KR20080025636A - Method and apparatus for encoding and decoding audio signal using band width extension technique - Google Patents

Abstract

A method and an apparatus for encoding/decoding an audio signal by using bandwidth expansion technique are provided to output an encoded bitplane and encoded bandwidth expansion information as an encoded result of an input signal, thereby efficiently encoding high frequency components at a limited bit rate and accordingly improving sound quality. An audio signal encoding method comprises the following steps of: splitting an input signal into a high band signal and a low band signal(1300); converting each of the high and low band signals from a time domain to a frequency domain(1310); performing quantization and context-dependent bitplane encoding of the converted low band signal(1320); generating and encoding bandwidth expansion information showing a characteristic of the converted high band signal by using the converted low band signal(1330); and outputting the encoded bitplane and the encoded bandwidth expansion information as an encoded result of the input signal(1340).

Description

Method and apparatus for encoding / decoding audio signal using bandwidth extension technique {Method and apparatus for encoding and decoding audio signal using band width extension technique}

The present invention relates to a method and apparatus for encoding / decoding an audio signal, and more particularly, to a method and apparatus for encoding / decoding an audio signal using a bandwidth extension technique.

In encoding or decoding an audio signal, it is required to improve sound quality as much as possible using a limited bit rate. At low bit rates, since the available bits are small, the frequency band of the audio signal to be encoded must be reduced so that the sound quality can be degraded.

In general, low frequency components have a significant effect on human perception compared to high frequency components. Therefore, there is a need for a method capable of improving sound quality while increasing the bits allocated for encoding low frequency components and reducing the bits allocated for encoding high frequency components.

SUMMARY OF THE INVENTION An object of the present invention is to provide a method and apparatus for encoding an audio signal capable of improving sound quality by efficiently encoding high frequency components at a limited bit rate.

Another object of the present invention is to provide a method and apparatus for decoding an audio signal which efficiently decodes high frequency components from a bitstream encoded at a limited bit rate.

According to an aspect of the present invention, there is provided a method of encoding an audio signal, the method including: (a) dividing an input signal into a high frequency band signal and a low frequency band signal; (b) converting the high frequency band signal and the low frequency band signal from the time domain to the frequency domain, respectively; (c) quantizing the converted low frequency band signal and encoding the bitplane based on a context; (d) generating and encoding bandwidth extension information indicating a characteristic of the converted high frequency band signal using the converted low frequency band signal; And (e) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal.

In addition, the object is (a) dividing the input signal into a high frequency band signal and a low frequency band signal; (b) converting the high frequency band signal and the low frequency band signal from the time domain to the frequency domain, respectively; (c) quantizing the transformed low frequency band signal and encoding the bitplane based on a context; (d) generating and encoding bandwidth extension information indicating a characteristic of the converted high frequency band signal using the converted low frequency band signal; And (e) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal. A computer-readable recording medium having recorded thereon a program for executing an encoding method of an audio signal. Is achieved by.

In addition, the method for encoding an audio signal according to the present invention for solving the other problem comprises the steps of: (a) dividing an input signal into a high frequency band signal and a low frequency band signal; (b) performing MDCT on the low frequency band signal and converting from the time domain to the frequency domain; (c) quantizing the signal on which the MDCT is performed and encoding the bitplane based on a context; (d) converting the high frequency band signal and the low frequency band signal from time domain to frequency domain or time / frequency domain, respectively; (e) generating and encoding bandwidth extension information representing a characteristic of the converted high frequency band signal using the converted low frequency band signal; And (f) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal.

In addition, the other problem is (a) dividing the input signal into a high frequency band signal and a low frequency band signal; (b) performing MDCT on the low frequency band signal and converting from the time domain to the frequency domain; (c) quantizing the signal on which the MDCT is performed and encoding the bitplane based on a context; (d) converting the high frequency band signal and the low frequency band signal from time domain to frequency domain or time / frequency domain, respectively; (e) generating and encoding bandwidth extension information representing a characteristic of the converted high frequency band signal using the converted low frequency band signal; And (f) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal. A computer-readable recording medium having recorded thereon a program for executing an encoding method of an audio signal. Is achieved by.

In addition, the method for encoding an audio signal according to the present invention for solving the another problem comprises the steps of (a) converting the input signal from the time domain to the frequency domain; (b) dividing the converted input signal into a high frequency band signal and a low frequency band signal; (c) quantizing the low frequency band signal and encoding the bitplane based on a context; (d) generating and encoding bandwidth extension information representing the characteristics of the high frequency band signal using the low frequency band signal; And (e) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal.

In addition, the further object is (a) converting the input signal from the time domain to the frequency domain; (b) dividing the converted input signal into a high frequency band signal and a low frequency band signal; (c) quantizing the low frequency band signal and encoding the bitplane based on a context; (d) generating and encoding bandwidth extension information representing the characteristics of the high frequency band signal using the low frequency band signal; And (e) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal. A computer-readable recording medium having recorded thereon a program for executing an encoding method of an audio signal. Is achieved by.

In addition, according to another aspect of the present invention, there is provided a method of decoding an audio signal, the method comprising: (a) receiving an encoding result of an audio signal; Decoding and inverse quantizing to generate a low frequency band signal; (c) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; (d) performing an Inverse Modified Discrete Cosine Transform (MDCT) on each of the low frequency band signal and the high frequency band signal to convert from the frequency domain to the time domain; And (e) synthesizing the converted low frequency band signal and the converted high frequency band signal.

In addition, the another object is (a) receiving an encoding result of the audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; (d) performing an Inverse Modified Discrete Cosine Transform (MDCT) on each of the low frequency band signal and the high frequency band signal to convert from the frequency domain to the time domain; And (e) synthesizing the converted low frequency band signal and the converted high frequency band signal, by a computer readable recording medium having recorded thereon a program for executing a method of decoding an audio signal.

In addition, the decoding method of an audio signal according to the present invention for solving the another problem comprises the steps of: (a) receiving an encoding result of the audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) performing inverse MDCT on the low frequency band signal and converting from the frequency domain to the time domain; (d) converting the low frequency band signal subjected to the inverse MDCT into a frequency domain or a time / frequency domain; (e) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal converted into the frequency domain or the time / frequency domain using the decoded bandwidth information; (f) inversely converting the high frequency band signal into a time domain; And (g) synthesizing the inverse transformed high frequency band signal and the converted low frequency band signal.

In addition, the another object is (a) receiving an encoding result of the audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) performing inverse MDCT on the low frequency band signal to convert from frequency domain to time domain; (d) converting the low frequency band signal subjected to the inverse MDCT into a frequency domain or a time / frequency domain; (e) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal converted into the frequency domain or the time / frequency domain using the decoded bandwidth extension information; (f) inversely converting the high frequency band signal into a time domain; And (g) synthesizing the inversely transformed high frequency band signal and the converted low frequency band signal, by a computer readable recording medium having recorded thereon a program for executing a method for decoding an audio signal.

In addition, the decoding method of an audio signal according to the present invention for solving the another problem comprises the steps of: (a) receiving an encoding result of the audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; (d) synthesizing the low frequency band signal and the high frequency band signal; And (d) performing inverse MDCT on the synthesized signal to convert from the frequency domain to the time domain.

In addition, the another object is (a) receiving an encoding result of the audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth information; (d) synthesizing the low frequency band signal and the high frequency band signal; And (d) performing inverse MDCT on the synthesized signal and converting the frequency signal from the frequency domain to the time domain, by means of a computer-readable recording medium having recorded thereon a program for executing a method of decoding an audio signal. .

In addition, an apparatus for encoding an audio signal according to the present invention for solving the above another problem comprises: a band dividing unit for dividing an input signal into a high frequency band signal and a low frequency band signal; A converter for converting the high frequency band signal and the low frequency band signal from a time domain to a frequency domain, respectively; A low frequency band encoder which quantizes the converted low frequency band signal and encodes the bit frequency into a bit plane based on a context; And a bandwidth extension encoder for generating and encoding bandwidth extension information indicating a characteristic of the converted high frequency band signal by using the converted low frequency band signal.

In addition, an apparatus for encoding an audio signal according to the present invention for solving the above another problem comprises: a band dividing unit for dividing an input signal into a high frequency band signal and a low frequency band signal; An MDCT application unit performing an MDCT on the low frequency band signal and converting the time domain from the time domain to the frequency domain; A low frequency band encoder for quantizing the output of the MDCT applier and encoding the bitplane based on a context; A converter for converting the high frequency band signal and the low frequency band signal from a time domain into a frequency domain or a time / frequency domain; And a bandwidth extension encoder for generating and encoding bandwidth extension information indicating a characteristic of the converted high frequency band signal by using the converted low frequency band signal.

In addition, an apparatus for encoding an audio signal according to an embodiment of the present invention for solving the another problem includes a converter for converting the input signal from the time domain to the frequency domain; A band dividing unit dividing the converted input signal into a high frequency band signal and a low frequency band signal; A low frequency band encoder for quantizing the low frequency band signal and encoding the bit frequency based on a context; And a bandwidth extension encoder for generating and encoding bandwidth extension information indicating a characteristic of the high frequency band signal by using the low frequency band signal.

In addition, an apparatus for decoding an audio signal according to the present invention for solving the another object is a low frequency band decoder for generating a low frequency band signal by decoding and inverse quantization of the encoded bit plane based on the context; A bandwidth extension decoder for decoding encoded bandwidth extension information and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; An inverse MDCT performing unit performing inverse MDCT on each of the low frequency band signal and the high frequency band signal and converting from the frequency domain to the time domain; And a band synthesizer for synthesizing the converted low frequency band signal and the converted high frequency band signal.

In addition, an apparatus for decoding an audio signal according to the present invention for solving the another object is a low frequency band decoder for generating a low frequency band signal by decoding and inverse quantization of the encoded bit plane based on the context; An inverse MDCT application unit performing inverse MDCT on the low frequency band signal and converting from the frequency domain to the time domain; A converter for converting the low frequency band signal subjected to the inverse MDCT to a frequency domain or a time / frequency domain; A bandwidth extension decoder for decoding encoded bandwidth extension information and generating a high frequency band signal from the low frequency band signal converted into the frequency domain or the time / frequency domain using the decoded bandwidth extension information; An inverse transformer for inversely transforming the decoded high frequency band signal into a time domain; And a band synthesizer configured to synthesize the inverse transformed high frequency band signal and the converted low frequency band signal.

