained by multiplying the low band subband power power (kb, J) of each low band on the low band side supplied as the feature amount by the coefficient A _ib (kb) for each subband and multiplying the coefficient A coefficient B _ib is further added to the sum of the subband powers, and a pseudo high band subband power power _est (ib, J) is obtained. The pseudo high frequency sub-band power is calculated for each high frequency side sub-band whose index is sb + 1 to eb.

  Also, the pseudo high frequency sub-band power calculation circuit 35 calculates the pseudo high frequency sub-band power for each decoded high frequency sub-band power estimation coefficient recorded in advance. For example, it is assumed that K decoded high band sub-band power estimation coefficients having a coefficient index of 1 to K (where 2 ≦ K) are prepared in advance. In this case, the pseudo high band sub-band power of each sub band is calculated for each of the K decoded high band sub-band power estimation coefficients.

  In step S 186, the pseudo high band sub-band power difference calculation circuit 36 is based on the high band sub-band signal from the sub-band division circuit 33 and the pseudo high band sub-band power from the pseudo high band sub-band power calculation circuit 35. And calculate the pseudo high band sub-band power difference.

  Specifically, the pseudo high frequency sub-band power difference calculation circuit 36 performs the same operation as the above equation (1) on the high frequency sub-band signal from the sub-band division circuit 33 and Calculate band power power (ib, J). In the present embodiment, it is assumed that all subbands of the lowband subband signal and the subbands of the highband subband signal are identified using the index ib.

Next, the pseudo high band sub-band power difference calculation circuit 36 performs the same operation as the above equation (14) to obtain the high band sub-band power power (ib, J) in the frame J and the pseudo high band sub-band Find the difference from power _est (ib, J). As a result, for each of the high band sub-band power estimation coefficients, the pseudo high band sub-band power difference power _diff (ib, J) is obtained for each high band sub-band whose index is sb + 1 to eb.

  In step S187, the pseudo high band sub-band power difference calculation circuit 36 calculates the following equation (15) for each decoded high band sub-band power estimation coefficient, and calculates the sum of squares of the pseudo high band sub-band power difference.

In equation (15), the sum of squared differences E (J, id) is the square of the pseudo high band sub-band power difference of frame J obtained for the decoded high band sub-band power estimation coefficient whose coefficient index is id It shows the sum. Also, in equation (15), power _diff (ib, J, id) is a pseudo of frame J of the subband whose index is ib, which is obtained for the decoded high frequency subband power estimation coefficient whose coefficient index is id The high band sub-band power difference power _diff (ib, J) is shown. The sum of squared differences E (J, id) is calculated for each of the K decoded high band sub-band power estimation coefficients.

  The difference square sum E (J, id) obtained in this way uses the high band sub-band power calculated from the actual high band signal and the decoded high band sub-band power estimation coefficient whose coefficient index is id It shows the degree of similarity with the pseudo high band sub-band power calculated.

  That is, it shows the error of the estimated value with respect to the true value of the high frequency sub-band power. Therefore, as the difference square sum E (J, id) is smaller, a decoded high band signal closer to the actual high band signal can be obtained by the calculation using the decoded high band sub-band power estimation coefficient. In other words, it can be said that the decoded high band sub-band power estimation coefficient that minimizes differential sum of squares E (J, id) is the estimation coefficient most suitable for frequency band expansion processing performed at the time of decoding of the output code string.

  Therefore, the pseudo high band sub-band power difference calculation circuit 36 selects a difference sum of squares with the smallest value among the K difference sums of squares E (J, id), and the decoding height corresponding to the difference sum of squares is selected. The coefficient index indicating the region subband power estimation coefficient is supplied to the high frequency encoding circuit 37.

  In step S188, the high band encoding circuit 37 encodes the coefficient index supplied from the pseudo high band sub-band power difference calculation circuit 36, and supplies the high band encoded data obtained as a result to the multiplexing circuit 38. .

  For example, in step S188, entropy coding or the like is performed on the coefficient index. Thereby, the amount of information of the high band encoded data output to the decoding device 40 can be compressed. The high band encoded data may be any information as long as it can obtain an optimal decoded high band sub-band power estimation coefficient. For example, the coefficient index is directly used as the high band encoded data. May be

  In step S189, the multiplexing circuit 38 multiplexes the low band encoded data supplied from the low band encoding circuit 32 and the high band encoded data supplied from the high band encoding circuit 37, and the result is obtained. The output code string is output, and the encoding process is completed.

  Thus, by outputting the high-pass encoded data obtained by encoding the coefficient index together with the low-pass encoded data as an output code string, the decoding apparatus 40 that receives the input of the output code string has a frequency band Decoded high band sub-band power estimation coefficients that are most suitable for the enlargement process can be obtained. This makes it possible to obtain a higher quality signal.

[Example of Functional Configuration of Decoding Device]
Also, a decoding device 40 that receives and decodes the output code string output from the encoding device 30 in FIG. 18 as an input code string is configured as shown in FIG. 20, for example. In FIG. 20, parts corresponding to the case in FIG. 13 are assigned the same reference numerals and descriptions thereof will be omitted.

  The decoding device 40 of FIG. 20 is the same as the decoding device 40 of FIG. 13 in that it comprises the demultiplexing circuit 41 to the combining circuit 48, but the decoded low band signal from the low band decoding circuit 42 is a feature amount This differs from the decoding device 40 of FIG. 13 in that it is not supplied to the calculation circuit 44.

In the decoding device 40 of FIG. 20, the high band decoding circuit 45 is a decoded high band sub-band power estimation coefficient the same as the decoded high band sub-band power estimation coefficient recorded in the pseudo high band sub-band power calculation circuit 35 of FIG. In advance. That is, a set of the coefficient A _ib (kb) as a decoded high band sub-band power estimation coefficient obtained in advance by regression analysis and the set of the coefficient B _ib are recorded in association with the coefficient index.

  The high band decoding circuit 45 decodes the high band encoded data supplied from the demultiplexing circuit 41, and the decoded high band sub-band power estimation coefficient indicated by the coefficient index obtained as a result is decoded into the high band sub band It is supplied to the power calculation circuit 46.

[Decryption process of decryption device]
Next, the decoding process performed by the decoding device 40 of FIG. 20 will be described with reference to the flowchart of FIG.

  This decoding process is started when the output code string output from the encoding device 30 is supplied to the decoding device 40 as an input code string. Note that the processes in steps S211 to S213 are the same as the processes in steps S131 to S133 in FIG.

  In step S 214, the feature amount calculation circuit 44 calculates a feature amount using the decoded low band sub-band signal from the sub-band division circuit 43, and supplies the feature amount to the decoded high band sub-band power calculation circuit 46. Specifically, the feature quantity calculation circuit 44 performs the calculation of the equation (1) described above, and the low band sub-band power power of frame J (where 0 ≦ J) for each low band sub-band ib. (ib, J) is calculated as the feature amount.

  In step S215, the high band decoding circuit 45 decodes the high band encoded data supplied from the demultiplexing circuit 41, and the decoded high band sub-band power estimation coefficient indicated by the coefficient index obtained as a result is The decoded high band sub-band power calculation circuit 46 is supplied. That is, among the plurality of decoded high band sub-band power estimation coefficients stored in advance in high band decoding circuit 45, the decoded high band sub-band power estimation coefficient indicated by the coefficient index obtained by the decoding is output.

  In step S 216, the decoded high band sub-band power calculation circuit 46 is based on the feature quantity supplied from the feature quantity calculation circuit 44 and the decoded high band sub-band power estimation coefficient supplied from the high band decoding circuit 45. The decoded high band sub-band power is calculated and supplied to the decoded high band signal generation circuit 47.

That is, the decoded high band sub-band power calculation circuit 46 calculates the coefficients A _ib (kb) and B _ib as the decoded high band sub-band power estimation coefficients and the low band sub-band power power (kb, J) as the feature quantity. (Where sb−3 ≦ kb ≦ sb) is used to calculate the above-mentioned equation (2) to calculate the decoded high band sub-band power. As a result, the decoded high frequency sub-band power is obtained for each high frequency side sub-band whose index is sb + 1 to eb.

  In step S 217, the decoded high band signal generation circuit 47 generates the decoded low band sub-band signal supplied from the sub-band dividing circuit 43 and the decoded high band sub-band power supplied from the decoded high band sub-band power calculation circuit 46. To generate a decoded high band signal.

