KR20090100664A - Apparatus and method for encoding/decoding using bandwidth extension in portable terminal - Google Patents

Abstract

PURPOSE: An encoding device using a bandwidth extension technique of a portable terminal and a method thereof are provided to improving the encoding performance by increasing the amount of information between an upper band signal and a lower band signal in a signal to be encoded. CONSTITUTION: An encoding device using a bandwidth extension technique comprises a lower band encoder(311), a filter calculating unit(305), an upper band mutual information amount filter(303) and a lower band mutual information amount filter(309). The filter calculating unit calculates a filter coefficient for increasing the mutual information amount of an upper band signal of an input signal and an encoded lower band signal, and then increases the mutual information amount of the upper band signal by using the calculated filter coefficient.

Description

Coding Apparatus and Method Using Bandwidth Expansion Technique of Portable Terminal {APPARATUS AND METHOD FOR ENCODING / DECODING USING BANDWIDTH EXTENSION IN PORTABLE TERMINAL}

The present invention relates to an apparatus and method for encoding and decoding a portable terminal, and more particularly, to an apparatus and method for adjusting a correlation (amount of mutual information) influencing coding efficiency in an encoding apparatus of a portable terminal using a band extension technique. It is about. In other words, the present invention relates to an apparatus and a method for processing to increase the amount of information between the upper band signal and the lower band signal in order to increase the coding efficiency of the portable terminal.

With the recent development of digital signal processing technology, audio signals are mostly stored and reproduced as digital data. Digital audio storage / playback equipment samples and quantizes analog audio signals, converts them into digital coded pulse code modulation (PCM) audio data, and stores them on information storage media such as CDs and DVDs. Allows you to listen.

The digital method uses an artificial bandwidth extension (BWE) to estimate and recover an upper band signal from a lower band signal or a feature vector extracted from the signal at the receiver to reproduce only the lower band signal. The sound quality can be greatly improved.

For example, when the sampling frequency (Fs) of the input signal is 16 kHz, the bandwidth extension method restores an upper band signal of 4 k to 8 kHz from a lower band signal of 0 to 4 kHz and passes only a signal of 4 kHz or less. Finally, the receiving side of the signal outputs the same 16 kHz signal as the original input signal, which is closely related to the correlation between the frequency bands (upper band and lower band) of the audio or audio signal.

 That is, when the audio signal and the voice signal of one frame are divided into the lower band and the upper band for each frequency band, since the signals of the two bands are generated from the same human voice structure, there is a close correlation between the two bands. If is large or if the mutual information is large, the upper band signal restored by the bandwidth extension method will have good sound quality close to the original sound.

However, there is a problem in that the upper band signal cannot be completely restored by the bandwidth extension method when there is no information on the upper band signal because the amount of mutual information between the two bands is not large.

1 is a block diagram illustrating a configuration of an apparatus for performing a bandwidth extension method in a general portable terminal.

Referring to FIG. 1, the portable terminal may include an encoder 100 and a decoder 110. Here, the encoder 100 may include a lower band encoder 101 and an upper band signal estimator 103. (BWE) and a quantizer 105, and the decoder 103 may include an inverse quantizer 112, a lower band decoder 114, and a higher band signal estimator 116. .

The encoder 100 is a BWE-based encoder that improves coding efficiency by using a BWE system in encoding an upper band signal. The lower band encoder 101 processes to receive and encode a lower band signal, and the upper band signal estimator 103. ) Estimates an upper band signal by using the signal encoded by the lower band encoder 101.

Accordingly, the encoder 100 outputs a residual upper band signal from which a signal overlapping with a lower band is removed by subtracting the upper band signal estimated by the upper band signal estimator 103 from the input upper band signal. do. Here, the residual upper band signal is quantized by quantizer 105 and then transmitted over a communication channel. In this case, the quantizer 105 uses a scalar or vector quantizer depending on the application purpose.

The decoder 110 processes the lower band decoder 114 to decode the encoded lower band signal to reproduce the lower band signal, and the higher band signal estimator 116 uses the decoded lower band signal. To estimate the higher-band signal. Further, the inverse quantizer 112 processes to inversely quantize the received residual upper band signal and output the decoded residual upper band signal.

Accordingly, the decoder 110 processes the sum of the decoded residual upper band signal and the upper band signal estimated by the upper band signal estimator 116 to output the decoded upper band signal.