In addition, an apparatus for decoding an audio signal according to the present invention for solving the another object is a low frequency band decoder for generating a low frequency band signal by decoding and inverse quantization of the encoded bit plane based on the context; A bandwidth extension decoder for decoding encoded bandwidth extension information and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; A band synthesizer for synthesizing the low frequency band signal and the high frequency band signal; And an inverse MDCT application unit performing inverse MDCT on the synthesized signal and converting from the frequency domain to the time domain.

According to the present invention, the input signal is divided into a high frequency band signal and a low frequency band signal, the high frequency band signal and the low frequency band signal are respectively converted from the time domain to the frequency domain, and the converted low frequency band signal is quantized to perform bit based on context. And encoding and generating bandwidth extension information representing the characteristics of the transformed high frequency band signal using the transformed low frequency band signal, and outputting the encoded bitplane and the encoded bandwidth extension information as the encoding result of the input signal. As a result, the sound quality can be improved by efficiently encoding the high frequency components at a limited bit rate.

In addition, according to the present invention, the encoding result of the audio signal is input, the encoded bitplane included in the encoding result is decoded and dequantized based on a context to generate a low frequency band signal, and the encoded bandwidth included in the encoding result. Decode the extension information, generate a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information, and perform an Inverse Modified Discrete Cosine Transform (Inverse MDCT) on each of the low frequency band signal and the high frequency band signal in the frequency domain By converting to the time domain and synthesizing the converted low frequency band signal and the converted high frequency band signal, it is possible to efficiently decode the high frequency component from the bitstream encoded at the limited bit rate.

With respect to the embodiments of the present invention disclosed in the text, specific structural to functional descriptions are merely illustrated for the purpose of describing embodiments of the present invention, embodiments of the present invention may be implemented in various forms and It should not be construed as limited to the embodiments described in.

As the inventive concept allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the text. However, this is not intended to limit the present invention to the specific disclosed form, it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. In describing the drawings, similar reference numerals are used for the components.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in the commonly used dictionaries should be construed as having meanings consistent with the meanings in the context of the related art and shall not be construed in ideal or excessively formal meanings unless expressly defined in this application. Do not.

Hereinafter, with reference to the accompanying drawings, it will be described in detail a preferred embodiment of the present invention. The same reference numerals are used for the same elements in the drawings, and duplicate descriptions of the same elements are omitted.

1 is a block diagram illustrating an apparatus for encoding an audio signal according to an embodiment of the present invention.

Referring to FIG. 1, an apparatus for encoding an audio signal includes a band splitter 100, a first MDCT applier 110, a frequency linear prediction performer 120, a multi-resolution analyzer 130, and a quantization unit ( 140, a PQ-SPSC module 150, a context-based bitplane encoder 160, a second MDCT applier 170, a bandwidth extension encoder 180, and a multiplexer 190.

The band splitting unit 100 divides the input signal IN into a low band signal (LB) and a high frequency band signal (HB). Here, the input signal IN may be a pulse code modulation (PCM) signal obtained by modulating an analog voice signal or an audio signal into a digital signal. Here, the low frequency band signal may be a signal corresponding to a frequency lower than an arbitrary threshold value, and the high frequency band signal may be a signal corresponding to a frequency higher than an arbitrary threshold value.

The first MDCT applying unit 110 performs a modified discrete cosine transform (MDCT) on the low frequency band signal LB divided by the band dividing unit 100 to convert the low frequency band signal LB from the time domain to the frequency domain. To convert.

The frequency linear prediction performing unit 120 performs frequency linear prediction on the low frequency band signal converted into the frequency domain by the first MDCT applying unit 110. Here, frequency linear prediction is a method of approximating a signal at a current frequency to a linear combination of signals at a previous frequency. In detail, the frequency linear prediction performing unit 120 calculates a coefficient of the linear prediction filter such that a prediction error, which is a difference between the signal on which the linear prediction is performed and the signal at the current frequency, is minimized, Linear predictive filtering is performed on the low frequency band signal converted into the frequency domain according to the calculated coefficients. In this case, the frequency linear prediction performing unit 120 may improve the efficiency of encoding by performing vector quantization expressed by a vector index on values corresponding to coefficients of the linear prediction filter.

In detail, the frequency linear prediction performing unit 120 performs frequency linear prediction when the low frequency band signal converted into the frequency domain by the first MDCT applying unit 110 is a speech signal or a pitched signal. can do. In other words, the frequency linear prediction performing unit 120 may improve the efficiency of encoding by performing frequency linear prediction according to the characteristics of the input signal.

The multi-resolution analysis unit 130 receives the low-frequency band signal converted into the frequency domain from the first MDCT application unit 110 or the signal filtered by the frequency linear prediction execution unit 120 and instantly. Analyze audio spectral coefficients of varying signals in multi-resolution. Specifically, the multi-resolution analysis unit 130 may filter the signal filtered by the frequency linear prediction performing unit 120 according to the intensity of the change in the audio spectrum in two types (eg, a stabilization type and a short). (short) type can be analyzed.

Specifically, the multi-resolution analysis unit 130 is a low-frequency band signal transformed into the frequency domain in the first MDCT application unit 110 or a signal filtered by the frequency linear prediction performing unit 120 is a transient signal. In the case of multi-resolution. In other words, the multi-resolution analyzer 130 may improve the efficiency of encoding by performing multi-resolution analysis according to the characteristics of the input signal.

The quantization unit 140 quantizes the result output from the frequency linear prediction execution unit 120 or the multi-resolution analysis unit 130.

The PQ-SPSC module 150 performs a side-polar coordination stereo encoding of a frequency spectrum on the result output from the quantization unit 140.

The context-based bitplane coding unit 160 encodes the output of the PQ-SPSC module 150 into a bitplane based on a context. Here, the context-based bitplane encoder 160 may encode a result output from the PQ-SPSC module 150 into a bitplane by using Huffman coding.

The second MDCT applying unit 170 performs MDCT on the high frequency band signal HB divided by the band dividing unit 100 to convert the high frequency band signal HB from the time domain to the frequency domain.

The bandwidth extension encoder 180 may convert the low frequency band signal converted into the frequency domain by the first MDCT applier 110 to transmit components of the high frequency band signal converted into the frequency domain by the second MDCT applier 170. Bandwidth extension information representing the characteristics of the high frequency band signal is generated and encoded. Here, it will be understood by those skilled in the art that the bandwidth extension information may include various information such as energy level or envelope for the high frequency band signal. More specifically, the bandwidth extension encoder 180 may generate bandwidth extension information by using a low frequency band signal based on a property that a high correlation exists between a high frequency and a low frequency band of an audio signal.

The multiplexer 190 multiplexes the results of the encoding performed by the frequency linear prediction performer 120, the PQ-SPSC module 150, the context-based bitplane encoder 160, and the bandwidth extension encoder 180. Generates a bit stream and outputs it through the output terminal OUT.

2 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 2, the apparatus for encoding an audio signal includes a band splitter 200, an MDCT applier 210, a frequency linear prediction performer 220, a multi-resolution analyzer 230, and a quantizer 240. The PQ-SPSC module 250, the context-based bitplane encoder 260, the low frequency band converter 270, the high frequency band converter 275, the bandwidth extension encoder 280, and the multiplexer 290. Include.

The band dividing unit 200 divides the input signal IN into a low frequency band signal LB and a high frequency band signal HB. The input signal IN may be a PCM signal obtained by modulating an analog voice signal or an audio signal into a digital signal. Here, the low frequency band signal may be a signal corresponding to a frequency lower than an arbitrary threshold value, and the high frequency band signal may be a signal corresponding to a frequency higher than an arbitrary threshold value.

The MDCT application unit 210 performs a modified discrete cosine transform (MDCT) on the low frequency band signal LB divided by the band divider 200 to convert the low frequency band signal LB from the time domain to the frequency domain. .

The frequency linear prediction performing unit 220 performs frequency linear prediction on the low frequency band signal converted into the frequency domain by the MDCT application unit 210. Here, frequency linear prediction is a method of approximating a signal at a current frequency to a linear combination of signals at a previous frequency. Specifically, the frequency linear prediction performing unit 220 calculates the coefficients of the linear prediction filter such that the prediction error, which is the difference between the signal on which the linear prediction is performed and the signal at the current frequency, is minimal, and according to the calculated coefficients Linear predictive filtering is performed on the low-frequency band signal transformed by. In this case, the frequency linear prediction performing unit 520 may improve the efficiency of encoding by performing vector quantization expressed by a vector index on a value corresponding to the coefficient of the linear prediction filter.

Specifically, the frequency linear prediction performing unit 220 may perform frequency linear prediction when the low frequency band signal converted into the frequency domain by the MDCT application unit 210 is a speech signal or a pitched signal. have. In other words, the frequency linear prediction performing unit 220 may improve the efficiency of encoding by performing frequency linear prediction according to the characteristics of the input signal.

The multi-resolution analyzer 230 receives the filtered signal from the frequency linear prediction performer 220 and analyzes the audio spectral coefficients of the instantaneously changing signal with multi-resolution. In detail, the multi-resolution analyzer 230 may filter the audio spectrum filtered by the frequency linear prediction performer 220 according to the intensity of the change in the audio spectrum in two types (for example, a stabil type and a short type). It can be analyzed by dividing by.

Specifically, the multi-resolution analysis unit 230 is a case where the low-frequency band signal converted into the frequency domain in the MDCT application unit 210 or the signal filtered by the frequency linear prediction performing unit 220 is a transient signal. It can be analyzed by multi-resolution. In other words, the multi-resolution analyzer 230 may improve the efficiency of encoding by performing multi-resolution analysis according to the characteristics of the input signal.

The quantization unit 240 quantizes the signal filtered by the frequency linear prediction execution unit 220 or the result output from the multi-resolution analyzer 230.

The PQ-SPSC module 250 performs side-polar coordination stereo encoding of the frequency spectrum on the result output from the quantization unit 240.

The context-based bitplane encoder 260 encodes the output of the PQ-SPSC module 250 into a bitplane based on the context. Here, the context-based bitplane encoder 260 may encode the quantized result in the PQ-SPSC module 250 into a bitplane using Huffman Coding.