  Specifically, the decoded high band signal generation circuit 47 performs the calculation of the above equation (1) using the decoded low band sub-band signal to calculate the low band sub-band power for each low band sub-band. . Then, the decoded high band signal generation circuit 47 performs the calculation of the equation (3) described above using the obtained low band sub-band powers and the decoded high band sub-band powers to generate the high band sub-bands. The amount of gain G (ib, J) is calculated.

  Furthermore, the decoded high band signal generation circuit 47 performs the calculation of the above-mentioned equations (5) and (6) using the gain amount G (ib, J) and the decoded low band sub-band signal to obtain the high band A high-frequency sub-band signal x3 (ib, n) is generated for each side sub-band.

  That is, the decoded high band signal generation circuit 47 amplitude-modulates the decoded low band sub-band signal x (ib, n) according to the ratio of the low band sub-band power and the decoded high band sub-band power, The obtained decoded low band sub-band signal x2 (ib, n) is further frequency modulated. Thereby, the signal of the frequency component of the low band side subband is converted to the signal of the frequency component of the high band side subband, and the high band sub-band signal x3 (ib, n) is obtained.

  The process of obtaining the high-frequency sub-band signal of each sub-band in this way is, more specifically, the following process.

  Four subbands arranged in series in the frequency domain are referred to as band blocks, and one band block (hereinafter, particularly the low band) from four subbands on the low band side whose index is sb to sb-3 It is assumed that the frequency band is divided so as to constitute a block). At this time, for example, a band composed of sub-bands whose high-frequency side indexes are sb + 1 to sb + 4 is taken as one band block. Hereinafter, a band block on the high band side, that is, a sub-band whose index is sb + 1 or more will be particularly referred to as a high band block.

  Now, it is assumed that one sub-band constituting a high-pass block is focused and a high-pass sub-band signal of the sub-band (hereinafter, referred to as a target sub-band) is generated. First, the decoded high band signal generation circuit 47 specifies the sub band of the low band block in the same positional relationship as the position of the target sub band in the high band block.

  For example, if the index of the sub-band of interest is sb + 1, the sub-band of interest is the lowest frequency band of the high band block, and so the sub-bands of the low band block having the same positional relationship as the sub-band of interest Is a sub-band whose index is sb-3.

  In this way, when a low band block subband in the same positional relationship as the target subband is identified, the low band subband power of the subband and the decoded low band subband signal, and the decoded high band of the target subband are identified. The sub-band power is used to generate a high-frequency sub-band signal of the sub-band of interest.

  That is, the decoded high band sub-band power and the low band sub-band power are substituted into equation (3), and the amount of gain according to the ratio of those powers is calculated. Then, the decoded low band sub-band signal is multiplied by the calculated gain amount, and the decoded low band sub-band signal multiplied by the gain amount is frequency-modulated by the calculation of equation (6), and the high of the noted sub-band Region sub-band signal.

  By the above processing, high band sub-band signals of each high band sub-band can be obtained. Then, the decoded high band signal generation circuit 47 further performs the calculation of the equation (7) described above, obtains the sum of the obtained high band sub-band signals, and generates a decoded high band signal. The decoded high band signal generation circuit 47 supplies the obtained decoded high band signal to the synthesis circuit 48, and the process proceeds from step S217 to step S218.

  In step S218, the synthesis circuit 48 synthesizes the decoded low band signal from the low band decoding circuit 42 and the decoded high band signal from the decoded high band signal generation circuit 47 and outputs it as an output signal. Then, the decoding process ends.

  As described above, according to the decoding apparatus 40, the coefficient index is obtained from the high band encoded data obtained by the demultiplexing of the input code string, and the decoded high band sub-band power estimation coefficient indicated by the coefficient index is obtained. Since the decoded high band sub-band power is calculated using this, the estimation accuracy of the high band sub-band power can be improved. This makes it possible to reproduce music signals with higher sound quality.

<4. Fourth Embodiment>
[Coding Process of Coding Device]
Further, although the case where only the coefficient index is included in the high-pass encoded data has been described above as an example, other information may be included.

  For example, if the coefficient index is included in the high band encoded data, the decoded high band sub-band power estimation coefficient that provides the decoded high band sub-band power closest to the high band sub-band power of the actual high band signal Can be known on the decoding device 40 side.

However, the actual high frequency sub-band power (true value) and the decoded high frequency sub-band power (estimated value) obtained on the decoding device 40 side are calculated by the pseudo high frequency sub-band power difference calculation circuit 36 A difference occurs by approximately the same value as the pseudo high band sub-band power difference power _diff (ib, J).

  Therefore, if high frequency encoded data includes not only the coefficient index but also the pseudo high frequency sub-band power difference of each sub-band, the decoding apparatus 40 can decode high relative to the actual high frequency sub-band power. It is possible to know the approximate error of the sub-band power. Then, this error can be used to further improve the estimation accuracy of the high band sub-band power.

  The encoding process and the decoding process in the case where the high frequency band encoded data includes the pseudo high frequency sub-band power difference will be described below with reference to the flowcharts of FIGS. 22 and 23.

  First, the encoding process performed by the encoding device 30 of FIG. 18 will be described with reference to the flowchart of FIG. Note that the processes of steps S241 to S246 are the same as the processes of steps S181 to S186 of FIG. 19, and thus the description thereof will be omitted.

  In step S247, the pseudo high band sub-band power difference calculation circuit 36 performs the calculation of the above equation (15) to calculate the difference square sum E (J, id) for each decoded high band sub-band power estimation coefficient. Do.

  Then, the pseudo high frequency sub-band power difference calculation circuit 36 selects a difference square sum with the smallest value among the difference square sums E (J, id), and the decoded high frequency sub-band corresponding to the difference square sum The coefficient index indicating the power estimation coefficient is supplied to the high band encoding circuit 37.

Furthermore, the pseudo high band sub-band power difference calculation circuit 36 calculates the pseudo high band sub-band power difference power _{diff of} each sub band obtained for the decoded high band sub-band power estimation coefficient corresponding to the selected difference square sum. , J) are supplied to the high band encoding circuit 37.

  In step S248, the high band encoding circuit 37 encodes the coefficient index and the pseudo high band sub-band power difference supplied from the pseudo high band sub-band power difference calculation circuit 36, and the resulting high band coding The data is supplied to the multiplexing circuit 38.

  As a result, the pseudo high band sub-band power difference of each high band side index whose index is sb + 1 to eb, that is, the estimation error of the high band sub-band power is supplied to the decoding device 40 as high band encoded data. It will be done.

  After the high band encoded data is obtained, the process of step S249 is performed to end the encoding process. However, the process of step S249 is the same as the process of step S189 of FIG. I omit it.

  As described above, if the pseudo high band sub-band power difference is included in the high band encoded data, the estimation accuracy of the high band sub-band power can be further improved in the decoding apparatus 40, and the sound quality can be further improved. Music signal can be obtained.

[Decryption process of decryption device]
Next, the decoding process performed by the decoding device 40 of FIG. 20 will be described with reference to the flowchart of FIG. Note that the processes of steps S 271 to S 274 are the same as the processes of steps S 211 to S 214 of FIG. 21, and thus the description thereof will be omitted.

  In step S275, the high frequency band decoding circuit 45 decodes the high frequency band encoded data supplied from the demultiplexing circuit 41. Then, the high band decoding circuit 45 decodes the high band sub-band power estimation coefficient indicated by the coefficient index obtained by the decoding and the pseudo high band sub-band power difference of each sub band obtained by the decoding. It supplies to the sub band power calculation circuit 46.

  In step S 276, the decoded high band sub-band power calculation circuit 46 is based on the feature quantity supplied from the feature quantity calculation circuit 44 and the decoded high band sub-band power estimation coefficient supplied from the high band decoding circuit 45. Calculate the decoded high band sub-band power. In step S276, the same process as step S216 in FIG. 21 is performed.

  In step S277, the decoded high band sub-band power calculation circuit 46 adds the pseudo high band sub-band power difference supplied from the high band decoding circuit 45 to the decoded high band sub-band power to obtain the final decoded high band Subband power is supplied to the decoded high band signal generation circuit 47. That is, the pseudo high band sub-band power difference of the same sub band is added to the calculated high band sub-band power of each sub band.