In the above method, the coding efficiency of the upper band signal can be increased by removing the upper band signal overlapping with the lower band by using the upper band signal estimator in the receiving side as well as the transmitting side, and encoding the remaining upper band part. There is a problem that a performance change occurs depending on the size of correlation or mutual information. In other words, in the above method, if the correlation is large, the coding efficiency may be high, but when the correlation is small, the coding efficiency is decreased.

Therefore, there is a need for an apparatus and method for solving the above problems.

The present invention was derived to solve the above problems, and an object of the present invention is to provide an apparatus and method for improving the performance of the encoding apparatus using a band extension technique of a portable terminal.

Another object of the present invention is to provide an apparatus and method for controlling mutual information amount between band signals in an encoding apparatus using a band extension technique of a portable terminal.

According to a first aspect of the present invention for achieving the above objects, an encoding apparatus using a band extension technique includes a lower band encoder that encodes a lower band signal of an input signal, an upper band signal of the input signal, and the encoded lower band A filter calculation unit for calculating a filter coefficient for increasing the mutual information amount of a band signal, an upper band mutual information amount filter for processing to increase the mutual information amount of the upper band signal by using the filter coefficient calculated by the filter coefficient calculation unit; And a lower band mutual information amount filter for processing to increase the mutual information amount of the lower band signal by using the filter coefficient calculated by the filter coefficient calculating unit.

According to a second aspect of the present invention for achieving the above objects, a decoding apparatus using a band extension technique includes an inverse quantizer for receiving and decoding an encoded residual upper band signal output in the encoding process, and a filter using a decoded signal. And a filter coefficient calculator for calculating coefficients, and an inverse filter for inverse-quantity information amount inversely filtering the decoded upper band signal by using the calculated filter coefficients and outputting the original signal as an original signal.

According to a third aspect of the present invention for achieving the above object, the encoding method using the band extension technique is to classify the input signal into an upper band signal and a lower band signal, and then to encode the classified lower band signal; Calculating a filter coefficient for increasing the mutual information amount of the classified upper band signal and the encoded lower band signal, and using the calculated filter coefficient to increase the mutual information amount of the upper band signal and the lower band signal. It characterized in that it comprises a process of converting.

According to a fourth aspect of the present invention for achieving the above objects, a decoding method using a band extension technique includes the steps of receiving an encoded residual upper band signal output during an encoding process, and receiving the encoded residual upper band signal. Decoding the signal; calculating a filter coefficient using a decoded signal; and reverse filtering the decoded upper band signal using the calculated filter coefficient and outputting the original signal as an original signal. .

As described above, the present invention increases the mutual information amount of the upper band signal and the lower band signal of a signal to be encoded in a portable terminal for encoding a voice and audio signal by using an artificial bandwidth extension (BWE), It is possible to obtain an improved encoding performance than the conventional encoding apparatus using the bandwidth extension method.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In describing the present invention, when it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

In the following description, an apparatus and method for adjusting a correlation (amount of mutual information) influencing coding efficiency in an encoding apparatus of a portable terminal using a band extension technique will be described.

2 is a block diagram illustrating an encoder of a portable terminal according to an exemplary embodiment of the present invention.

Referring to FIG. 2, the encoder includes an upper band mutual information amount filter 201, a quantizer 203, a filter coefficient calculator 205, an upper band signal estimator 207, a lower band mutual information amount filter 209, A lower band encoder 211, a lower band decoder 213, and an upper band mutual information amount inverse filter 215 may be included.

The lower band encoder 211 of the encoder encodes and processes a lower band signal to be transmitted through a communication channel, and causes the higher band signal estimator 207 to estimate the upper band signal using the encoded lower band signal. Do it.

The lower band mutual information amount filter 209 processes to increase the mutual information amount of the encoded lower band signal using the filter coefficients provided from the filter coefficient calculating unit 205.

The upper band mutual information amount filter 201 processes to convert the received upper band signal using the filter coefficient provided from the filter coefficient calculating unit 205. That is, the upper band mutual information amount filter processes to convert an input upper band signal (upper band signal 1) into an output upper band signal (upper band signal 2) having increased mutual information amount.

The filter coefficient calculation unit 205 is a lower order necessary to increase the mutual information amount of the two input signals by using the decoded upper band signal provided from the upper band mutual information amount inverse filter and the lower band signal decoded by the lower band decoder. Band and upper band filter coefficients are calculated and processed to provide these coefficients to each corresponding filter.

Here, the decoded upper band signal provided from the upper band cross information amount inverse filter 215 is a signal that is fed back after the encoded upper band signal of the previous frame is decoded, and the decoded lower band signal is the encoding of the current frame. The lower band signal is a decoded signal.