The low frequency band converter 270 converts the low frequency band signal LB divided by the band divider 200 from a time domain to a frequency domain or a time / frequency domain using a conversion technique other than MDCT. For example, the low frequency band converter 270 uses the Modified Discrete Sine Transform (MDST), Fast Fourier Transform (FFT), Quadrature Mirror Filters (QMF), and the like to generate a low frequency band signal LB in the time domain Or convert to the time / frequency domain.

The high frequency band converter 275 converts the high frequency band signal HB split by the band divider 200 from the time domain to the frequency domain or the time / frequency domain using a conversion technique other than MDCT. Here, the high frequency band converter 275 uses the same conversion technique as the low frequency band converter 270. For example, the high frequency band converter 275 may convert the high frequency band signal HB from the time domain to the frequency domain or the time / frequency domain using MDST, FFT, and QMF.

The bandwidth extension encoder 280 uses the low frequency band signal converted into the frequency domain by the low frequency band converter 270 to increase the bandwidth of the high frequency band signal converted into the frequency domain by the high frequency band converter 275. Generate and encode information. Here, it will be understood by those skilled in the art that the bandwidth extension information may include various information such as energy level or envelope for the high frequency band signal. More specifically, the bandwidth extension encoder 280 may generate bandwidth extension information by using the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal.

The multiplexer 290 multiplexes the result of the encoding performed by the frequency linear prediction performer 220, the PQ-SPSC module 250, the context-based bitplane encoder 260, and the bandwidth extension encoder 280. To generate a bit stream and output it through the output terminal OUT.

3 is a block diagram showing an apparatus for encoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 3, the apparatus for encoding an audio signal includes an MDCT application unit 300, a band divider 310, a frequency linear prediction performer 320, a multi-resolution analyzer 330, a quantizer 340, A PQ-SPSC module 350, a context-based bitplane encoder 360, a bandwidth extension encoder 370, and a multiplexer 380 are included.

The MDCT application unit 300 performs an MDCT on the input signal IN to convert the input signal IN from the time domain to the frequency domain. The input signal IN may be a PCM signal obtained by modulating an analog voice signal or an audio signal into a digital signal.

The band dividing unit 310 divides the signal converted into the frequency domain by the MDCT applying unit 300 into a low frequency band signal LB and a high frequency band signal HB.

The frequency linear prediction performer 320 performs frequency linear prediction on the low frequency band signal LB divided by the band divider 310. Here, frequency linear prediction is a method of approximating a signal at a current frequency to a linear combination of signals at a previous frequency. Specifically, the frequency linear prediction performing unit 320 calculates coefficients of the linear prediction filter such that the prediction error, which is the difference between the signal on which the linear prediction is performed and the signal at the current frequency, is minimized, and according to the calculated coefficients Linear predictive filtering is performed on the low-frequency band signal transformed by. In this case, the frequency linear prediction performing unit 320 may improve the efficiency of encoding by performing vector quantization expressed by a vector index on values corresponding to coefficients of the linear prediction filter.

In detail, the frequency linear prediction performing unit 320 may perform frequency linear prediction when the low frequency band signal LB divided by the band dividing unit 310 is a speech signal or a pitched signal. have. In other words, the frequency linear prediction performing unit 320 may improve the efficiency of encoding by performing frequency linear prediction according to the characteristics of the input signal.

The multi-resolution analyzer 330 receives the filtered signal from the frequency linear prediction performer 320 and analyzes the audio spectral coefficients of the signal that is instantaneously changed in multi resolution. In detail, the multi-resolution analyzer 330 may filter the audio spectrum filtered by the frequency linear prediction performer 320 according to the intensity of the change in the audio spectrum in two types (for example, a stabil type and a short type). It can be analyzed by dividing by.

In detail, the multi-resolution analyzer 330 is a low frequency band signal LB divided by the band divider 310 or a signal filtered by the frequency linear prediction performer 320 when the signal is a transient signal. It can be analyzed by multi-resolution. In other words, the multi-resolution analyzer 330 may improve the encoding efficiency by performing multi-resolution analysis according to the characteristics of the input signal.

The quantizer 340 quantizes the signal filtered by the frequency linear prediction performer 320 or the result output from the multi-resolution analyzer 330.

The PQ-SPSC module 350 performs side-polar coordination stereo encoding of the frequency spectrum on the result output from the quantization unit 340.

The context-based bitplane encoder 360 encodes a result output from the PQ-SPSC module 350 into a bitplane based on the context. Here, the context-based bitplane encoder 360 may encode a result output from the PQ-SPSC module 350 into a bitplane by using Huffman coding.

The bandwidth extension encoder 370 uses the low frequency band signal LB divided by the band divider 310 to obtain bandwidth extension information indicating the characteristics of the high frequency band signal HB split by the band divider 310. Generate and encode. Here, it will be understood by those skilled in the art that the bandwidth extension information may include various information such as energy level or envelope for the high frequency band signal. More specifically, the bandwidth extension encoder 370 may generate bandwidth extension information by using the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal.

The multiplexer 380 multiplexes the results of the encoding performed by the frequency linear prediction performer 320, the PQ-SPSC module 350, the context-based bitplane encoder 360, and the bandwidth extension encoder 370. Generates a bit stream and outputs it through the output terminal OUT.

4 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 4, the apparatus for encoding an audio signal includes a band splitter 400, a first MDCT applier 410, a frequency linear prediction performer 420, a multi-resolution analyzer 430, and a quantizer 440. ), A context-based bitplane encoder 450, a second MDCT applier 460, a bandwidth extension encoder 470, and a multiplexer 480.

The band divider 400 divides the input signal IN into a low frequency band signal LB and a high frequency band signal HB. The input signal IN may be a PCM signal obtained by modulating an analog voice signal or an audio signal into a digital signal. Here, the low frequency band signal may be a signal corresponding to a frequency lower than an arbitrary threshold value, and the high frequency band signal may be a signal corresponding to a frequency higher than an arbitrary threshold value.

The first MDCT applying unit 410 performs MDCT on the low frequency band signal LB divided by the band dividing unit 400 to convert the low frequency band signal LB from the time domain to the frequency domain. Here, the time domain is a domain representing the magnitude of the input signal IN (for example, energy or sound pressure) over time, and the frequency domain is a domain representing the magnitude of the input signal IN according to the change of frequency. to be.

The frequency linear prediction performing unit 420 performs frequency linear prediction on the low frequency band signal converted into the frequency domain by the first MDCT applying unit 410. Here, frequency linear prediction is a method of approximating a signal at a current frequency to a linear combination of signals at a previous frequency. Specifically, the frequency linear prediction performing unit 420 calculates a coefficient of the linear prediction filter such that a prediction error, which is a difference between the signal on which the linear prediction is performed and the signal at the current frequency, is minimized, Linear predictive filtering is performed on the low frequency band signal converted into the frequency domain according to the calculated coefficients. In this case, the frequency linear prediction performing unit 420 may improve the efficiency of encoding by performing vector quantization expressed by a vector index on a value corresponding to the coefficient of the linear prediction filter.

In detail, the frequency linear prediction performing unit 420 performs frequency linear prediction when the low frequency band signal converted into the frequency domain by the first MDCT applying unit 410 is a speech signal or a pitched signal. can do. In other words, the frequency linear prediction performing unit 420 may improve the efficiency of encoding by performing frequency linear prediction according to the characteristics of the input signal.

The multi-resolution analyzer 430 receives the filtered signal from the frequency linear prediction performer 420 and analyzes the audio spectral coefficients of the instantaneously changing signal in multi-resolution. In detail, the multi-resolution analyzer 430 may filter the signal filtered by the frequency linear prediction performer 420 according to the intensity of the change in the audio spectrum in two types (eg, a stabilization type and a short). (short) type can be analyzed.

Specifically, the multi-resolution analysis unit 430 is a low-frequency band signal transformed into a frequency domain in the first MDCT application unit 410 or a signal filtered by the frequency linear prediction performing unit 420 is a transient signal. In the case of multi-resolution. In other words, the multi-resolution analyzer 430 may improve encoding efficiency by performing multi-resolution analysis according to the characteristics of the input signal.

The quantizer 440 quantizes the signal filtered by the frequency linear prediction performer 420 or the result output from the multi-resolution analyzer 430.

The context-based bitplane encoder 450 encodes the result of the quantization in the quantizer 440 into a bitplane based on the context. In detail, the context-based bit plane encoder 450 may encode the result quantized by the quantizer 440 into a bitplane using Huffman coding.

As such, the frequency linear prediction execution unit 420, the multi-resolution analysis unit 430, the quantization unit 440, and the context-based bitplane encoder 450 are converted low frequencies output from the MDCT application unit 410. Since the encoding is performed on the band signal, it may be referred to as a low frequency band encoder.

The second MDCT application unit 460 converts the high frequency band signal HB from the time domain to the frequency domain by performing MDCT on the high frequency band signal HB divided by the band divider 400.

The bandwidth extension encoder 470 indicates the characteristics of the high frequency band signal converted into the frequency domain by the second MDCT applier 460 using the low frequency band signal converted into the frequency domain by the first MDCT applier 410. Generate and encode bandwidth extension information. Here, it will be understood by those skilled in the art that the bandwidth extension information may include various information such as energy level or envelope for the high frequency band signal. More specifically, the bandwidth extension encoder 470 may generate bandwidth extension information by using the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal.

The multiplexer 480 multiplexes the result of the encoding performed by the frequency linear prediction performer 420, the context-based bitplane encoder 450, and the bandwidth extension encoder 470 to generate a bit stream, and output terminal OUT. Output through

5 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 5, the apparatus for encoding an audio signal includes a band splitter 500, an MDCT applier 510, a frequency linear prediction performer 520, a multi-resolution analyzer 530, a quantizer 540, A context-based bitplane encoder 550, a low frequency band converter 560, a high frequency band converter 570, a bandwidth extension encoder 580, and a multiplexer 590 are included.

The band divider 500 divides the input signal IN into a low frequency band signal LB and a high frequency band signal HB. The input signal IN may be a PCM signal obtained by modulating an analog voice signal or an audio signal into a digital signal. Here, the low frequency band signal may be a signal corresponding to a frequency lower than an arbitrary threshold value, and the high frequency band signal may be a signal corresponding to a frequency higher than an arbitrary threshold value.

The MDCT application unit 510 performs MDCT on the low frequency band signal LB divided by the band divider 400 to convert the low frequency band signal LB from the time domain to the frequency domain.