  Thereafter, the processes of steps S278 and S279 are performed, and the decoding process ends. However, since these processes are similar to steps S217 and S218 of FIG. 21, the description thereof is omitted.

  As described above, the decoding device 40 obtains the coefficient index and the pseudo high band sub-band power difference from the high band encoded data obtained by the demultiplexing of the input code string. Then, the decoding device 40 calculates the decoded high frequency sub-band power using the decoded high frequency sub-band power estimation coefficient indicated by the coefficient index and the pseudo high frequency sub-band power difference. As a result, the estimation accuracy of the high band sub-band power can be improved, and the music signal can be reproduced with higher sound quality.

  Note that the difference between estimated values of high band sub-band power occurring between encoding device 30 and decoding device 40, that is, the difference between pseudo high band sub-band power and decoded high band sub-band power (hereinafter referred to as inter-device estimated difference) ) May be considered.

  In such a case, for example, the pseudo high band sub-band power difference to be high band encoded data is corrected with the inter-device estimated difference, or the high band encoded data includes the inter-device estimated difference, On the device 40 side, the pseudo high band sub-band power difference is corrected by the inter-device estimated difference. Furthermore, the inter-apparatus estimated difference is recorded in advance on the decoding apparatus 40 side, and the inter-apparatus estimated difference is added to the pseudo high frequency sub-band power difference to perform the correction. Good. This makes it possible to obtain a decoded high band signal closer to the actual high band signal.

<5. Fifth Embodiment>
In the encoding device 30 of FIG. 18, the pseudo high band sub-band power difference calculation circuit 36 has been described as selecting an optimum one from a plurality of coefficient indexes using the difference square sum E (J, id) as an index. The coefficient index may be selected using an index different from the sum of squared differences.

  For example, as an index for selecting a coefficient index, an evaluation value may be used in consideration of the root mean square value, the maximum value, the average value, and the like of the residuals of the high band sub-band power and the pseudo high band sub-band power. In such a case, the encoding device 30 of FIG. 18 performs the encoding process shown in the flowchart of FIG.

  Hereinafter, the encoding process by the encoding device 30 will be described with reference to the flowchart in FIG. The processes in steps S301 to S305 are the same as the processes in steps S181 to S185 in FIG. 19, and thus the description thereof will be omitted. When the processes of steps S301 to S305 are performed, the pseudo high band sub-band power of each sub band is calculated for each of the K decoded high band sub-band power estimation coefficients.

  In step S306, the pseudo high frequency sub-band power difference calculation circuit 36 calculates an evaluation value Res (id, J) using the current frame J to be processed for each of the K decoded high frequency sub-band power estimation coefficients. Calculate

  Specifically, the pseudo high frequency sub-band power difference calculation circuit 36 performs the same operation as the above equation (1) using the high frequency sub-band signal of each sub-band supplied from the sub-band division circuit 33. Then, the high band sub-band power power (ib, J) in the frame J is calculated. In the present embodiment, it is assumed that all subbands of the lowband subband signal and the subbands of the highband subband signal are identified using the index ib.

When the high-frequency sub-band power power (ib, J) is obtained, the pseudo high-frequency sub-band power difference calculation circuit 36 calculates the following equation (16) to obtain the residual mean square value Res _std (id, J) calculate.

That is, for each subband on the high band side whose index is sb + 1 to eb, the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) of frame J And the sum of squares of the differences is taken as the residual mean square value Res _std (id, J). The pseudo high band sub-band power power _est (ib, id, J) is a pseudo band J of the sub band whose index is ib, which is obtained for the decoded high band sub-band power estimation coefficient whose coefficient index is id. The high band sub-band power is shown.

Subsequently, the pseudo high frequency sub-band power difference calculation circuit 36 calculates the following equation (17) to calculate the residual maximum value Res _max (id, J).

In equation (17), max _ib {| power (ib, J) -power _est (ib, id, J) |} is the high-frequency sub-band power of each subband whose index is sb + 1 to eb The largest one of the absolute values of the difference between power (ib, J) and the pseudo high band sub-band power _est (ib, id, J) is shown. Therefore, the maximum absolute value of the difference between the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) in frame J is the residual maximum value Res _max (id, J).

Also, the pseudo high frequency sub-band power difference calculation circuit 36 calculates the following equation (18), and calculates the residual average value Res _ave (id, J).

That is, for each subband on the high band side whose index is sb + 1 to eb, the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) of frame J The differences of are calculated and the sum of the differences is calculated. Then, the absolute value of the value obtained by dividing the sum of the obtained differences by the number of subbands on the high frequency side (eb−sb) is taken as the residual average value Res _ave (id, J). The residual average value Res _ave (id, J) indicates the magnitude of the average value of the estimation error of each subband in which the code is considered.

Furthermore, when the residual mean square value Res _std (id, J), the residual maximum value Res _max (id, J), and the residual mean value Res _ave (id, J) are obtained, the pseudo high band sub-band power The difference calculation circuit 36 calculates the following expression (19), and calculates a final evaluation value Res (id, J).

That is, the residual mean square value Res _std (id, J), the residual maximum value Res _max (id, J), and the residual mean value Res _ave (id, J) are weighted and added, and the final evaluation is performed. The value Res (id, J) is taken. In Equation (19), W _max and W _ave are predetermined weights, for example, W _max = 0.5, W _ave = 0.5, and the like.

  The pseudo high band sub-band power difference calculation circuit 36 performs the above-described processing, and evaluates the evaluation value Res (id, J) for each of the K decoded high band sub-band power estimation coefficients, that is, for each of the K coefficient indices id. Calculate).

  In step S307, the pseudo high frequency sub-band power difference calculation circuit 36 selects the coefficient index id based on the evaluation value Res (id, J) for each of the obtained coefficient indexes id.

  The evaluation value Res (id, J) obtained by the above processing is calculated using the high band sub-band power calculated from the actual high band signal and the decoded high band sub-band power estimation coefficient whose coefficient index is id. It shows the degree of similarity with the calculated pseudo high band sub-band power. That is, it indicates the magnitude of the estimation error of the high frequency component.

  Therefore, as the evaluation value Res (id, J) is smaller, a decoded high band signal closer to the actual high band signal can be obtained by the calculation using the decoded high band sub-band power estimation coefficient. Therefore, the pseudo high frequency sub-band power difference calculation circuit 36 selects an evaluation value with the smallest value among the K evaluation values Res (id, J), and the decoded high frequency sub-band corresponding to the evaluation value. The coefficient index indicating the power estimation coefficient is supplied to the high band encoding circuit 37.

  After the coefficient index is output to the high band encoding circuit 37, the processes of steps S308 and S309 are performed and the encoding process ends, but these processes are the same as steps S188 and S189 of FIG. Therefore, the description is omitted.

As described above, the encoding device 30 calculates the residual square average value Res _std (id, J), the residual maximum value Res _max (id, J), and the residual average value Res _ave (id, J). The estimated value Res (id, J) is used to select the coefficient index of the optimal decoded high band sub-band power estimation coefficient.

  By using the evaluation value Res (id, J), it is possible to evaluate the estimation accuracy of the high band sub-band power using more evaluation scales than in the case of using the difference square sum, and hence a more appropriate decoded high band Subband power estimation coefficients can be selected. As a result, the decoding device 40 receiving the input of the output code string can obtain the decoded high band sub-band power estimation coefficient most suitable for the frequency band expansion processing, so that a higher quality sound signal can be obtained. Become.

Modified Example 1
In addition, when the encoding process described above is performed for each frame of the input signal, in the stationary part where the temporal variation of the high band sub-band power of each sub band of the high band side of the input signal is small In some cases, different coefficient indexes may be selected.

  That is, since high frequency sub-band powers of the respective frames have almost the same value in successive frames constituting a stationary part of the input signal, the same coefficient index should be continuously selected in those frames. However, in the section of these continuous frames, the coefficient index selected for each frame changes, and as a result, the high frequency component of the sound reproduced on the decoding device 40 side may not be steady. As a result, the reproduced sound is unnatural in hearing.

  Therefore, when the coefficient index is selected in the encoding device 30, the estimation result of the high-frequency component in the temporally previous frame may be considered. In such a case, the encoding device 30 of FIG. 18 performs the encoding process shown in the flowchart of FIG.