The encoder is a residual upper band signal (residual) obtained by subtracting an upper band signal (upper band signal 2) converted by the upper band mutual information amount filter 201 and an upper band signal estimated by the upper band signal estimator 207. After processing the high band signal 2)), the quantizer 203 processes to quantize the signal.

3 is a block diagram illustrating a decoder of a portable terminal according to an exemplary embodiment of the present invention.

Referring to FIG. 3, the decoder includes an inverse quantizer 301, an upper band mutual information amount inverse filter 303, a filter calculator 305, an upper band signal estimator 307, and a lower band mutual information amount filter 309. ) And a lower band decoder 311.

The lower band decoder 311 of the decoder decodes the encoded lower band signal, and processes the higher band signal estimator 307 to estimate the upper band signal using the decoded lower band signal.

The inverse quantizer 301 receives an encoded residual upper band signal (coded residual upper band signal 2) and inversely quantizes it to process a decoded residual upper band signal (decoded residual upper band signal 2).

The filter coefficient calculating unit 305 calculates a lower band filter coefficient and an upper band inverse filter coefficient by using the decoded upper band signal and the decoded lower band signal, and provides the coefficients to respective corresponding filters. do. Here, the lower band filter coefficient calculated by the filter calculation unit 305 is the same as the transmitting side lower band filter coefficient for the same frame, and the upper band inverse filter coefficient is the same as the upper band inverse filter coefficient at the transmitting side. Therefore, it has an inverse relationship with the upper band filter coefficient of the transmitting side.

The lower band cross information amount filter 309 processes the cross information amount of the decoded lower band signal to be increased by using the coefficient provided from the filter coefficient calculating unit 305, and then the upper band signal estimating unit 307 To estimate the higher-band signal.

Accordingly, the decoder adds the residual upper band signal decoded by the inverse quantizer 301 and the upper band signal estimated by the upper band signal estimator 307 to decode the upper band signal (decoded upper band signal). 2) is output to the upper band mutual information amount inverse filter.

The upper band mutual information inverse filter 303 performs an inverse filter process on the decoded upper band signal (decoded upper band signal 2) by using an inverse filter coefficient provided from the filter coefficient calculator 305. Process to reproduce the band signal.

4 is a block diagram illustrating a configuration of a filter coefficient calculator applied to an encoder according to an exemplary embodiment of the present invention.

Referring to FIG. 4, the filter coefficient calculator 400 applied to the encoder includes an upper band mutual information amount filter coefficient calculator 401 and a lower band mutual information amount filter coefficient calculator 403.

Here, the upper band mutual information amount filter coefficient calculator 401 may be calculated using a decoded upper band signal and a decoded lower band signal as a block for calculating a coefficient of the upper band mutual information amount filter.

In addition, the lower band mutual information amount filter coefficient calculator 403 may calculate the coefficients of the lower band mutual information amount filter using a decoded upper band signal and a decoded lower band signal.

Here, the filter coefficient calculator 400 of the encoder must satisfy the following conditions when calculating the filter coefficient for increasing the mutual information amount of the higher band signal.

First, the lower-band signal X, the higher-band signal Y, the higher-band filter mutual information H [], the higher-band inverse filter mutual information for the higher-band signal converted by the H ^-1 [], H [] to Y2 Assuming

First, H [] must be reversible and H- ¹ [] must be present to establish Y = H ^-1 [Y2] = H ^-1 [H [Y]]. In other words, the original signal Y must be reproducible from the converted signal Y2 .

Second, mutual information amount I [ X; Y2 ]> I [ X; Y ] must hold.

Third, in the statistical sense, the dynamic range of Y2 should not be greater than at least the dynamic range of Y.

Description of the conditions to be satisfied when calculating the filter coefficients as described above will be described in detail with reference to FIG.

5 is a block diagram illustrating a configuration of a filter coefficient calculator applied to a decoder according to an exemplary embodiment of the present invention.

Referring to FIG. 5, the filter coefficient calculator 500 applied to the decoder includes an upper band mutual information amount inverse filter coefficient calculator 501 and a lower band mutual information amount filter coefficient calculator 513.

Here, the upper band cross information amount inverse filter coefficient calculator 501 may calculate a coefficient of the upper band cross information amount inverse filter using a decoded upper band signal and a decoded lower band signal.

Also, the lower band mutual information amount filter coefficient calculator 513 may calculate the coefficients of the lower band mutual information amount filter using a decoded upper band signal and a decoded lower band signal.