The frequency linear prediction performing unit 520 performs frequency linear prediction on the low frequency band signal converted into the frequency domain by the MDCT application unit 510. Here, frequency linear prediction is a method of approximating a signal at a current frequency to a linear combination of signals at a previous frequency. Specifically, the frequency linear prediction performing unit 520 calculates coefficients of the linear prediction filter such that the prediction error, which is a difference between the signal on which the linear prediction is performed and the signal at the current frequency, is minimized, and according to the calculated coefficients, the frequency domain Linear predictive filtering is performed on the low-frequency band signal transformed by. In this case, the frequency linear prediction performing unit 520 may improve the efficiency of encoding by performing vector quantization expressed by a vector index on a value corresponding to the coefficient of the linear prediction filter.

In detail, the frequency linear prediction performing unit 520 may perform frequency linear prediction when the low frequency band signal converted into the frequency domain by the MDCT application unit 510 is a speech signal or a pitched signal. have. In other words, the frequency linear prediction performing unit 520 may improve the efficiency of encoding by performing frequency linear prediction according to the characteristics of the input signal.

The multi-resolution analyzer 530 receives the filtered signal from the frequency linear prediction performer 520 and analyzes the audio spectral coefficients of the instantaneously changing signal in multi-resolution. In detail, the multi-resolution analyzer 530 may filter the audio spectrum filtered by the frequency linear prediction performer 420 according to the intensity of the change in the audio spectrum in two types (for example, a stabil type and a short type). It can be analyzed by dividing by.

In detail, the multi-resolution analyzer 530 is a case where the low frequency band signal transformed into the frequency domain by the MDCT application unit 510 or the signal filtered by the frequency linear prediction performer 520 is a transient signal. It can be analyzed by multi-resolution. In other words, the multi-resolution analyzer 530 may improve the encoding efficiency by performing multi-resolution analysis according to the characteristics of the input signal.

The quantizer 540 quantizes the signal filtered by the frequency linear prediction performer 520 or the result output from the multi-resolution analyzer 530.

The context-based bitplane encoder 550 encodes the quantized result in the quantizer 540 into a bitplane based on the context. In detail, the context-based bit plane encoder 550 may encode the quantized result in the quantizer 540 into a bit plane using Huffman Coding.

As such, the frequency linear prediction execution unit 520, the multi-resolution analysis unit 530, the quantization unit 540, and the context-based bitplane encoder 550 are converted low frequencies output from the MDCT application unit 510. Since the encoding is performed on the band signal, it may be referred to as a low frequency band encoder.

The low frequency band converter 560 converts the low frequency band signal LB divided by the band divider 500 from the time domain to the frequency domain or the time / frequency domain using a conversion technique other than MDCT. For example, the low frequency band converter 560 may convert the low frequency band signal LB from the time domain to the frequency domain or the time / frequency domain using MDST, FFT, and QMF. Here, the time domain is a domain representing the magnitude (eg, energy or sound pressure) of the low frequency band signal LB over time. In contrast, the frequency domain is a domain indicating the magnitude of the low frequency band signal LB according to the change of frequency. The time / frequency domain is a domain indicating the magnitude of the low frequency band signal LB according to the passage of time and the change of frequency.

The high frequency band converter 570 converts the high frequency band signal HB split by the band divider 500 from the time domain to the frequency domain or the time / frequency domain using a conversion technique other than MDCT. Here, the high frequency band converter 570 uses the same conversion technique used by the low frequency band converter 560. For example, the high frequency band converter 570 may convert the high frequency band signal HB from the time domain to the frequency domain or the time / frequency domain using MDST, FFT, and QMF.

The bandwidth extension encoder 580 uses the low frequency band signal converted into the frequency domain by the low frequency band converter 560 to increase the bandwidth of the high frequency band signal converted into the frequency domain by the high frequency band converter 570. Generate and encode information. Here, it will be understood by those skilled in the art that the bandwidth extension information may include various information such as energy level or envelope for the high frequency band signal. More specifically, the bandwidth extension encoder 580 may generate bandwidth extension information using a low frequency band signal based on a property that a high correlation exists between a high frequency and a low frequency band of the audio signal.

The multiplexer 590 multiplexes the result of the encoding performed by the frequency linear prediction performer 520, the context-based bitplane encoder 550, and the bandwidth extension encoder 580 to generate a bit stream, and outputs the output terminal OUT. Output through

6 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 6, the apparatus for encoding an audio signal includes an MDCT application unit 600, a band divider 610, a frequency linear prediction performer 620, a multi-resolution analyzer 630, a quantizer 640, The context-based bitplane encoder 650, the bandwidth extension encoder 660, and the multiplexer 670 are included.

The MDCT application unit 600 performs an MDCT on the input signal IN to convert the input signal IN from the time domain to the frequency domain. The input signal IN may be a PCM signal obtained by modulating an analog voice signal or an audio signal into a digital signal.

The band dividing unit 610 divides the signal converted into the frequency domain by the MDCT applying unit 600 into a low frequency band signal LB and a high frequency band signal HB. Here, the low frequency band signal may be a signal corresponding to a frequency lower than an arbitrary threshold value, and the high frequency band signal may be a signal corresponding to a frequency higher than an arbitrary threshold value.

The frequency linear prediction performer 620 performs frequency linear prediction on the low frequency band signal LB divided by the band divider 610. Here, frequency linear prediction is a method of approximating a signal at a current frequency to a linear combination of signals at a previous frequency. Specifically, the frequency linear prediction performing unit 620 calculates coefficients of the linear prediction filter such that the prediction error, which is the difference between the signal on which the linear prediction is performed and the signal at the current frequency, is minimized, and according to the calculated coefficients, the frequency domain Linear predictive filtering is performed on the low-frequency band signal transformed by. In this case, the frequency linear prediction performing unit 620 may improve the efficiency of encoding by performing vector quantization expressed by a vector index on values corresponding to coefficients of the linear prediction filter.

In detail, the frequency linear prediction performing unit 620 may perform frequency linear prediction when the low frequency band signal divided by the band dividing unit 610 is a speech signal or a pitched signal. In other words, the frequency linear prediction performing unit 620 may improve the efficiency of encoding by performing frequency linear prediction according to the characteristics of the input signal.

The multi-resolution analyzer 630 receives the filtered signal from the frequency linear prediction performer 620 and analyzes the audio spectral coefficients of the signal that is instantaneously changed in multi resolution. In detail, the multi-resolution analyzer 630 may filter the audio spectrum filtered by the frequency linear prediction performer 620 according to the intensity of the change in the audio spectrum in two types (for example, a stabil type and a short type). It can be analyzed by dividing by.

In detail, the multi-resolution analyzer 630 performs the multi-resolution when the low frequency band signal divided by the band divider 610 or the signal filtered by the frequency linear prediction performer 620 is a transient signal. Can be analyzed. In other words, the multi-resolution analyzer 630 may improve the efficiency of encoding by performing multi-resolution analysis according to the characteristics of the input signal.

The quantizer 640 quantizes the signal filtered by the frequency linear prediction performer 620 or the result output from the multi-resolution analyzer 630.

The context-based bitplane encoder 650 encodes the quantized result in the quantizer 640 into a bitplane based on the context. In detail, the context-based bitplane encoder 650 may encode the quantized result in the quantizer 640 into a bitplane using Huffman Coding.

As such, the frequency linear prediction performing unit 620, the multi-resolution analysis unit 630, the quantization unit 640, and the context-based bitplane encoder 650 are converted low frequencies output from the MDCT application unit 610. Since the encoding is performed on the band signal, it may be referred to as a low frequency band encoder.

The bandwidth extension encoder 660 generates and encodes bandwidth extension information indicating a characteristic of the high frequency band signal HB using the low frequency band signal LB. Here, it will be understood by those skilled in the art that the bandwidth extension information may include various information such as energy level or envelope for the high frequency band signal. More specifically, the bandwidth extension encoder 660 may generate bandwidth extension information by using the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal.

The multiplexer 670 multiplexes the result of the encoding performed by the frequency linear prediction performer 620, the context-based bitplane encoder 650, and the bandwidth extension encoder 660 to generate a bit stream, and output terminal OUT. Output through

7 is a block diagram illustrating an apparatus for decoding an audio signal according to an embodiment of the present invention.

Referring to FIG. 7, an apparatus for decoding an audio signal includes a demultiplexer 700, a context-based bitplane decoder 710, a PQ-SPSC module 720, a dequantizer 730, and a multi-resolution synthesizer. 740, an inverse frequency linear prediction execution unit 750, a first inverse MDCT application unit 760, a bandwidth extension decoding unit 770, a second inverse MDCT application unit 780, and a band synthesis unit 790. It is done by

The demultiplexer 700 demultiplexes a bit stream output from an encoding stage. In detail, the demultiplexer 700 separates each part of the data level into data parts corresponding to each unit, and analyzes and outputs information of a bitstream related thereto in the corresponding unit. Here, the information output from the demultiplexer 700 includes description analysis of the audio spectrum stream to be used by the PQ-SPSC module 720, quantization values and other reconstruction information, reconstruction information of the quantization spectrum, and context-based bitplanes. Information of decoding, information of quantization wide-polar coordination stereo decoding, signal type information, information of frequency-domain linear prediction and vector quantization, and coded bandwidth extension information have.

The context-based bitplane decoding unit 710 decodes the encoded bitplane based on the context. Here, the context-based bitplane decoder 710 receives information output from the demultiplexer 700 and performs Huffman decoding to perform frequency spectrum, coding band mode information, and scale. Restore the scale factor. The context-based bitplane decoder 710 receives Prejudice coding band mode information, a scale factor of Prejudice coding, and a frequency spectrum of Prejudice coding, and receives a coding band mode value, a decoding cosmetic display of a scale factor, and a quantization value of a frequency spectrum. Outputs

The PQ-SPSC module 720 receives a result output from the context-based bitplane decoder 710 and performs side-polar coordination stereo decoding of the frequency spectrum. Here, the PQ-SPSC module 720 receives side-polar coordination stereo and coupling information of the frequency spectrum, performs side-polar coordination stereo decoding, and then outputs a quantized frequency spectrum.

The inverse quantization unit 730 inverse quantizes the result output from the PQ-SPSC module 720.