  Hereinafter, the encoding process by the encoding device 30 will be described with reference to the flowchart in FIG. The processes in steps S331 to S336 are the same as the processes in steps S301 to S306 in FIG. 24, and thus the description thereof will be omitted.

  In step S337, the pseudo high frequency sub-band power difference calculation circuit 36 calculates an evaluation value ResP (id, J) using the past frame and the current frame.

  Specifically, the pseudo high frequency sub-band power difference calculating circuit 36 decodes the decoding height of the finally selected coefficient index for the frame (J-1) temporally before the frame J to be processed. The pseudo high band sub-band power of each sub band obtained by using the sub band power estimation coefficient is recorded. Here, the finally selected coefficient index is a coefficient index which is encoded by the high band encoding circuit 37 and output to the decoding device 40.

In the following, in particular, the coefficient index id selected in the frame (J-1) is set as id _selected (J-1). Also, the pseudo high band sub of the sub-band whose index is ib (where sb + 1 ≦ ib ≦ eb) obtained using the decoded high band sub-band power estimation coefficient of the coefficient index id _selected (J−1) We will continue to explain band power as power _est (ib, id _selected (J-1), J-1).

The pseudo high frequency sub-band power difference calculation circuit 36 first calculates the following equation (20), and calculates the estimated residual mean square value ResP _std (id, J).

That is, the pseudo high band sub-band power power _est (ib, id _selected (J-1), J-1) of the frame (J-1) for each high band sub-band whose index is sb + 1 to eb And the difference between the pseudo high band sub-band power power _est (ib, id, J) of the frame J is obtained. Then, the sum of squares of these differences is taken as the estimated residual mean square value ResP _std (id, J). The pseudo high band sub-band power power _est (ib, id, J) is a pseudo band J of the sub band whose index is ib, which is obtained for the decoded high band sub-band power estimation coefficient whose coefficient index is id. The high band sub-band power is shown.

Since this estimated residual mean square value ResP _std (id, J) is the difference square sum of pseudo high band sub-band powers between temporally consecutive frames, the estimated residual mean square value ResP _std (id, J) The smaller the), the smaller the temporal change of the estimated value of the high-frequency component.

Subsequently, the pseudo high frequency sub-band power difference calculating circuit 36 calculates the following equation (21) to calculate the estimated residual maximum value ResP _max (id, J).

In equation (21), max _ib {| power _est (ib, id _selected (J-1), J-1) -power _est (ib, id, J) |} has an index of sb + 1 to eb. Absolute value of the difference between the pseudo high band sub-band power power _est (ib, id _selected (J-1), J-1) and the pseudo high band sub-band power power _est (ib, id, J) of each sub-band being It shows the largest of them. Therefore, the maximum value of the absolute value of the difference between the pseudo high band sub-band powers between temporally consecutive frames is taken as the estimated residual maximum value ResP _max (id, J).

The smaller the value of the estimated residual maximum value ResP _max (id, J), the closer the estimation result of the high-frequency component between consecutive frames is.

When the estimated residual maximum value ResP _max (id, J) is obtained, the pseudo high frequency sub-band power difference calculating circuit 36 next calculates the following equation (22), and the estimated residual average value ResP _ave (id, Calculate J).

That is, the pseudo high band sub-band power power _est (ib, id _selected (J-1), J-1) of the frame (J-1) for each high band sub-band whose index is sb + 1 to eb And the difference between the pseudo high band sub-band power power _est (ib, id, J) of the frame J is obtained. Then, the absolute value of the value obtained by dividing the sum of the differences of each subband by the number of subbands on the high frequency side (eb−sb) is taken as the estimated residual average value ResP _ave (id, J). . The estimated residual average value ResP _ave (id, J) indicates the magnitude of the average value of the difference between estimated values of subbands between frames in which the code is considered.

Furthermore, when the estimated residual mean square value ResP _std (id, J), the estimated residual maximum value ResP _max (id, J), and the estimated residual mean value ResP _ave (id, J) are obtained, the pseudo high region is obtained. The subband power difference calculation circuit 36 calculates the following expression (23), and calculates an evaluation value ResP (id, J).

That is, the estimated residual mean square value ResP _std (id, J), the estimated residual maximum value ResP _max (id, J), and the estimated residual average value ResP _ave (id, J) are weighted and added, and evaluation is performed. The value ResP (id, J) is taken. In equation (23), W _max and W _ave are predetermined weights, for example, W _max = 0.5, W _ave = 0.5, and the like.

  Thus, when the evaluation value ResP (id, J) using the past frame and the current frame is calculated, the process proceeds from step S337 to step S338.

In step S338, the pseudo high frequency sub-band power difference calculation circuit 36 calculates the following equation (24) to calculate the final evaluation value Res _all (id, J).

That is, the obtained evaluation value Res (id, J) and the evaluation value ResP (id, J) are weighted and added. In equation (24), W _p (J) is, for example, a weight defined by the following equation (25).

Further, power _r (J) in equation (25) is a value determined by the following equation (26).

This power _r (J) indicates the average of the difference between the high frequency sub-band powers of the frame (J-1) and the frame J. Further, W _p (J) from formulas (25), when power _r (J) is a value within the predetermined range near 0 becomes a value close to about 1 power _r (J) is small, power _r When (J) is larger than a predetermined range of values, it becomes zero.

Here, in the case where power _r (J) is a value within a predetermined range near 0, the average of differences in high band sub-band powers between consecutive frames will be small to some extent. In other words, the temporal variation of the high frequency component of the input signal is small, and the current frame of the input signal is a steady part.

The weight W _p (J) has a value closer to 1 as the high frequency component of the input signal is more steady, and conversely, the value becomes closer to 0 when the high frequency component is not steady. Therefore, in the evaluation value Res _all (id, J) shown in the equation (24), the smaller the temporal variation of the high-pass component of the input signal, the comparison result with the estimation result of the high-pass component in the immediately preceding frame The contribution rate of the evaluation value ResP (id, J) with the above as the evaluation scale is increased.

As a result, in the stationary part of the input signal, a decoded high band sub-band power estimation coefficient that can obtain an approximation close to the estimation result of the high band component in the immediately preceding frame is selected. Can play high quality voices. Conversely, in the non-stationary part of the input signal, the term of the evaluation value ResP (id, J) in the evaluation value Res _all (id, J) is 0, and a decoded high band signal closer to the actual high band signal is obtained.

The pseudo high frequency sub-band power difference calculation circuit 36 performs the above processing to calculate an evaluation value Res _all (id, J) for each of the K decoded high frequency sub-band power estimation coefficients.

In step S339, the pseudo high frequency sub-band power difference calculation circuit 36 selects the coefficient index id based on the evaluation value Res _all (id, J) for each of the obtained high frequency sub-band power estimation coefficients.

The evaluation value Res _all (id, J) obtained by the above processing is a linear combination of the evaluation value Res (id, J) and the evaluation value ResP (id, J) using weights. As described above, as the evaluation value Res (id, J) is smaller, a decoded high band signal closer to the actual high band signal can be obtained. In addition, as the evaluation value ResP (id, J) decreases, a decoded high band signal closer to the decoded high band signal of the immediately preceding frame can be obtained.

Therefore, as the evaluation value Res _all (id, J) is smaller, a more appropriate decoded high band signal can be obtained. Therefore, the pseudo high frequency sub-band power difference calculation circuit 36 selects an evaluation value with the smallest value among the K evaluation values Res _all (id, J), and the decoded high frequency sub corresponding to the evaluation value is selected. The coefficient index indicating the band power estimation coefficient is supplied to the high band encoding circuit 37.

  After the coefficient index is selected, the processes of steps S340 and S341 are performed to end the encoding process. However, since these processes are similar to steps S308 and S309 of FIG. I omit it.

As described above, in the encoding device 30, the evaluation value Res _all (id, J) obtained by linear combination of the evaluation value Res (id, J) and the evaluation value ResP (id, J) is used to perform an optimum. The coefficient index of the proper decoded high band sub-band power estimation coefficient is selected.