Here, the filter coefficient calculation unit 500 of the decoder may filter the filter coefficients for increasing the amount of mutual information of the upper band signal in a state that satisfies the conditions as in the filter coefficient calculation unit 400 of the encoder described with reference to FIG. 4. Calculate

The foregoing has described an apparatus for adjusting a correlation (amount of mutual information) influencing coding efficiency in an encoding apparatus of a portable terminal using a band extension technique. Hereinafter, the apparatus according to an exemplary embodiment of the present invention will be described. It will be described a method for adjusting the correlation (interaction amount) affecting the coding efficiency.

6 is a flowchart illustrating an operation of an encoder according to an exemplary embodiment of the present invention. Here, the encoder extends the bandwidth of the upper band of the input signal (BWE) in order to increase the amount of mutual information between the upper band and the lower band. Accordingly, the coder encodes and outputs the low-band signal by low-band encoding.

Referring to FIG. 6, the encoder first calculates mutual information amount filter coefficients in step 601 and then proceeds to step 603 and converts the received upper band signal (upper band signal 1) using the calculated filter coefficients. Do it. Here, the mutual information amount filter coefficient may be calculated by a filter coefficient unit using coefficients of an upper band mutual information amount filter and coefficients of a lower band mutual information amount filter, and the converted upper band signal (upper band signal 2) is input upper band. The output upper band signal whose mutual information amount is increased compared to the signal.

Here, the filter coefficient calculator must satisfy the following conditions.

First, the lower-band signal X, the higher-band signal Y, the higher-band filter mutual information H [], the higher-band inverse filter mutual information for the higher-band signal converted by the H ^-1 [], H [] to Y2 Assuming

First, H [] must be reversible and H- ¹ [] must be present to establish Y = H ^-1 [Y2] = H ^-1 [H [Y]]. In other words, the original signal Y must be reproducible from the converted signal Y2 .

Second, mutual information amount I [ X; Y2 ]> I [ X; Y ] must hold.

Third, in the statistical sense, the dynamic range of Y2 should not be greater than at least the dynamic range of Y.

The first condition means that the signal transformed by the sender upper band cross information amount filter should be reconstructed by the receiver high band cross information amount inverse filter. The second and third conditions indicate that the conversion result by filter H [] is encoded. It means that it should contribute to the improvement of efficiency.

In order for the first condition, that is, H- ¹ [] to exist, the filter H [] basically represents a monotonic and differentiable function, but the amount of mutual information does not change with respect to this conversion function. That is, I [ X; Y2 ] = I [ X; Y ]. Therefore, this does not satisfy the second condition.

In view of this problem, the present invention uses a term ("reversible") to refer to a function that makes it possible to finally reproduce the original transmitted information Y using other transmitted information, such as a lower band vector X. Let's do it.

For example, Y2 = H [ X, Y ] = X ** Y = {x1y1, ..., xNyN} ′. Here, "*" means a multiplication operation between components, and x1, y1, etc. mean components of X and Y, respectively. The inverse function H ^-1 [], that is, a function that regenerates Y from Y2 again under the condition that X is given can be defined as follows. That is, Y = H ^-1 [ X, Y2 ] = X · Y2 = {x1 / y21, ..., xN / y2N} ′. Here, "· /" denotes a division operation between components, and x1 and y21 mean components of X and Y2, respectively.

In this way, Y2 sent from the transmitting side using the function "*" can be restored to Y again using the function "· /" at the receiving side. In relation to the second condition, in general, two random variables (or vectors) increase in mutual information when they have mutual dependence, that is, a mutual function relationship. That is, if Y2 transformed by the filter H [] has a functional relationship with X , for example, Y2 = f [ X ], the amount of mutual information of the two random vectors Y2 and X increases.

Thereafter, the encoder proceeds to step 605 to process the low-band cross information amount filter to estimate an upper band signal, and then proceeds to step 607 to output a residual upper band signal (residual upper band signal 2).

Here, the residual upper band output signal refers to a signal obtained by subtracting the upper band signal (upper band signal 2) converted in step 603 and the upper band signal estimated in step 605.

In step 609, the encoder processes the residual signal to be quantized, and then transmits it through a communication channel (coded residual upper band signal 2). Here, the quantizer for quantizing the residual signal may use a scalar or vector quantizer depending on the application purpose.

The encoder then terminates this algorithm.

7 is a flowchart illustrating an operation of a decoder according to an exemplary embodiment of the present invention. Here, the decoder processes to decode an input signal having an increased amount of mutual information between the upper band and the lower band. Accordingly, the decoder causes the lower band decoder to decode the encoded lower band signal received through the communication channel to reproduce the lower band signal.