The multi-resolution synthesis unit 740 receives the output of the inverse quantization unit 730 and synthesizes the multi-resolution of the audio spectral coefficients of the instantaneously changing signal. In detail, the multi-resolution synthesis unit 740 may improve the decoding efficiency by synthesizing the output of the dequantization unit 730 into the multi-resolution when the audio signal is analyzed as the multi-resolution at the encoding end. Here, the multi-resolution synthesis unit 740 receives a reserve quantization spectrum / difference spectrum and outputs a reconstruction spectrum / difference spectrum.

An inverse frequency linear prediction performing unit (750) is a frequency output from an inverse quantizer 730 or an output of the multi-resolution synthesis unit 740 and an encoder received from the demultiplexer 700. Synthesize the result of performing the linear prediction. In detail, the inverse frequency linear prediction execution unit 750 outputs the result of the frequency linear prediction when the audio signal is encoded in the encoding stage, or multi-resolution synthesis of the inverse quantization unit 730. The efficiency of the decoding may be improved by combining with the output of the unit 740. Here, the inverse frequency linear prediction performing unit 750 effectively improves coding efficiency by employing a frequency domain prediction technique and a vector quantization technique of prediction coefficients. The inverse frequency linear prediction performing unit 750 receives a difference spectrum coefficient and an index of a vector and outputs an MDCT spectrum coefficient.

The first inverse MDCT application unit 760 performs an inverse MDCT (inverse MDCT) on the low frequency band signal output from the multi-resolution synthesis unit 740 or the inverse frequency linear prediction execution unit 750 as an inverse transform process for the encoding end. Perform inverse transform from frequency domain to time domain. Here, the first inverse MDCT applying unit 760 receives the frequency spectrum coefficients obtained as a result of the inverse quantization by the multi-resolution synthesis unit 740 or the inverse frequency linear prediction performing unit 750, and restores the frequency spectrum coefficients corresponding to the low frequency bands. Output as audio data.

The bandwidth extension decoder 770 decodes the encoded bandwidth extension information and uses the decoded bandwidth extension information to output the low frequency band signal output from the multi-resolution synthesis unit 740 or the inverse frequency linear prediction performer 750. Generate a high frequency band signal. Here, the bandwidth extension decoder 770 generates the high frequency band signal by applying the decoded bandwidth extension information to the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal. Here, the bandwidth extension information is information that can indicate the characteristics of the high frequency band of the high frequency band and is information such as energy level or envelope of the high frequency band signal.

The second inverse MDCT application unit 780 inversely converts the high frequency band signal decoded by the bandwidth extension decoder 770 from the frequency domain to the time domain by inverse MDCT.

The band synthesizing unit 790 synthesizes the low frequency band signal converted into the time domain in the first inverse MDCT applying unit 760 and the high frequency band signal converted into the time domain in the second inverse MDCT applying unit 780 and outputs the output terminal OUT. Output through

8 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 8, the audio data decoding apparatus includes a demultiplexer 800, a context-based bitplane decoder 810, a PQ-SPSC module 820, an inverse quantizer 830, and a multi-resolution synthesizer ( 840, an inverse frequency linear prediction execution unit 850, an inverse MDCT application unit 860, a transformer 865, a bandwidth extension decoder 870, an inverse transformer 880, and a band synthesizer 890. Is done.

The demultiplexer 800 demultiplexes a bit stream output from an encoding stage. The demultiplexer 800 separates each part of the data level into data parts corresponding to each unit, and analyzes and outputs information of a bitstream related thereto in the corresponding unit. In this case, the information output from the demultiplexer 800 may include description analysis of an audio spectrum stream to be used by the PQ-SPSC module 820, quantization values and other reconstruction information, reconstruction information of the quantization spectrum, and context-based bitplanes. Information of context-dependent bit plane decoding without decoding, information of quantization wide-polar coordination stereo decoding, signal type information, information of frequency-domain linear prediction and vector quantization, and coded bandwidth extension information.

The context-based bitplane decoder 810 decodes the encoded bitplane based on the context. Here, the context-based bitplane decoder 810 receives the information output from the demultiplexer 800 and performs Huffman decoding to obtain frequency spectrum, coding band mode information, and Restore the scale factor. The context-based bitplane decoder 810 receives prejudice coding band mode information, a scale factor of prejudice coding, and a frequency spectrum of prejudice coding to obtain a coding band mode value, a decoding cosmetic display of a scale factor, and a quantization value of a frequency spectrum. Output

The PQ-SPSC module 820 receives the output of the context-based bitplane decoder 810 and performs side-polar coordination stereo decoding of the frequency spectrum. Here, the PQ-SPSC module 820 receives side-polar coordination stereo and coupling information of the frequency spectrum, performs side-polar coordination stereo decoding, and then outputs a quantized frequency spectrum.

The dequantization unit 830 dequantizes the result output from the PQ-SPSC module 820.

The multi-resolution synthesizer 840 receives the output of the inverse quantizer 830 and processes multi-resolution for the audio spectral coefficients of the instantaneously changing signal. In detail, the multi-resolution synthesis unit 840 may improve the decoding efficiency by synthesizing the output of the dequantization unit 830 into a multi-resolution when the audio signal is analyzed as a multi-resolution at the encoding end. have. Here, the multi-resolution synthesis unit 840 receives a reserve quantization spectrum / difference spectrum and outputs a reconstruction spectrum / difference spectrum.

The inverse frequency linear prediction performing unit 850 synthesizes the output of the multi-resolution synthesis unit 840 and the result of performing the frequency linear prediction in the encoding stage from the demultiplexing unit 800 and performs inverse-vector quantization. In detail, the inverse frequency linear prediction performing unit 850 outputs the result of the frequency linear prediction when the audio signal is encoded in the encoding stage or outputs the inverse quantization unit 830 or the multi-resolution synthesis unit. In combination with the output of 840, the efficiency of decoding may be improved. Here, the inverse frequency linear prediction execution unit 850 effectively improves coding efficiency by employing a frequency domain prediction technique and a vector quantization technique of prediction coefficients. The inverse frequency linear prediction performing unit 850 receives a difference spectrum coefficient and an index of a vector and outputs an MDCT spectrum coefficient.

The inverse MDCT application unit 860 converts the low frequency band signal output from the multi-resolution synthesis unit 840 or the inverse frequency linear prediction unit 850 from the frequency domain to the time domain by inverse MDCT as an inverse transform process on the encoding end. Invert Here, the inverse MDCT application unit 860 receives the frequency spectrum coefficients obtained as a result of inverse quantization by the multi-resolution synthesis unit 840 or the inverse frequency linear prediction performing unit 850, and then reconstructed audio data corresponding to the low frequency band. Will output

The conversion unit 865 converts the low frequency band signal converted into the time domain in the inverse MDCT application unit 860 from the time domain to the frequency domain or the time-frequency domain using a conversion technique other than MDCT. Examples of conversion techniques used by the conversion unit 865 include MDST, FFT, and QMF. Although the transform unit 865 may not be implemented by applying MDCT, the embodiment shown in FIG. 7 is more efficient when using MDCT.

The bandwidth extension decoder 870 decodes the encoded bandwidth extension information and generates a high frequency band signal from the low frequency band signal output from the converter 865 using the decoded bandwidth extension information. Here, the bandwidth extension decoder 870 generates the high frequency band signal by applying the decoded bandwidth extension information to the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal. Here, the bandwidth extension information is information that can indicate the characteristics of the high frequency band of the high frequency band and is information such as energy level or envelope of the high frequency band signal.

The inverse transformer 880 inversely converts the high frequency band signal decoded by the bandwidth extension decoder 870 from the frequency domain or the time-frequency domain to the time domain using a conversion technique other than MDCT. Here, the inverse transformer 880 uses the same transformation technique used by the transformer 865. Examples of the conversion technique used by the inverse transform unit 880 include MDST, FFT, and QMF.

The band synthesizer 890 synthesizes the low frequency band signal converted into the time domain by the inverse MDCT application unit 860 and the high frequency band signal converted into the time domain by the inverse converter 880 and outputs the same through the output terminal OUT.

9 is a block diagram showing an apparatus for decoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 9, an audio data decoding apparatus includes a demultiplexer 900, a context-based bitplane decoder 910, a PQ-SPSC module 920, an inverse quantizer 930, and a multi-resolution synthesis unit ( 940, an inverse frequency linear prediction performer 950, a bandwidth extension decoder 960, a band synthesizer 970, and an inverse MDCT application 980.

The demultiplexer 900 demultiplexes a bit stream output from an encoding stage. The demultiplexer 900 separates each part of the data level into data parts corresponding to each unit, and analyzes and outputs information of a bitstream related thereto in the corresponding unit. Here, the information output from the demultiplexer 900 includes description analysis of the audio spectrum stream to be used by the PQ-SPSC module 920, quantization values and other reconstruction information, reconstruction information of the quantization spectrum, and context-based bitplanes. Information of context-dependent bit plane decoding without decoding, information of quantization wide-polar coordination stereo decoding, signal type information, information of frequency-domain linear prediction and vector quantization, and coded bandwidth extension information.

The context-based bitplane decoder 910 decodes the encoded bitplane based on the context. Here, the context-based bitplane decoder 910 receives the information output from the demultiplexer 900 and performs Huffman decoding to obtain frequency spectrum, coding band mode information, and Restore the scale factor. The context-based bitplane decoder 910 receives prejudice coding band mode information, a scale factor of prejudice coding, and a frequency spectrum of prejudice coding, and receives a coding band mode value, a decoding cosmetic display of a scale factor, and a quantization value of a frequency spectrum. Output

The PQ-SPSC module 920 receives the output of the context-based bitplane decoder 910 and performs side-polar coordination stereo decoding of the frequency spectrum. Here, the PQ-SPSC module 920 receives side-polar coordination stereo and coupling information of the frequency spectrum, performs side-polar coordination stereo decoding, and then outputs a quantized frequency spectrum.

The dequantizer 930 dequantizes the result output from the PQ-SPSC module 920.

The multi-resolution synthesis unit 940 receives the output of the inverse quantization unit 930 and processes multi-resolution of audio spectral coefficients of a signal that is instantaneously changed. In detail, the multi-resolution synthesis unit 90 may improve the decoding efficiency by synthesizing the output of the dequantization unit 930 into a multi-resolution when the audio signal is analyzed as a multi-resolution at the encoding end. Here, the multi-resolution synthesis unit 940 receives a reserve quantization spectrum / difference spectrum and outputs a reconstruction spectrum / difference spectrum.