If evaluation value Res _all (id, J) is used, as in the case where evaluation value Res (id, J) is used, more evaluation measures are used to select more appropriate decoded high band sub-band power estimation coefficients be able to. Moreover, by using the evaluation value Res _all (id, J), it is possible to suppress temporal fluctuation in the steady part of the high frequency component of the signal to be reproduced on the decoding device 40 side, and a signal with higher sound quality You can get

<Modification 2>
By the way, in frequency band expansion processing, in order to obtain voice with higher sound quality, the lower band side becomes more important for hearing. That is, the higher the estimation accuracy of the sub-band closer to the low band side among the high band side sub-bands, the higher the sound quality can be reproduced.

  Therefore, when the evaluation value of each decoded high band sub-band power estimation coefficient is calculated, the lower band may be weighted. In such a case, the encoding device 30 of FIG. 18 performs the encoding process shown in the flowchart of FIG.

  Hereinafter, the encoding process by the encoding device 30 will be described with reference to the flowchart in FIG. Note that the processes of step S371 to step S375 are the same as the processes of step S331 to step S335 of FIG. 25, and thus the description thereof will be omitted.

In step S 376, the pseudo high band sub-band power difference calculation circuit 36 calculates an evaluation value ResW _band (id, J) using the current frame J to be processed for each of the K decoded high band sub-band power estimation coefficients. Calculate).

  Specifically, the pseudo high frequency sub-band power difference calculation circuit 36 performs the same operation as the above equation (1) using the high frequency sub-band signal of each sub-band supplied from the sub-band division circuit 33. Then, the high band sub-band power power (ib, J) in the frame J is calculated.

When the high band sub-band power power (ib, J) is obtained, the pseudo high band sub-band power difference calculating circuit 36 calculates the following equation (27), and the residual mean square value Res _std W _band (id, J Calculate).

That is, for each subband on the high band side whose index is sb + 1 to eb, the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) of frame J The differences of are calculated, and the differences W are multiplied by the weight W _band (ib) for each subband. Then, the sum of squares of differences multiplied by the weight W _band (ib) is taken as the residual mean square value Res _std W _band (id, J).

Here, the weight W _band (ib) (where sb + 1 ≦ ib ≦ eb) is defined by, for example, the following equation (28). The value of this weight W _band (ib) increases in the lower side subbands.

Subsequently, the pseudo high frequency sub-band power difference calculation circuit 36 calculates the residual maximum value Res _max W _band (id, J). Specifically, weight is added to the difference between the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) of each sub-band whose index is sb + 1 to eb The maximum value of the absolute values of those multiplied by W _band (ib) is taken as the residual maximum value Res _max W _band (id, J).

Further, the pseudo high frequency sub-band power difference calculation circuit 36 calculates the residual average value Res _ave W _band (id, J).

Specifically, for each subband having an index of sb + 1 to eb, the difference between the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) is obtained The weight W _band (ib) is multiplied, and the sum of the differences multiplied by the weight W _band (ib) is obtained. Then, the absolute value of the value obtained by dividing the obtained sum of differences by the number of sub-bands on the high frequency side (eb−sb) is taken as the residual average value Res _ave W _band (id, J).

Furthermore, the pseudo high frequency sub-band power difference calculation circuit 36 calculates an evaluation value ResW _band (id, J). That is, residual mean square value Res _std W _band (id, J), residual maximum value Res _max W _band (id, J) multiplied by weight W _max , and residual mean value multiplied by weight W _ave the sum of the _{Res ave W band (id, J} ) is the evaluation value ResW _band (id, J).

In step S377, the pseudo high frequency sub-band power difference calculation circuit 36 calculates an evaluation value ResPW _band (id, J) using the past frame and the current frame.

  Specifically, the pseudo high frequency sub-band power difference calculating circuit 36 decodes the decoding height of the finally selected coefficient index for the frame (J-1) temporally before the frame J to be processed. The pseudo high band sub-band power of each sub band obtained by using the sub band power estimation coefficient is recorded.

The pseudo high frequency sub-band power difference calculation circuit 36 first calculates an estimated residual square average value ResP _std W _band (id, J). That is, the pseudo high band sub-band power power _est (ib, id _selected (J-1), J-1) and the pseudo high band sub-band for each high band sub-band having an index of sb + 1 to eb The difference of power power _est (ib, id, J) is obtained and multiplied by the weight W _band (ib). Then, the sum of squares of differences multiplied by the weight W _band (ib) is taken as the estimated residual mean square value ResP _std W _band (id, J).

Subsequently, the pseudo high frequency sub-band power difference calculating circuit 36 calculates the estimated residual maximum value ResP _max W _band (id, J). Specifically, the pseudo high band sub-band power _est (ib, id _selected (J-1), J-1) and the pseudo high band sub-band power power _{est of} each sub-band whose index is sb + 1 to eb Of the difference between (ib, id, J) and the weight W _band (ib) multiplied, the maximum value of the absolute values is taken as the estimated residual maximum value ResP _max W _band (id, J).

Next, the pseudo high frequency subband power difference calculation circuit 36 calculates an estimated residual mean value _{ResP ave W band (id, J} ). Specifically, for each subband whose index is sb + 1 to eb, the pseudo high band sub-band power power _est (ib, id _selected (J-1), J-1) and the pseudo high band sub-band power The difference of power _est (ib, id, J) is obtained and multiplied by the weight W _band (ib). Then, the absolute value of the value obtained by dividing the sum of the differences multiplied by the weight W _band (ib) by the number of subbands on the high frequency side (eb−sb) is the estimated residual average value ResP _ave W _band (Id, J).

Further, the pseudo high band sub-band power difference calculation circuit 36 calculates the estimated residual maximum value ResP _max W _band (id, J) multiplied by the estimated residual square average value ResP _std W _band (id, J) and the weight W _max. And the estimated residual average value ResP _ave W _band (id, J) multiplied by the weight W _ave is obtained as an evaluation value ResPW _band (id, J).

In step S378, the pseudo high frequency sub-band power difference calculation circuit 36 calculates an evaluation value ResPW _band (id, J) obtained by multiplying the evaluation value ResW _band (id, J) by the weight W _p (J) in equation (25). And the final evaluation value Res _all W _band (id, J) is calculated. The evaluation value Res _all W _band (id, J) is calculated for each of the K decoded high frequency sub-band power estimation coefficients.

Then, after that, the processing of step S379 to step S381 is performed and the encoding processing ends, but these processing is the same as the processing of step S339 to step S341 of FIG. In step S379, among the K coefficient indexes, the one with the smallest evaluation value Res _all W _band (id, J) is selected.

  As described above, by weighting each subband so that the lower subbands are weighted, it is possible to obtain higher quality voice at the decoding device 40 side.

Although it has been described above that the selection of the decoded high band sub-band power estimation coefficient is performed based on the evaluation value Res _all W _band (id, J), the decoded high band sub-band power estimation coefficient is It may be selected based on the value ResW _band (id, J).

<Modification 3>
Furthermore, since human auditory sense has a characteristic that the higher the amplitude (power) frequency band is perceived well, each decoded high band sub-band power estimation is carried out so that the higher power sub-band is emphasized. An evaluation value for the coefficient may be calculated.

  In such a case, the encoding device 30 of FIG. 18 performs the encoding process shown in the flowchart of FIG. The coding process by the coding device 30 will be described below with reference to the flowchart in FIG. The processes of steps S401 to S405 are the same as the processes of steps S331 to S335 of FIG. 25, and thus the description thereof is omitted.

In step S406, the pseudo high band sub-band power difference calculation circuit 36 calculates an evaluation value ResW _power (id, J) using the current frame J to be processed for each of the K decoded high band sub-band power estimation coefficients. Calculate).

  Specifically, the pseudo high frequency sub-band power difference calculation circuit 36 performs the same operation as the above equation (1) using the high frequency sub-band signal of each sub-band supplied from the sub-band division circuit 33. Then, the high band sub-band power power (ib, J) in the frame J is calculated.

When the high frequency sub-band power power (ib, J) is obtained, the pseudo high frequency sub-band power difference calculation circuit 36 calculates the following equation (29), and the residual mean square value Res _std W _power (id, J Calculate).

That is, for each subband on the high frequency side having an index of sb + 1 to eb, the difference between the high frequency sub-band power power (ib, J) and the pseudo high frequency sub-band power power _est (ib, id, J) is These differences are then multiplied by the per-subband weight W _power (power (ib, J)). Then, the sum of squares of differences multiplied by the weight W _power (power (ib, J)) is taken as the residual mean square value Res _std W _power (id, J).