Referring to FIG. 7, the decoder first receives a higher-band residual signal (coded residual upper-band signal 2) converted by the encoder in step 701.

In step 703, the decoder processes the received residual signal to be quantized, and then proceeds to step 705 to decode the higher-band signal. That is, the decoder processes the received encoded residual upper band signal 2 to be output as a decoded residual upper band signal 2.

In operation 707, the decoder calculates a filter coefficient using a decoded signal. Here, the filter coefficients refer to lower band filter coefficients and upper band inverse filter coefficients, and the decoder may calculate the filter coefficients by using the decoded upper band signal and the decoded lower band signal. In addition, the lower band filter coefficients are the same as the transmitting side lower band filter coefficients for the same frame, and the upper band inverse filter coefficients are the same as the upper band inverse filter coefficients of the transmitting side and have an inverse relationship with the upper band filter coefficients of the transmitting side. .

Thereafter, the decoder proceeds to step 709 to decode the original upper band signal.

In the decoding of the original upper band signal, the upper-band signal decoded by adding the residual upper-band signal 2 decoded in step 705 and the upper-band signal estimated by the upper-band signal estimator is decoded. After outputting the upper band signal 2), a process of inverse filtering the decoded upper band signal to decode the original upper band signal.

The decoder then terminates this algorithm.

8 is a flowchart illustrating a process of calculating filter coefficients of a filter calculator according to an exemplary embodiment of the present invention.

Referring to FIG. 8, the filter calculator first checks the decoded upper band signal of the previous frame in step 801 and then proceeds to step 803 to calculate the upper band filter coefficients. That is, the filter calculator calculates a filter for increasing the amount of mutual information of the input upper band signal using the decoded signal upper band signal of the previous frame.

In operation 805, the filter calculator determines the decoded lower band signal, and then proceeds to operation 807 to calculate the lower band filter coefficients. Here, the filter calculator calculates a filter for increasing the amount of mutual information of the input signal by using the decoded lower band signal in which the encoded lower band signal of the current frame is a decoded signal.

Thereafter, the filter calculation unit terminates the present algorithm.

In the above description, a method of using a filter coefficient of an upper band signal and a filter coefficient of an upper band signal as an apparatus and method for increasing the amount of mutual information of an upper band signal and a lower band signal according to the present invention has been described. The mutual information amount of the upper band signal and the lower band signal may be increased by applying only the mutual information amount filter or only the lower band mutual information amount filter.

As described above, the method of applying only the upper band cross information amount filter or only the lower band cross information amount filter is the same as the method of using the upper band cross information amount filter and the lower band cross information amount filter described with reference to FIGS. The amount of mutual information of the upper band signal and the lower band signal can be increased.

9 is a graph showing the performance of a decoder according to an embodiment of the present invention.

The present invention is a method for increasing the mutual information amount of the upper band vector and the lower band vector, as described above, a method of applying an upper band mutual information amount filter and a lower band mutual information amount filter, and an upper band mutual information amount filter and a lower band mutual information amount filter. Only one of them can be applied.

9 illustrates a method of applying only an upper band mutual information amount filter and a case of applying only a lower band mutual information amount filter according to an embodiment of the present invention.

First, in order to compare the performance of a coding apparatus of a general portable terminal and a coding apparatus of a portable terminal according to the present invention, a low-band signal of a PCM speech signal sampled at 16 kHz is combined with a total of 14-order Mel Frequency Cepstral Coefficient (MFCC) feature vectors. Log-scaled energy, that is, X (n) = {x1 (n), ..., x14 (n), lnELB (n)} ', and the corresponding higher-band signal is the fourth-order MFCC coefficients and the logarithm. Scaled energy, that is, Y (n) = {y1 (n), ..., y4 (n), lnEHB (n)} '. Here, n indicates a frame number and the frame size is 20 ms. In this case, the encoding problem is to efficiently encode the fourth-order higher-band MFCC coefficients and energy information Y (n).

First, when the encoding apparatus encodes only the E _HB of the higher band signal among the Y (n) information, the upper band signal is Y (n). ) = {lnE _HB (n)}. In this case, an example of an upper band mutual information amount filter for converting the original upper band signal Y (n) into Y2 (n) may be expressed by Equation 1 below.