The inverse frequency linear prediction performing unit 950 synthesizes the output of the multi-resolution synthesis unit 940 and the result of performing the frequency linear prediction in the coding stage received from the demultiplexing unit 900 and performs inverse-vector quantization. . In detail, the inverse frequency linear prediction performing unit 950 outputs the result of the frequency linear prediction when the audio signal is encoded in the encoding stage, or outputs the multi-resolution synthesis of the inverse quantization unit 930. The efficiency of decoding may be improved by combining with the output of the unit 940. Here, the inverse frequency linear prediction performing unit 950 effectively improves coding efficiency by employing a frequency domain prediction technique and a vector quantization technique of prediction coefficients. The inverse frequency linear prediction performing unit 950 receives a difference spectrum coefficient and an index of a vector and outputs an MDCT spectrum coefficient.

The bandwidth extension decoder 960 decodes the encoded bandwidth extension information and uses the decoded bandwidth extension information to output the low frequency band signal output from the multi-resolution synthesis unit 940 or the inverse frequency linear prediction performer 950. Generate a high frequency band signal. Here, the bandwidth extension decoder 960 generates the high frequency band signal by applying the decoded bandwidth extension information to the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal. Here, the bandwidth extension information is information that can indicate the characteristics of the high frequency band of the high frequency band and is information such as energy level or envelope of the high frequency band signal.

The band synthesis unit 970 synthesizes the low frequency band signal output from the multi-resolution synthesis unit 940 or the inverse frequency linear prediction execution unit 950 and the high frequency band signal decoded by the bandwidth extension decoder 960.

The inverse MDCT application unit 980 inversely converts the signal output from the band synthesis unit 970 from the frequency domain to the time domain by an inverse MDCT and outputs it through the output terminal OUT. Here, the inverse MDCT application unit 980 receives the frequency spectrum coefficients obtained as the result of inverse quantization by the inverse frequency linear prediction performing unit 950 and outputs the recovered audio data corresponding to the low frequency band.

10 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 10, an apparatus for decoding an audio signal includes a demultiplexer 1000, a context-based bitplane decoder 1010, a dequantizer 1020, a multi-resolution synthesizer 1030, and inverse frequency linear prediction. An execution unit 1040, a bandwidth extension decoding unit 1050, a first inverse MDCT application unit 1060, a second inverse MDCT application unit 1070, and a band synthesis unit 1080 are included.

The demultiplexer 1000 demultiplexes a bit stream output from an encoding stage. Specifically, the demultiplexer 1000 separates each part of the data level into data parts corresponding to each unit, and analyzes and outputs information of a bitstream related thereto in the corresponding unit. In this case, the information output from the demultiplexer 1000 includes an explanatory analysis of the audio spectrum stream, quantization values and other reconstruction information, reconstruction information of the quantization spectrum, and context-dependent bit plane decoding without. information of decoding, signal type information, information of frequency-linear prediction and vector quantization, and coded bandwidth extension information.

The context-based bitplane decoder 1010 decodes the encoded bitplane based on the context. Here, the context-based bitplane decoder 1010 receives the information output from the demultiplexer 1000 and performs Huffman decoding to perform frequency spectrum, coding band mode information, and scale. Restore the scale factor. The context-based bitplane decoder 1010 receives prejudice coding band mode information, a scale factor of prejudice coding, and a frequency spectrum of prejudice coding, and displays a coding band mode value, a decoding cosmetic display of a scale factor, and a frequency spectrum. Output the quantization value of.

The inverse quantizer 1020 dequantizes the result output from the context-based bitplane decoder 1010.

The multi-resolution synthesis unit 1030 receives the dequantized signal from the inverse quantization unit 1020 and processes multi-resolution of audio spectral coefficients of a signal that is instantaneously changed. In detail, the multi-resolution synthesis unit 1030 may improve the decoding efficiency by synthesizing the output of the inverse quantizer 1020 into multi-resolution when the audio signal is analyzed as multi-resolution at the encoding end. Here, the multi-resolution synthesis unit 1030 receives a reserve quantization spectrum / difference spectrum and outputs a reconstruction spectrum / difference spectrum.

The inverse frequency linear prediction execution unit 1040 synthesizes the output of the multi-resolution synthesis unit 1030 and the result of performing the frequency linear prediction in the coding stage received from the demultiplexer 1000. In detail, the inverse frequency linear prediction execution unit 1040 outputs the result of the frequency linear prediction when the audio signal is encoded in the encoding stage, or outputs the inverse quantization unit 1020 or the multi-resolution synthesis unit. Combined with the output of 1030, the efficiency of copying can be improved. Here, the inverse frequency linear prediction execution unit 1040 effectively improves coding efficiency by employing a frequency domain prediction technique and a vector quantization technique of prediction coefficients. The inverse frequency linear prediction execution unit 1040 receives a difference spectrum coefficient and an index of a vector and outputs an MDCT spectrum coefficient.

As such, the context-based bitplane decoder 1010, the inverse quantizer 1020, the multi-resolution synthesizer 1030, and the inverse frequency linear prediction performer 1040 perform decoding on the encoded low frequency band signal. Therefore, it can be referred to as a low frequency band decoder.

The bandwidth extension decoder 1050 decodes the encoded bandwidth extension information and uses the decoded bandwidth extension information to output the low frequency band signal output from the multi-resolution synthesis unit 1030 or the inverse frequency linear prediction execution unit 1040. Generate a high frequency band signal. Here, the bandwidth extension decoder 1050 generates the high frequency band signal by applying the decoded bandwidth extension information to the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal. Here, the bandwidth extension information is information that can indicate the characteristics of the high frequency band of the high frequency band and is information such as energy level or envelope of the high frequency band signal.

The first inverse MDCT application unit 1060 performs an inverse MDCT on the low frequency band signal output from the multi-resolution synthesis unit 1030 or the inverse frequency linear prediction unit 1040 as an inverse transform process for the encoding end. Perform inverse transform from frequency domain to time domain. Here, the first inverse MDCT application unit 1060 receives a frequency spectrum coefficient obtained as a result of inverse quantization by the multi-resolution synthesis unit 1030 or the inverse frequency linear prediction execution unit 1040, and restores the frequency spectrum coefficient corresponding to the low frequency band. Output as audio data.

The second inverse MDCT application unit 1070 inversely converts the high frequency band signal generated by the bandwidth extension decoding unit 1050 from the frequency domain to the time domain by inverse MDCT.

The band synthesizing unit 1080 synthesizes the low frequency band signal converted into the time domain in the first inverse MDCT applying unit 1060 and the high frequency band signal converted into the time domain in the second inverse MDCT applying unit 1070 and outputs the terminal OUT. Output through

11 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 11, the audio data decoding apparatus performs demultiplexer 1100, context-based bitplane decoder 1110, dequantizer 1120, multi-resolution synthesizer 1130, and inverse frequency linear prediction. A unit 1140, an inverse MDCT application unit 1150, a transform unit 1160, a bandwidth extension decoder 1170, an inverse transform unit 1180, and a band synthesizer 1190 are included.

The demultiplexer 1100 receives a bit stream output from an encoding stage and demultiplexes the bit stream. The demultiplexer 1100 divides each part of the data level into data parts corresponding to each unit, and analyzes and outputs information of a bitstream related thereto in the corresponding unit. In this case, the information output from the demultiplexer 1100 may include descriptive analysis of the audio spectrum stream, quantization values and other reconstruction information, reconstruction information of the quantization spectrum, and context-dependent bit plane decoding without. information of decoding, signal type information, information of frequency-linear prediction and vector quantization, and coded bandwidth extension information.

The context-based bitplane decoder 1110 decodes the bitplane encoded based on the context. Here, the context-based bitplane decoder 1110 receives the information output from the demultiplexer 1100 and performs Huffman decoding to perform frequency spectrum, coding band mode information, and Restore the scale factor. The context-based bitplane decoder 1110 receives prejudice coding band mode information, a scale factor of prejudice coding, and a frequency spectrum of prejudice coding, and receives a coding band mode value, a decoding cosmetic display of a scale factor, and a quantization value of a frequency spectrum. Output

The dequantizer 1120 dequantizes the result output from the context-based bitplane decoder 1110.

The multi-resolution synthesis unit 1130 receives the dequantized signal from the inverse quantization unit 1120 and processes multi-resolution with respect to the audio spectral coefficients of a signal that is instantaneously changed. In detail, the multi-resolution synthesis unit 1130 may improve the decoding efficiency by synthesizing the output of the inverse quantization unit 1120 into a multi-resolution when the audio signal is analyzed as a multi-resolution at the encoding end. Here, the multi-resolution synthesis unit 1130 receives a reserve quantization spectrum / difference spectrum and outputs a reconstruction spectrum / difference spectrum.

The inverse frequency linear prediction performing unit 1140 synthesizes the result output from the multi-resolution synthesis unit 1130 and the result of performing the frequency linear prediction in the encoding stage from the demultiplexing unit 1100 and performs inverse-vector quantization. . In detail, the inverse frequency linear prediction performing unit 1140 outputs the result of the frequency linear prediction when the audio signal is performed at the encoding end by the inverse quantization unit 1120 or the multi-resolution synthesis unit. The efficiency of the decoding may be improved by combining with the output of 1130. Here, the inverse frequency linear prediction performing unit 1140 effectively improves coding efficiency by employing a frequency domain prediction technique and a vector quantization technique of prediction coefficients. The inverse frequency linear prediction performing unit 1140 receives a difference spectrum coefficient and an index of a vector and outputs an MDCT spectrum coefficient.

As such, the context-based bitplane decoder 1110, the inverse quantizer 1120, the multi-resolution synthesizer 1130, and the inverse frequency linear prediction performer 1140 decode the encoded low frequency band signal. Therefore, it can be referred to as a low frequency band decoder.