Here, the weight W _power (power (ib, J)) (where sb + 1 ≦ ib ≦ eb) is defined by the following equation (30), for example. The value of the weight W _power (power (ib, J)) increases as the high band sub-band power power (ib, J) of the sub-band increases.

Subsequently, the pseudo high frequency sub-band power difference calculating circuit 36 calculates the residual maximum value Res _max W _power (id, J). Specifically, weight is added to the difference between the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) of each sub-band whose index is sb + 1 to eb The maximum value of the absolute values of those multiplied by W _power (power (ib, J)) is taken as the residual maximum value Res _max W _power (id, J).

Further, the pseudo high frequency sub-band power difference calculation circuit 36 calculates the residual average value Res _ave W _power (id, J).

Specifically, for each subband having an index of sb + 1 to eb, the difference between the high band sub-band power power (ib, J) and the pseudo high band sub-band power power _est (ib, id, J) is obtained are by weight _{W power (power (ib, J} )) is multiplied by the weight _{W power (power (ib, J} )) there is obtained the sum of the multiplied difference. Then, the absolute value of the value obtained by dividing the sum of the obtained differences by the number of subbands on the high frequency side (eb−sb) is taken as the residual average value Res _ave W _power (id, J).

Furthermore, the pseudo high frequency sub-band power difference calculation circuit 36 calculates an evaluation value ResW _power (id, J). That is, the residual mean square value Res _std W _power (id, J), the residual maximum value Res _max W _power (id, J) multiplied by the weight W _max , and the residual mean value multiplied by the weight W _{ave Res ave W power (id, J} ) the sum of is the evaluation value ResW _power (id, J).

In step S407, the pseudo high frequency sub-band power difference calculation circuit 36 calculates an evaluation value ResPW _power (id, J) using the past frame and the current frame.

  Specifically, the pseudo high frequency sub-band power difference calculating circuit 36 decodes the decoding height of the finally selected coefficient index for the frame (J-1) temporally before the frame J to be processed. The pseudo high band sub-band power of each sub band obtained by using the sub band power estimation coefficient is recorded.

The pseudo high frequency sub-band power difference calculating circuit 36 first calculates an estimated residual square average value ResP _std W _power (id, J). That is, the pseudo high band sub-band power power _est (ib, id _selected (J-1), J-1) and the pseudo high band sub-band for each high band sub-band having an index of sb + 1 to eb The difference of the power power _est (ib, id, J) is obtained and multiplied by the weight W _power (power (ib, J)). Then, the sum of squares of differences multiplied by the weight W _power (power (ib, J)) is taken as the estimated residual mean square value ResP _std W _power (id, J).

Subsequently, the pseudo high frequency sub-band power difference calculating circuit 36 calculates the estimated residual maximum value ResP _max W _power (id, J). Specifically, the pseudo high band sub-band power _est (ib, id _selected (J-1), J-1) and the pseudo high band sub-band power power _{est of} each sub-band whose index is sb + 1 to eb The absolute value of the maximum value of the difference of (ib, id, J) multiplied by the weight W _power (power (ib, J)) is the estimated residual maximum value ResP _max W _power (id, J) It is assumed.

Next, the pseudo high frequency subband power difference calculation circuit 36 calculates an estimated residual mean value _{ResP ave W power (id, J} ). Specifically, for each subband whose index is sb + 1 to eb, the pseudo high band sub-band power power _est (ib, id _selected (J-1), J-1) and the pseudo high band sub-band power The difference of power _est (ib, id, J) is obtained and multiplied by the weight W _power (power (ib, J)). Then, the absolute value of the value obtained by dividing the sum of the differences multiplied by the weight W _power (power (ib, J) by the number of subbands on the high frequency side (eb−sb) is the estimated residual average The value ResP _ave W _power (id, J) is used.

Further, the pseudo high band sub-band power difference calculation circuit 36 calculates the estimated residual maximum value ResP _max W _power (id, J) multiplied by the estimated residual square mean value ResP _std W _power (id, J) and the weight W _max. And the estimated residual average value ResP _ave W _power (id, J) multiplied by the weight W _ave is obtained as an evaluation value ResPW _power (id, J).

In step S408, the pseudo high band sub-band power difference calculation circuit 36 calculates an evaluation value ResPW _power (id, J) obtained by multiplying the evaluation value ResW _power (id, J) by the weight W _p (J) in equation (25). And the final evaluation value Res _all W _power (id, J) is calculated. The evaluation value Res _all W _power (id, J) is calculated for each of the K decoded high frequency sub-band power estimation coefficients.

Then, after that, the processing of step S409 to step S411 is performed and the encoding processing ends, but these processing is the same as the processing of step S339 to step S341 of FIG. In step S409, the one with the smallest evaluation value Res _all W _power (id, J) is selected from the K coefficient indexes.

  As described above, by weighting each sub-band so that emphasis is placed on the sub-bands with high power, it is possible to obtain higher quality sound at the decoding device 40 side.

Although it has been described above that the selection of the decoded high band sub-band power estimation coefficient is performed based on the evaluation value Res _all W _power (id, J), the decoded high band sub-band power estimation coefficient is It may be selected based on the value ResW _power (id, J).

<6. Sixth embodiment>
[Configuration of coefficient learning apparatus]
By the way, in the decoding device 40 of FIG. 20, a set of the coefficient A _ib (kb) as the decoded high band sub-band power estimation coefficient and the coefficient B _ib is recorded in association with the coefficient index. For example, when decoded high band sub-band power estimation coefficients of 128 coefficient indexes are recorded in the decoding device 40, a large area is required as a recording area such as a memory for recording those decoded high band sub-band power estimation coefficients It becomes.

  Therefore, a part of several decoded high frequency sub-band power estimation coefficients may be used as a common coefficient to make the recording area necessary for recording the decoded high frequency sub-band power estimation coefficients smaller. In such a case, a coefficient learning apparatus for obtaining the decoded high band sub-band power estimation coefficient by learning is configured, for example, as shown in FIG.

  The coefficient learning device 81 includes a subband dividing circuit 91, a high frequency subband power calculating circuit 92, a feature amount calculating circuit 93, and a coefficient estimating circuit 94.

  A plurality of music data or the like used for learning is supplied to the coefficient learning device 81 as a wide-band teacher signal. The wide band teacher signal is a signal including a plurality of high frequency sub-band components and a low frequency sub-band component.

  The subband dividing circuit 91 is formed of a band pass filter or the like, divides the supplied wide band teacher signal into a plurality of subband signals, and supplies the divided signals to the high band subband power calculating circuit 92 and the feature quantity calculating circuit 93. Specifically, the high band sub-band signal of each high band sub-band whose index is sb + 1 to eb is supplied to the high band sub-band power calculation circuit 92, and the low band side of which the index is sb-3 to sb The low band sub-band signal of each sub-band is supplied to the feature value calculation circuit 93.

  The high frequency sub-band power calculation circuit 92 calculates the high frequency sub-band power of each high frequency sub-band signal supplied from the sub-band division circuit 91, and supplies the high frequency sub-band power to the coefficient estimation circuit 94. The feature quantity calculating circuit 93 calculates low band sub-band power as a feature quantity on the basis of each low band sub-band signal supplied from the sub-band dividing circuit 91, and supplies it to the coefficient estimating circuit 94.

  The coefficient estimation circuit 94 performs regression analysis using the high band sub-band power from the high band sub-band power calculation circuit 92 and the feature quantity from the feature quantity calculation circuit 93 to obtain a decoded high band sub-band power estimation factor Are generated and output to the decoding device 40.

[Description of coefficient learning process]
Next, coefficient learning processing performed by the coefficient learning device 81 will be described with reference to the flowchart in FIG.

  In step S431, the sub-band division circuit 91 divides each of the plurality of supplied wide band teacher signals into a plurality of sub-band signals. Then, the sub-band dividing circuit 91 supplies the high-frequency sub-band signal of the sub-band whose index is sb + 1 to eb to the high-frequency sub-band power calculating circuit 92, and the low sub-band whose index is sb-3 to sb. The region subband signal is supplied to the feature value calculation circuit 93.

  In step S432, the high frequency sub-band power calculation circuit 92 performs the same calculation as the above equation (1) for each high frequency sub-band signal supplied from the sub-band division circuit 91 to obtain high frequency sub-band power. It is calculated and supplied to the coefficient estimation circuit 94.