Y2 (n) = H [X (n), Y (n-1)] = lnE _HB (n) -lnE _LB (n) = ln (E _HB (n) / E _LB (n))

Here, X is a lower band signal, Y is an upper band signal, H [] is an upper band mutual information amount filter, Y2 is an upper band signal converted by H [], and E _HB Is the energy of the upper band signal, E _LB is the energy of the lower band signal. In addition, Y (n-1) indicates the encoded highband vector of the feedback n-1th frame.

Referring to Equation 1, the upper band mutual information amount filter corresponds to a differential operation on a logarithmic scale and a division operation on a linear scale.

In the high-band cross information amount filter as described above, the first condition (H [] must be reversible among the three conditions presented in the present invention, and since H ⁻¹ [] is present, Y = H ⁻¹ [Y2] = H ^−1). [H [Y]] must be established, i.e., the original signal Y must be reproducible from the converted signal Y2 ). That is, the original upper band signal can be recovered from lnELB (n) which is one component of the upper band signal Y2 and the lower band signal X (n) converted by the upper band mutual information amount filter. This can be expressed as in Equation 2.

lnE _HB (n) = Y2 (n) + lnE _LB (n)

Here, E _HB Is the energy of the upper band signal, E _LB is the energy of the lower band signal, and Y2 is the upper band signal converted by the upper band mutual information amount filter.

Equation 1 is expressed as Equation 3 below, and thus the second condition (interaction amount I [ X; Y2 ]> I [ X; Y ] is established among the three conditions presented in the present invention. Satisfactory).

I [X (n); Y2 (n)] = 1.27> I [X (n); Y (n)] = 0.71

Here, X is a lower band signal, Y is an upper band signal, and Y2 is an upper band signal converted by the upper band mutual information amount filter.

Referring to Equation 3, the mutual information amount of Y2 is increased by about 0.56 bit compared to the mutual information amount of Y. This indicates an improvement in coding efficiency of 0.56 bits per frame and 28 bits per second in the encoding problem of higher band energy Y (n) = {lnE _HB (n)}. Finally, the dispersion of Y was calculated to be about 74.44, and the dispersion of Y2 to about 35.06. Therefore, the above Equation 1 satisfies the third condition (the dynamic range of Y2 should not be larger than the dynamic range of Y at least) in the present invention.

In addition, the mutual information amount between the two vectors shown in Equation 3 may be defined as Equation 4 below.

Here, X is a feature vector of the lower band signal, Y is a feature vector of the upper band signal. In addition, f _X (x) represents a probability density function of X, f _Y (y) represents a probability density function of Y, and f _XY (x, y) represents a combined probability density function of X and Y.

As described above, the filter defined in Equation 1 satisfies all three conditions of the present invention as the upper band mutual information amount filter while encoding efficiency of higher band energy Y (n) = {lnE _HB (n)}. To increase.

<Table 1> is a result of comparing the performance of the general encoding device (BWE) and the performance of the encoding device (eBWE) according to the present invention, and Figure 9 (a) is the amount of high-band mutual information according to an embodiment of the present invention It is a graph comparing the performance of the encoding device to which only a filter is applied and the encoding device of a general portable terminal.

Quantization Bit Count BWE eBWE 0 74.44139682 35.04823669 One 29.9638428 14.41622256 2 10.48636178 4.902777868 3 3.237544251 1.483840889 4 0.879504628 0.412519595 5 0.228972655 0.108508489 6 0.058430905 0.028144785 7 0.014838623 0.007102297 8 0.003612276 0.001717147

The numerical values in Table 1 refer to encoding error energy, and it can be seen that the encoding apparatus according to the present invention is improved over the encoding performance of a general encoding apparatus.

Another embodiment of the apparatus will be described with reference to the operation of the encoding apparatus by applying only the low-band cross information amount filter.

As described above, assuming that the encoding apparatus encodes only the MFCC coefficients of the upper band 4th order of the Y (n) information, the method of increasing encoding efficiency by applying only the lower band mutual information amount filter is performed. n) = {y1 (n), ..., y4 (n)}. In this case, an example of a low-band cross information amount filter for converting the original low-band signal X (n) into X2 (n) may be expressed as Equation 5 below.

X2 (n) = G [ X (n), Y (n-1)] = { X (n): Y (n-1)} ′

= {x1 (n), ..., x14 (n): y1 (n-1), ..., y4 (n-1)} ′

Where X is a low-band signal, Y is a high-band signal, G [] is a low-band cross-quantity filter, X2 is a low-band signal converted by G [], and ":" is an extension operator in matrix and vector, Y (n-1) indicates the encoded highband vector of the feedback n-1th frame.

As shown in Equation 5, the low-band cross information amount filter indicates an augmentation operator that outputs an augmented vector.