The inverse MDCT application unit 1150 converts the low frequency band signal output from the multi-resolution synthesis unit 1130 or the inverse frequency linear prediction unit 1140 from the frequency domain to the time domain by inverse MDCT as an inverse transform process for the encoding end. Invert Here, the inverse MDCT application unit 1150 receives the frequency spectrum coefficients obtained as a result of inverse quantization by the multi-resolution synthesis unit 1130 or the inverse frequency linear prediction execution unit 1140, and restores the audio data corresponding to the low frequency band. Will output

The transform unit 1160 converts the low frequency band signal converted into the time domain in the inverse MDCT application unit 1150 into the frequency domain or the time / frequency domain using a conversion technique other than MDCT. Examples of the conversion technique used by the transform unit 1160 include MDST, FFT, and QMF. Although the transformation unit 1160 may not implement and apply MDCT, the embodiment shown in FIG. 10 is more efficient when using MDCT.

The bandwidth extension decoder 1170 decodes the encoded bandwidth extension information and generates a high frequency band signal from the low frequency band signal output from the converter 1160 using the decoded bandwidth extension information. Here, the bandwidth extension decoder 1170 generates the high frequency band signal by applying the decoded bandwidth extension information to the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal. Here, the bandwidth extension information is information that can indicate the characteristics of the high frequency band of the high frequency band and is information such as energy level or envelope of the high frequency band signal.

The inverse transformer 1180 inversely transforms the high frequency band signal decoded by the bandwidth extension decoder 1170 from the frequency domain or the time-frequency domain to the time domain using a conversion technique other than MDCT. Here, the inverse transform unit 1180 uses the same transform technique used by the transform unit 1160. Examples of the conversion technique used by the inverse transform unit 1180 include MDST, FFT, and QMF.

The band synthesizer 1190 synthesizes the low frequency band signal converted into the time domain in the inverse MDCT application unit 1150 and the high frequency band signal converted into the time domain in the inverse transform unit 1180 and outputs the same through the output terminal OUT.

12 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 12, the audio data decoding apparatus performs demultiplexer 1200, context-based bitplane decoder 1210, dequantizer 1220, multi-resolution synthesizer 1230, and inverse frequency linear prediction. A unit 1240, a bandwidth extension decoder 1250, a band synthesizer 1260, and an inverse MDCT application unit 1270 are included.

The demultiplexer 1200 demultiplexes a bit stream output from an encoding stage. The demultiplexer 1200 divides each part of the data level into data parts corresponding to each unit, and analyzes and outputs information of a bitstream related thereto in the corresponding unit. In this case, the information output from the demultiplexer 1200 includes an explanatory analysis of the audio spectrum stream, quantization values and other reconstruction information, reconstruction information of the quantization spectrum, and context-dependent bit plane decoding without information of decoding, signal type information, information of frequency-linear prediction and vector quantization, and coded bandwidth extension information.

The context-based bitplane decoder 1210 decodes the bitplane encoded based on the context. Here, the context-based bitplane decoder 1210 receives the information output from the demultiplexer 1200 and performs Huffman decoding to perform frequency spectrum, coding band mode information, and Restore the scale factor. The context-based bitplane decoder 1210 receives prejudice coding band mode information, a scale factor of prejudice coding, and a frequency spectrum of prejudice coding, and receives a coding band mode value, a decoding cosmetic display of a scale factor, and a quantization value of a frequency spectrum. Output

The dequantizer 1220 dequantizes the result output from the context-based bitplane decoder 1210.

The multi-resolution synthesis unit 1230 receives the dequantized result from the inverse quantization unit 1220 and processes multi-resolution of the audio spectral coefficients of a signal that is instantaneously changed. In detail, the multi-resolution synthesis unit 1230 may improve the decoding efficiency by synthesizing the output of the dequantization unit 1220 into multi-resolution when the audio signal is analyzed as multi-resolution at the encoding end. Here, the multi-resolution synthesis unit 1230 receives a reserve quantization spectrum / difference spectrum and outputs a reconstruction spectrum / difference spectrum.

The inverse frequency linear prediction performing unit 1240 synthesizes the result output from the multi-resolution synthesis unit 1230 and the result of performing the frequency linear prediction in the coding stage input from the demultiplexing unit 1200 and performs inverse-vector quantization. Perform. In detail, the inverse frequency linear prediction execution unit 1240 outputs the result of the frequency linear prediction when the audio signal is encoded in the encoding stage by the inverse quantization unit 1220 or the multi-resolution synthesis unit. The efficiency of decoding can be improved by combining with the output of 1230. Here, the inverse frequency linear prediction performing unit 1240 effectively improves coding efficiency by employing a frequency domain prediction technique and a vector quantization technique of prediction coefficients. The inverse frequency linear prediction performing unit 1240 receives a difference spectrum coefficient and an index of a vector and outputs an MDCT spectrum coefficient.

As such, the context-based bitplane decoder 1210, the inverse quantizer 1220, the multi-resolution synthesis unit 1230, and the inverse frequency linear prediction performer 1240 perform decoding on the encoded low frequency band signal. Therefore, it can be referred to as a low frequency band decoder.

The bandwidth extension decoder 1250 decodes the encoded bandwidth extension information and uses the decoded bandwidth extension information from the low frequency band signal output from the multi-resolution synthesis unit 1230 or the inverse frequency linear prediction performer 1240. Generate a high frequency band signal. Here, the bandwidth extension decoder 1250 generates the high frequency band signal by applying the decoded bandwidth extension information to the low frequency band signal based on the property that there is a high correlation between the high frequency and the low frequency band of the audio signal. Here, the bandwidth extension information is information that can indicate the characteristics of the high frequency band of the high frequency band and is information such as energy level or envelope of the high frequency band signal.

The band synthesizer 1260 synthesizes the low frequency band signal output from the multi-resolution synthesis unit 1230 or the inverse frequency linear prediction performer 1240 and the high frequency band signal generated by the bandwidth extension decoder 1250.

The inverse MDCT application unit 1270 converts the signal output from the band synthesizing unit 1260 from the frequency domain to the time domain by inverse MDCT and outputs it through the output terminal OUT. Here, the inverse MDCT application unit 1270 receives the frequency spectrum coefficients obtained as the result of inverse quantization by the inverse frequency linear prediction performing unit 1240 and outputs the restored audio data corresponding to the low frequency band.

13 is a flowchart illustrating a method of encoding an audio signal according to an embodiment of the present invention.

Referring to FIG. 13, the audio signal encoding method according to the present embodiment includes steps that are processed in time series in the audio signal encoding apparatus shown in FIG. 4. Therefore, even if omitted below, the above description of the audio signal encoding apparatus shown in FIG. 4 is also applied to the audio signal encoding method according to the present embodiment.

In operation 1300, the band divider 400 divides the input signal into a high frequency band signal and a low frequency band signal.

In operation 1310, the first and second MDCT applying units 410 and 460 convert the high frequency band signal and the low frequency band signal from the time domain to the frequency domain, respectively. Herein, the first and second MDCT applying units 410 and 460 may perform a Modified Discrete Cosine Transform (MDCT) on the high frequency band signal and the low frequency band signal to convert the time domain into the frequency domain.

In operation 1320, the low frequency band encoder quantizes the converted low frequency band signal and encodes the bit plane based on a context. Here, the low frequency band coder may include a frequency linear prediction performer 420, a multi-resolution analyzer 430, a quantizer 440, and a context-based bitplane encoder 450. In detail, the frequency linear prediction performing unit 420 performs the filtering by performing the frequency linear prediction according to the characteristics of the converted low frequency band signal, and the multi-resolution analyzer 430 performs the characteristics of the converted low frequency band signal or the filtered signal. According to the multi-resolution analysis. In addition, the quantization unit 440 quantizes the signal analyzed by the multi-resolution, and the context-based bitplane encoder 450 encodes the quantized signal into the bitplane based on the context.

In operation 1330, the bandwidth extension encoder 470 generates and encodes bandwidth extension information indicating a characteristic of the converted high frequency band signal using the converted low frequency band signal.

In operation 1340, the multiplexer 480 multiplexes the encoded bitplane and the encoded bandwidth extension information and outputs the result of encoding the input signal.

14 is a flowchart illustrating a method of encoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 14, the method of encoding an audio signal according to the present embodiment includes the steps of time-series processing in the apparatus for encoding an audio signal illustrated in FIG. 5. Therefore, even if omitted below, the above description of the audio signal encoding apparatus shown in FIG. 5 is also applied to the audio signal encoding method according to the present embodiment.

In operation 1400, the band divider 500 divides the input signal into a high frequency band signal and a low frequency band signal.

In step 1410, the MDCT application unit 510 performs an MDCT on the low frequency band signal to convert from the time domain to the frequency domain.

In step 1420, the low frequency band encoder quantizes the signal on which the MDCT is performed and encodes the bit plane based on the context. Here, the low frequency band encoder may include a frequency linear prediction performer 520, a multi-resolution analyzer 530, a quantizer 540, and a context-based bitplane encoder 550. Specifically, the frequency linear prediction performing unit 520 performs the filtering by performing the frequency linear prediction according to the characteristics of the converted low frequency band signal, and the multi-resolution analyzer 530 performs the characteristics of the converted low frequency band signal or the filtered signal. According to the multi-resolution analysis. In addition, the quantization unit 540 quantizes the signal analyzed by the multi-resolution, and the context-based bitplane encoder 550 encodes the quantized signal into the bitplane based on the context.

In operation 1430, the low frequency band converter 560 and the high frequency band converter 570 convert the high frequency band signal and the low frequency band signal from the time domain to the frequency domain or the time / frequency domain, respectively.

In operation 1440, the bandwidth extension encoder 580 generates and encodes bandwidth extension information indicating a characteristic of the converted high frequency band signal using the converted low frequency band signal.

In operation 1450, the multiplexer 590 multiplexes the encoded bitplane and the encoded bandwidth extension information and outputs the result of encoding the input signal.

15 is a flowchart illustrating a method of encoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 15, an audio signal encoding method according to the present embodiment includes steps that are processed in time series in an audio signal encoding apparatus of FIG. 6. Therefore, even if omitted below, the above description of the audio signal encoding apparatus shown in FIG. 6 is also applied to the audio signal encoding method according to the present embodiment.

In operation 1500, the MDCT applying unit 600 converts the input signal from the time domain to the frequency domain. In detail, the MDCT applying unit 600 may perform an MDCT on the input signal to convert the input signal from the time domain to the frequency domain.

In operation 1510, the band divider 610 divides the converted input signal into a high frequency band signal and a low frequency band signal.