  In step S 433, the feature amount calculation circuit 93 calculates the low band sub-band power as a feature amount by performing the calculation of the equation (1) described above for each low band sub band signal supplied from the sub band dividing circuit 91. , And the coefficient estimation circuit 94.

  Thereby, the high frequency sub-band power and the low frequency sub-band power are supplied to the coefficient estimation circuit 94 for each frame of a plurality of wide band teacher signals.

In step S 434, the coefficient estimation circuit 94 performs regression analysis using the least squares method, and the coefficient A is calculated for each high frequency side subband ib (where sb + 1 ≦ ib ≦ eb) whose index is sb + 1 to eb. Calculate _ib (kb) and the coefficient B _ib .

  In the regression analysis, the low band sub-band power supplied from the feature value calculation circuit 93 is used as an explanatory variable, and the high band sub-band power supplied from the high band sub-band power calculation circuit 92 is used as an explained variable. . Further, the regression analysis is performed using the low band sub-band power and the high band sub-band power of all the frames constituting all the wide band teacher signals supplied to the coefficient learning device 81.

In step S 435, the coefficient estimation circuit 94 obtains a residual vector of each frame of the broadband teacher signal, using the obtained coefficient A _ib (kb) and coefficient B _ib for each subband ib.

For example, the coefficient estimation circuit 94 is configured such that for each subband ib of the frame J (where sb + 1 ≦ ib ≦ eb), the low frequency subband power power (ib, J) is multiplied by the coefficient A _ib (kb). The sum of the sum of the region subband power power (kb, J) (where sb-3 kb kb s sb) and the coefficient _Bib is subtracted to obtain a residual. Then, a vector composed of the residual of each subband ib of the frame J is taken as the residual vector.

  The residual vector is calculated for all the frames that constitute all the broadband teacher signals supplied to the coefficient learning device 81.

  In step S436, the coefficient estimation circuit 94 normalizes the residual vector obtained for each frame. For example, the coefficient estimation circuit 94 obtains, for each subband ib, the variance of the residuals of the subband ib of the residual vector of the entire frame, and the square root of the variance produces the remainder of the subband ib in each residual vector. The residual vector is normalized by dividing the difference.

  In step S437, the coefficient estimation circuit 94 clusters the residual vectors of all the normalized frames by the k-means method or the like.

For example, _let us call the average frequency envelope of all the frames obtained when estimation of the high band sub-band power is performed using the coefficient A _ib (kb) and the coefficient B _i b as the average frequency envelope SA . Further, a predetermined frequency envelope whose power is greater than the average frequency envelope SA is taken as a frequency envelope SH, and a predetermined frequency envelope whose power is less than the average frequency envelope SA is taken as a frequency envelope SL.

  At this time, residual vectors of coefficients such that the average frequency envelope SA, the frequency envelope SH, and the residual vector of the coefficients for which the frequency envelope close to the frequency envelope SL is obtained belong to the cluster CA, the cluster CH, and the cluster CL. Clustering is performed. In other words, clustering is performed such that the residual vector of each frame belongs to either cluster CA, cluster CH, or cluster CL.

In frequency band expansion processing that estimates the high frequency component based on the correlation between the low frequency component and the high frequency component, the residual vector is obtained by using the coefficient A _ib (kb) and the coefficient B _ib obtained by regression analysis in terms of its characteristics Is calculated, the higher the frequency band is, the larger the residual becomes. Therefore, if the residual vector is clustered as it is, processing is performed with emphasis placed on higher subbands.

  On the other hand, the coefficient learning device 81 normalizes the residual vector with the dispersion value of the residual of each subband to make the dispersion of the residual of each subband apparent and equal to each subband. Clustering can be performed with equal weighting.

  In step S438, the coefficient estimation circuit 94 selects one of the cluster CA, the cluster CH, and the cluster CL as a processing target cluster.

In step S439, the coefficient estimation circuit 94 uses the frame of the residual vector belonging to the cluster selected as the cluster to be processed, and the coefficient A _ib of each subband ib (where sb + 1 ≦ ib ≦ eb) by regression analysis. Calculate kb) and the coefficient B _ib .

That is, assuming that a frame of a residual vector belonging to a cluster to be processed is called a frame to be processed, the low band sub-band power and the high band sub-band power of all processing target frames are an explanatory variable and an explanatory variable Then, regression analysis using the least squares method is performed. Thereby, the coefficient A _ib (kb) and the coefficient B _ib are obtained for each subband ib.

In step S440, the coefficient estimation circuit 94 obtains a residual vector for all processing target frames using the coefficients A _ib (kb) and the coefficients B _ib obtained by the process of step S439. In step S440, processing similar to that in step S435 is performed to obtain a residual vector of each processing target frame.

  In step S441, the coefficient estimation circuit 94 normalizes the residual vector of each process target frame obtained in the process of step S440 by performing the same process as in step S436. That is, for each subband, the residual is divided by the square root of the variance to normalize the residual vector.

  In step S 442, the coefficient estimation circuit 94 clusters the residual vectors of all the normalized processing target frames by the k-means method or the like. The number of clusters here is determined as follows. For example, in the case of generating the decoded high band sub-band power estimation coefficient of 128 coefficient indexes in the coefficient learning device 81, the number of processing target frames is multiplied by 128 and divided by the total number of frames. The number of clusters is the number of clusters. Here, the total number of frames is the total number of all frames of all the broadband teacher signals supplied to the coefficient learning device 81.

  In step S443, the coefficient estimation circuit 94 obtains the barycentric vector of each cluster obtained in the process of step S442.

  For example, the clusters obtained by the clustering in step S 442 correspond to the coefficient index, and the coefficient learning device 81 assigns a coefficient index to each cluster, and the decoded high band sub-band power estimation coefficient of each coefficient index is Desired.

Specifically, it is assumed that cluster CA is selected as a processing target cluster in step S438, and F clusters are obtained by clustering in step S442. Now, focusing on one cluster CF out of F clusters, the decoded high band sub-band power estimation coefficient of the coefficient index of cluster CF is linear in the coefficient A _ib (kb) obtained for cluster CA in step S439. The coefficient A _ib (kb) is a correlation term. Also, the sum of the vector obtained by applying the inverse process (inverse normalization) of the normalization performed in step S441 to the centroid vector of the cluster CF determined in step S443 and the coefficient B _ib determined in step S439 are The coefficient B _ib is a constant term of the decoded high band sub-band power estimation coefficient. The inverse normalization mentioned here is, for example, for each element of the barycentric vector of the cluster CF when the normalization performed in step S 441 is to divide the residual by the square root of the variance for each subband. The processing is to multiply the same value as in normalization (the square root of the variance value for each subband).

In other words, the coefficient A _{ib (kb)} obtained in step S439, sets the coefficient B _ib obtained as described above, the decoded high frequency sub-band power estimation coefficients of the coefficient index cluster CF. Therefore, each of the F clusters obtained by clustering has a coefficient A _{i b} (kb) obtained for the cluster CA in common as a linear correlation term of the decoded high band sub-band power estimation coefficient.

  In step S444, the coefficient learning device 81 determines whether all clusters of the cluster CA, the cluster CH, and the cluster CL have been processed as a processing target cluster. If it is determined in step S444 that all clusters have not been processed yet, the process returns to step S438, and the above-described process is repeated. That is, the next cluster is selected as a processing target, and the decoded high band sub-band power estimation coefficient is calculated.

On the other hand, if it is determined in step S444 that all clusters have been processed, a predetermined number of decoded high band sub-band power estimation coefficients to be obtained have been obtained, so the process proceeds to step S445.

  In step S445, the coefficient estimation circuit 94 outputs the obtained coefficient index and the decoded high band sub-band power estimation coefficient to the decoding device 40 for recording, and the coefficient learning process is finished.

For example, among the decoded high band sub-band power estimation coefficients output to the decoding device 40, there are some that have the same coefficient A _ib (kb) as a linear correlation term. Therefore, the coefficient learning device 81 associates a linear correlation term index (pointer) which is information for specifying the coefficient A _ib (kb) with the common coefficient A _ib (kb), and Thus, the linear correlation term index is associated with the coefficient B _ib which is a constant term.