The lower band mutual information amount filter as described above satisfies the second condition among the three requirements of the mutual information amount filter presented in the present invention, and the mutual information amount is increased by the extension vector X2 as shown in Equation 6 below.

I [ X2 ; Y ] = 2.369886> I [ X; Y ] = 1.439137 (unit bit)

According to the mutual information amount calculation result as shown in Equation 6, the mutual information amount is increased by about 1 bit by changing from the lower band signal X to X2 . This makes it possible to predict the encoding efficiency improvement of 1 bit per frame and 50 bits per second in the problem of encoding the fourth-order higher-band MFCC coefficient Y. In addition, the first and third conditions of the three requirements of the mutual information amount filter presented in the present invention are related to the upper band mutual information amount filter and are not applicable when only the lower band mutual information amount filter is used.

Table 2 shows a result of comparing the performance of the general encoding device (BWE) and the performance of the encoding device (eBWE) according to the present invention, and FIG. 9 (b) shows the amount of low-band cross information according to an embodiment of the present invention. It is a graph comparing the performance of the encoding device to which only a filter is applied and the encoding device of a general portable terminal.

Quantization Bit Count BWE (CD) eBWE (CD) 2 0.410244 0.3954 3 0.293181 0.260144 4 0.21433 0.176756 5 0.155101 0.122713 6 0.112254 0.0866 7 0.081301 0.061116 8 0.058495 0.043185

The numerical value of Table 2 means a Cepstral Distance value, and it can be seen that the encoding apparatus according to the present invention is improved over the encoding performance of the general encoding apparatus.

Meanwhile, in the detailed description of the present invention, specific embodiments have been described, but various modifications are possible without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the scope of the following claims, but also by the equivalents of the claims.

1 is a block diagram showing a configuration of an apparatus for performing a bandwidth extension method in a general portable terminal;

2 is a block diagram showing an encoder of a portable terminal according to an embodiment of the present invention;

3 is a block diagram illustrating a decoder of a portable terminal according to an embodiment of the present invention;

4 is a block diagram showing a configuration of a filter coefficient calculation unit applied to an encoder according to an embodiment of the present invention;

5 is a block diagram showing a configuration of a filter coefficient calculator applied to a decoder according to an exemplary embodiment of the present invention;

6 is a flowchart illustrating an operation of an encoder according to an embodiment of the present invention;

7 is a flowchart illustrating an operation process of a decoder according to an embodiment of the present invention;

8 is a flowchart illustrating a process of calculating a filter coefficient of a filter calculator according to an exemplary embodiment of the present invention.

9 (a) is a graph comparing performances of an encoding apparatus to which only an upper band mutual information amount filter is applied and an encoding apparatus of a general portable terminal according to an embodiment of the present invention;

9 (b) is a graph comparing the performance of an encoding device to which only the low-band cross-quantity information filter is applied and the encoding device of a general portable terminal according to an exemplary embodiment of the present invention.

Claims (22)

In the encoding apparatus using a band extension technique, A low band encoder for encoding a low band signal of the input signal, A filter calculator for calculating filter coefficients for increasing the amount of mutual information between the upper band signal of the input signal and the encoded lower band signal; An upper band mutual information amount filter which processes to increase the mutual information amount of the upper band signal by using the filter coefficient calculated by the filter coefficient calculating unit; And a lower band mutual information amount filter for processing to increase the mutual information amount of the lower band signal by using the filter coefficient calculated by the filter coefficient calculating unit. The method of claim 1, The encoding apparatus using the band extension technique, An upper band estimator for estimating an upper band signal using the lower band signal having the increased mutual information amount, and processing to output a residual upper band signal obtained by subtracting the estimated upper band signal and the upper band signal having the mutual information amount increased; And a quantizer for quantizing and outputting the residual upper band signal. The method of claim 2, The quantizer is, And a residual high band signal is quantized using at least one of a scalar quantizer and a vector quantizer. The method of claim 1, The filter coefficient calculation unit, From the converted signal Y2 , the original signal Y must be reproducible, the mutual information amount I [ X; Y2 ] must be greater than I [ X; Y ], and in the statistical sense the dynamic range of Y2 is at least Y. And calculating a filter coefficient that satisfies a condition that should not be large compared to the dynamic range. Here, X is a lower band signal, Y is The upper band signal, H [], is an upper band mutual information amount filter, H ^{- 1} [] is an upper band mutual information amount inverse filter, and Y2 is an upper band signal converted by H []. The method of claim 4, wherein The filter coefficient calculation unit, And encoding the filter coefficients by using the decoded upper band signal and the decoded lower band signal. The method of claim 1, The encoding apparatus using the band extension technique, And the mutual information amount is increased by using only one of the upper band mutual information amount filter and the lower band mutual information amount filter. In a decoding apparatus using a band extension technique, An inverse quantizer for receiving and decoding the encoded residual upper band signal output in the encoding process; A filter coefficient calculator for calculating filter coefficients using the decoded signal, And an upper band mutual information amount inverse filter which inversely filters the decoded upper band signal using the calculated filter coefficients and outputs the original signal as an original signal. The method of claim 7, wherein The upper band mutual information amount inverse filter, And decoding the received encoded upper band signal by decoding the higher-band signal obtained by summing the output-signal and the higher-band signal estimated by using the lower-band signal with increased mutual information amount. The method of claim 7, wherein The inverse quantizer, Inverse quantization using at least one of a scalar inverse quantizer and a vector inverse quantizer. The method of claim 7, wherein The filter coefficient calculation unit, From the converted signal Y2 , the original signal Y must be reproducible, the mutual information amount I [ X; Y2 ] must be greater than I [ X; Y ], and in the statistical sense the dynamic range of Y2 is at least Y. And a filter coefficient that satisfies a condition that should not be greater than that of the dynamic range. Here, X is a lower band signal, Y is The upper band signal, H [], is an upper band mutual information amount filter, H ^{- 1} [] is an upper band mutual information amount inverse filter, and Y2 is an upper band signal converted by H []. The method of claim 10, The filter coefficient calculation unit, And decoding the filter coefficients using the decoded upper band signal and the decoded lower band signal. In the coding method using a band extension method, Classifying an input signal into an upper band signal and a lower band signal, and then encoding the classified lower band signal; Calculating filter coefficients for increasing the amount of mutual information between the classified upper band signal and the encoded lower band signal; And converting the amount of mutual information of the upper band signal and the lower band signal into a signal having an increased value using the calculated filter coefficients. The method of claim 12, The encoding method using the band extension technique, Estimating an upper band signal using the lower band signal having the increased amount of mutual information; Outputting a residual upper band signal obtained by subtracting the estimated upper band signal and the upper band signal having increased mutual information amount; And transmitting the output residual upper band signal. The method of claim 12, The output residual upper band signal, A coding method characterized in that quantization is performed using at least one of a scalar quantizer and a vector quantizer. The method of claim 12, The filter coefficient for increasing the mutual information amount is From the converted signal Y2 , the original signal Y must be reproducible, the mutual information amount I [ X; Y2 ] must be greater than I [ X; Y ], and in the statistical sense the dynamic range of Y2 is at least Y. A coding method, characterized in that it should not be larger than the dynamic range. Here, X is a lower band signal, Y is The upper band signal, H [], is an upper band mutual information amount filter, H ^{- 1} [] is an upper band mutual information amount inverse filter, and Y2 is an upper band signal converted by H []. The method of claim 15, The filter coefficient for increasing the mutual information amount is An encoding method comprising calculating using a decoded upper band signal and a decoded lower band signal. The method of claim 12, The encoding method using the band extension technique, And increasing the mutual information amount of at least one of the upper band signal and the lower band signal. In the decoding method using the band extension method, Receiving the encoded residual upper band signal output from the encoding process; Decoding the received encoded residual upper band signal; Calculating a filter coefficient using a decoded signal, And inversely filtering the decoded upper band signal using the calculated filter coefficients and outputting the original signal as an original signal. The method of claim 18, Inversely filtering the decoded upper band signal, And decoding a higher-band signal obtained by decoding the encoded residual upper-band signal and adding the higher-band signal estimated by using the lower-band signal having increased mutual information amount. The method of claim 18, The encoded residual upper band signal is A quantization method using at least one of a scalar inverse quantizer and a vector inverse quantizer. The method of claim 18, The calculated filter coefficient is From the converted signal Y2 , the original signal Y must be reproducible, the mutual information amount I [ X; Y2 ] must be greater than I [ X; Y ], and in the statistical sense the dynamic range of Y2 is at least Y. A decoding method, characterized in that it should not be larger than the dynamic range. Here, X is a lower band signal, Y is The upper band signal, H [], is an upper band mutual information amount filter, H ^{- 1} [] is an upper band mutual information amount inverse filter, and Y2 is an upper band signal converted by H []. The method of claim 21, The calculated filter coefficient is And decoding using the decoded upper band signal and the decoded lower band signal.

Images