Then, the coefficient learning device 81 decodes the associated linear correlation term index (pointer) and coefficient A _ib (kb), and the associated coefficient index, linear correlation term index (pointer) and coefficient B _ib 40 to be recorded in the memory in the high band decoding circuit 45 of the decoding device 40. Thus, when recording a plurality of decoded high band sub-band power estimation coefficients, the linear correlation term index for the common linear correlation term in the recording area for each decoded high band sub-band power estimation coefficient By storing the pointer, the recording area can be made much smaller.

In this case, since the linear correlation term index and the coefficient A _ib (kb) are associated with each other and recorded in the memory in the high band decoding circuit 45, the linear correlation term index and the coefficient B _ib are obtained from the coefficient index. Then, the coefficient A _ib (kb) can be obtained from the linear correlation term index.

  In addition, as a result of analysis by the present applicant, even if the linear correlation terms of a plurality of decoded high band sub-band power estimation coefficients are made common to about three patterns, there is almost no deterioration of the sound quality on the audibility of the speech subjected to frequency band expansion processing. I know that. Therefore, according to the coefficient learning device 81, it is possible to further reduce the recording area necessary for recording the decoded high band sub-band power estimation coefficient without deteriorating the sound quality of the voice after the frequency band expansion processing.

  As described above, the coefficient learning device 81 generates and outputs the decoded high frequency sub-band power estimation coefficient of each coefficient index from the supplied wide band teacher signal.

  In the coefficient learning process of FIG. 29, although it has been described that the residual vector is normalized, normalization of the residual vector may not be performed in one or both of step S436 and step S441.

  In addition, normalization of the residual vector may be performed, and sharing of the linear correlation term of the decoded high band sub-band power estimation coefficient may not be performed. In such a case, after the normalization process in step S436, the normalized residual vector is clustered into clusters in the same number as the number of decoded high band sub-band power estimation coefficients to be obtained. Then, frames of residual vectors belonging to each cluster are used, regression analysis is performed for each cluster, and decoded high band sub-band power estimation coefficients of each cluster are generated.

  The series of processes described above can be performed by hardware or software. When a series of processes are executed by software, various functions are executed by installing a computer in which a program configuring the software is incorporated in dedicated hardware or various programs. Can be installed, for example, on a general-purpose personal computer from the program storage medium.

  FIG. 30 is a block diagram showing an example of a hardware configuration of a computer that executes the series of processes described above according to a program.

  In the computer, a CPU 101, a read only memory (ROM) 102, and a random access memory (RAM) 103 are mutually connected by a bus 104.

  Further, an input / output interface 105 is connected to the bus 104. The input / output interface 105 includes an input unit 106 such as a keyboard, a mouse and a microphone, an output unit 107 such as a display and a speaker, a storage unit 108 such as a hard disk and a non-volatile memory, and a communication unit 109 such as a network interface. A drive 110 for driving a removable medium 111 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is connected.

  In the computer configured as described above, for example, the CPU 101 loads the program stored in the storage unit 108 into the RAM 103 via the input / output interface 105 and the bus 104 and executes the program. Processing is performed.

The program executed by the computer (CPU 101) is, for example, a magnetic disk (including a flexible disk), an optical disk (CD-ROM (Compact Disc-Read Only Memory), DVD (Digital Versatile Disc), etc.), a magneto-optical disk, or a semiconductor It is recorded on a removable medium 111 which is a package medium including a memory or the like, or is provided via a wired or wireless transmission medium such as a local area network, the Internet, or digital satellite broadcasting.

  The program can be installed in the storage unit 108 via the input / output interface 105 by mounting the removable media 111 in the drive 110. The program can be received by the communication unit 109 via a wired or wireless transmission medium and installed in the storage unit 108. In addition, the program can be installed in advance in the ROM 102 or the storage unit 108.

  Note that the program executed by the computer may be a program that performs processing in chronological order according to the order described in this specification, in parallel, or when necessary, such as when a call is made. It may be a program to be processed.

  The embodiment of the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the present invention.

  Reference Signs List 10 frequency band expanding apparatus, 11 low pass filter, 12 delay circuit, 13, 13-1 to 13-N band pass filter, 14 feature amount calculating circuit, 15 high band sub-band power estimating circuit, 16 high band signal generating circuit , 17 high pass filters, 18 signal adders, 20 coefficient learning devices, 21 21-1 to 21-(K + N) band pass filters, 22 high band sub-band power calculation circuits, 23 feature quantity calculation circuits, 24 coefficients estimation Circuit, 30 encoder, 31 low pass filter, 32 low pass encoding circuit, 33 sub-band division circuit, 34 feature quantity calculation circuit, 35 pseudo high band sub-band power calculation circuit, 36 pseudo high band sub-band power difference Calculation circuit, 37 high-band coding circuit, 38 multiplexing circuit, 40 decoding device, 41 non-multiple , Low-pass decoding circuit, 43 sub-band division circuit, 44 feature amount calculation circuit, 45 high-pass decoding circuit, 46 decoded high-pass sub-band power calculation circuit, 47 decoded high-pass signal generation circuit, 48 synthesis circuit, 50 Coefficient learning device, 51 low pass filter, 52 sub-band division circuit, 53 feature quantity calculation circuit, 54 pseudo high band sub-band power calculation circuit, 55 pseudo high band sub-band power difference calculation circuit, 56 pseudo high band sub-band power Differential clustering circuit, 57 coefficient estimation circuit, 101 CPU, 102 ROM, 103 RAM, 104 bus, 105 input / output interface, 106 input unit, 107 output unit, 108 storage unit, 109 communication unit, 110 drives, 111 removable media

Claims (3)

Demultiplexing means for demultiplexing the input coded data into lower band coded data and an index indicating an estimated coefficient used to generate a high band signal;
Subband dividing means for dividing a low band signal obtained by decoding the low band encoded data into a plurality of low band sub-bands and generating a low band sub-band signal for each low band sub-band;
Feature amount calculating means for calculating low band sub-band power using the low band sub-band signal for each of the low band sub-bands;
The high band signal is obtained by multiplying each of the low band sub-band powers by each of the estimated coefficients indicated by the index, and calculating a sum of the low band sub-band powers of the respective estimated coefficients multiplied. High band sub-band power calculating means for calculating high band sub-band power of the high band sub-band signal of
High band signal generation means for generating the high band sub-band signal using the high band sub-band power and the low band sub-band signal of the mapping source;
A decoding device comprising: combining means for combining each of the low band sub-band signals and each of the high band sub-band signals generated by the high band signal generating means. A demultiplexing step of demultiplexing input coded data into low band coded data and an index indicating an estimated coefficient used to generate a high band signal;
A subband dividing step of dividing a low band signal obtained by decoding the low band encoded data into a plurality of low band sub-bands and generating a low band sub-band signal for each low band sub-band;
A feature amount calculating step of calculating low band sub-band power using the low band sub-band signal for each of the low band sub-bands;
The high band signal is obtained by multiplying each of the low band sub-band powers by each of the estimated coefficients indicated by the index, and calculating a sum of the low band sub-band powers of the respective estimated coefficients multiplied. High band sub-band power calculating step of calculating high band sub-band power of the high band sub-band signal of
A high band signal generation step of generating the high band sub-band signal using the high band sub-band power and the low band sub-band signal of the mapping source;
And a combining step of combining and outputting the low band sub-band signals and the high band sub-band signals generated by the high band signal generating step. A demultiplexing step of demultiplexing input coded data into low band coded data and an index indicating an estimated coefficient used to generate a high band signal;
A subband dividing step of dividing a low band signal obtained by decoding the low band encoded data into a plurality of low band sub-bands and generating a low band sub-band signal for each low band sub-band;
A feature amount calculating step of calculating low band sub-band power using the low band sub-band signal for each of the low band sub-bands;
The high band signal is obtained by multiplying each of the low band sub-band powers by each of the estimated coefficients indicated by the index, and calculating a sum of the low band sub-band powers of the respective estimated coefficients multiplied. High band sub-band power calculating step of calculating high band sub-band power of the high band sub-band signal of
A high band signal generation step of generating the high band sub-band signal using the high band sub-band power and the low band sub-band signal of the mapping source;
A program that causes a computer to execute a process including a combining step of combining each of the low band sub-band signals and each of the high band sub-band signals generated by the process of the high band signal generating step.

Images