In operation 1520, the low frequency band encoder quantizes the low frequency band signal and encodes the bit frequency based on the context. Here, the low frequency band encoder may include a frequency linear prediction performer 620, a multi-resolution analyzer 630, a quantizer 640, and a context-based bitplane encoder 650. Specifically, the frequency linear prediction performing unit 620 performs the frequency linear prediction according to the characteristics of the converted low frequency band signal, and filters the multi-resolution analyzer 630 of the converted low frequency band signal or the filtered signal. Depending on the characteristics, it is analyzed by multi-resolution. In addition, the quantization unit 640 quantizes the signal analyzed by the multi-resolution, and the context-based bitplane encoder 650 encodes the quantized signal into the bitplane based on the context.

In operation 1530, the bandwidth extension encoder 660 generates and encodes bandwidth extension information indicating characteristics of the high frequency band signal using the low frequency band signal.

In operation 1540, the multiplexer 670 multiplexes the encoded bitplane and the encoded bandwidth extension information and outputs the result of encoding the input signal.

16 is a flowchart illustrating a method of decoding an audio signal according to an embodiment of the present invention.

Referring to FIG. 16, an audio signal decoding method according to the present embodiment includes steps processed in time series in an audio signal decoding apparatus illustrated in FIG. 10. Therefore, even if omitted below, the above description of the audio signal decoding apparatus shown in FIG. 10 is also applied to the audio signal decoding method according to the present embodiment.

In operation 1600, the demultiplexer 1000 receives a result of encoding the audio signal. Here, the encoded result includes a bit plane encoded based on the context and encoded bandwidth extension information for the low frequency band.

In operation 1610, the low frequency band decoder generates a low frequency band signal by decoding and demultiplexing the encoded bitplane based on a context. Here, the low frequency band decoder may include a context-based bitplane decoder 1010, an inverse quantizer 1020, a multi-resolution synthesizer 1030, and an inverse frequency linear prediction performer 1040. In detail, the context-based bitplane decoder 1010 decodes the bitplane encoded based on the context with respect to the low frequency band signal, and the dequantizer 1020 dequantizes the decoded signal. In addition, when the multi-resolution analysis is performed in the encoding stage, the multi-resolution synthesis unit 1030 synthesizes the dequantized signal into multi-resolution, and the inverse frequency linear prediction execution unit 1040 performs frequency linear prediction in the encoding stage. In this case, the result of performing the frequency linear prediction in the encoding end using the vector index is synthesized into a quantized signal or a multi-resolution synthesized signal.

In operation 1620, the bandwidth extension decoder 1050 decodes the encoded bandwidth extension information and generates a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information.

In operation 1630, the first and second inverse MDCT applying units 1060 and 1070 perform inverse MDCT on each of the low frequency band signal and the high frequency band signal to convert from the frequency domain to the time domain.

In operation 1640, the band synthesis unit 1080 synthesizes the converted low frequency band signal and the converted high frequency band signal.

17 is a flowchart illustrating a method of decoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 17, an audio signal decoding method according to the present embodiment includes steps that are processed in time series in an audio signal decoding apparatus illustrated in FIG. 11. Therefore, even if omitted below, the above description of the audio signal decoding apparatus shown in FIG. 11 is also applied to the audio signal decoding method according to the present embodiment.

In operation 1700, the demultiplexer 1100 receives a result of encoding an audio signal. Here, the encoded result includes a bit plane encoded based on the context and encoded bandwidth extension information for the low frequency band.

In operation 1710, the low frequency band decoder generates a low frequency band signal by decoding and inverse quantizing the encoded bitplane based on a context. Here, the low frequency band decoder may include a context-based bitplane decoder 1110, an inverse quantizer 1120, a multi-resolution synthesizer 1130, and an inverse frequency linear prediction performer 1140. In detail, the context-based bitplane decoder 1110 decodes the bitplane encoded based on the context with respect to the low frequency band signal, and the inverse quantizer 1120 dequantizes the decoded signal. In addition, the multi-resolution synthesis unit 1130 synthesizes the dequantized signal into a multi-resolution when the multi-resolution analysis is performed at the encoding end. In addition, when frequency linear prediction is performed at the encoding end, the inverse frequency linear prediction performing unit 1140 may convert the result of frequency linear prediction at the encoding end into a dequantized signal or a multi-resolution synthesized signal using a vector index. Synthesize

In operation 1720, the inverse MDCT application unit 1150 performs inverse MDCT on the low frequency band signal to convert from the frequency domain to the time domain.

In operation 1730, the converter 1160 converts the low frequency band signal in which the inverse MDCT is performed into the frequency domain or the time / frequency domain.

In operation 1740, the bandwidth extension decoder 1170 decodes the encoded bandwidth extension information and generates a high frequency band signal from the low frequency band signal converted into the frequency domain or the time / frequency domain using the decoded bandwidth extension information.

In operation 1750, the inverse transformer 1180 inversely converts the high frequency band signal into the time domain.

In operation 1760, the band synthesizer 1190 synthesizes the inversely transformed high frequency band signal and the converted low frequency band signal.

18 is a flowchart illustrating a method of decoding an audio signal according to another embodiment of the present invention.

Referring to FIG. 18, an audio signal decoding method according to the present embodiment includes steps processed in time series in an audio signal decoding apparatus shown in FIG. 12. Therefore, even if omitted below, the above description of the audio signal decoding apparatus shown in FIG. 12 is also applied to the audio signal decoding method according to the present embodiment.

In operation 1800, the demultiplexer 1200 receives a result of encoding an audio signal. Here, the encoded result includes a bit plane encoded based on the context and encoded bandwidth extension information for the low frequency band.

In step 1810, the low frequency band decoder generates a low frequency band signal by decoding and inverse quantizing the encoded bitplane based on the context. Here, the low frequency band decoder may include a context-based bitplane decoder 1210, an inverse quantizer 1220, a multi-resolution synthesizer 1230, and an inverse frequency linear prediction performer 1240. In detail, the context-based bitplane decoder 1210 decodes the bitplane encoded based on the context with respect to the low frequency band signal, and the dequantizer 1220 dequantizes the decoded signal. In addition, the multi-resolution synthesis unit 1230 synthesizes the dequantized signal into a multi-resolution when the multi-resolution analysis is performed at the encoding end. In addition, when frequency linear prediction is performed at the encoding end, the inverse frequency linear prediction performing unit 1240 may convert the result of frequency linear prediction at the encoding end to a dequantized signal or a multi-resolution synthesized signal using a vector index. Synthesize

In operation 1820, the bandwidth extension decoder 1250 decodes the encoded bandwidth extension information and generates a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information.

In operation 1830, the band synthesizer 1260 synthesizes the low frequency band signal and the high frequency band signal.

In operation 1840, the inverse MDCT applying unit 1270 performs inverse MDCT on the synthesized signal and converts the frequency domain from the frequency domain to the time domain.

The present invention is not limited to the above-described embodiment, and of course, modifications may be made by those skilled in the art within the spirit of the present invention.

The invention can also be embodied as computer readable code on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, hard disk, floppy disk, flash memory, optical data storage device, and also carrier waves (for example, transmission over the Internet). It also includes the implementation in the form of. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

1 is a block diagram illustrating an apparatus for encoding an audio signal according to an embodiment of the present invention.

2 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

3 is a block diagram showing an apparatus for encoding an audio signal according to another embodiment of the present invention.

4 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

5 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

6 is a block diagram illustrating an apparatus for encoding an audio signal according to another embodiment of the present invention.

7 is a block diagram illustrating an apparatus for decoding an audio signal according to an embodiment of the present invention.

8 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

9 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

10 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

11 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

12 is a block diagram illustrating an apparatus for decoding an audio signal according to another embodiment of the present invention.

13 is a flowchart illustrating a method of encoding an audio signal according to an embodiment of the present invention.

14 is a flowchart illustrating a method of encoding an audio signal according to another embodiment of the present invention.

15 is a flowchart illustrating a method of encoding an audio signal according to another embodiment of the present invention.

16 is a flowchart illustrating a method of decoding an audio signal according to an embodiment of the present invention.

17 is a flowchart illustrating a method of decoding an audio signal according to another embodiment of the present invention.

18 is a flowchart illustrating a method of decoding an audio signal according to another embodiment of the present invention.

Claims (43)

(a) dividing an input signal into a high frequency band signal and a low frequency band signal; (b) converting the high frequency band signal and the low frequency band signal from the time domain to the frequency domain, respectively; (c) quantizing the converted low frequency band signal and encoding the bitplane based on a context; (d) generating and encoding bandwidth extension information indicating a characteristic of the converted high frequency band signal using the converted low frequency band signal; And and (e) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal. The method of claim 1, Step (b) is And encoding the high frequency band signal and the low frequency band signal by performing a Modified Discrete Cosine Transform (MDCT) to transform them from the time domain to the frequency domain. The method of claim 1, (f) performing frequency linear prediction on the transformed low frequency band signal and filtering the result; And (g) further analyzing at least one of the converted low frequency band signals with multi-resolution, Step (c) is And encoding the filtered signal or the signal analyzed by the multi-resolution by bit quantization based on a context. The method of claim 3, Step (f) A coefficient of a linear prediction filter is calculated by performing frequency linear prediction on the transformed low frequency band signal, and a value corresponding to the coefficient is expressed as a vector index. Step (e) is And encoding the encoded bitplane, the encoded bandwidth extension information, and the vector index as an encoding result of the input signal. A computer-readable recording medium having recorded thereon a program for executing an audio signal encoding method according to any one of claims 1 to 4. (a) dividing an input signal into a high frequency band signal and a low frequency band signal; (b) performing MDCT on the low frequency band signal and converting from the time domain to the frequency domain; (c) quantizing the signal on which the MDCT is performed and encoding the bitplane based on a context; (d) converting the high frequency band signal and the low frequency band signal from time domain to frequency domain or time / frequency domain, respectively; (e) generating and encoding bandwidth extension information representing a characteristic of the converted high frequency band signal using the converted low frequency band signal; And (f) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal. The method of claim 6, Step (c) is (g) performing frequency linear prediction on the signal on which the MDCT is performed and filtering the signal; And (h) further analyzing at least one step of analyzing the signal on which the MDCT has been performed by multi-resolution, Step (c) is And encoding the filtered signal or the signal analyzed by the multi-resolution to a bit plane based on a context. The method of claim 7, wherein Step (g) A coefficient of a linear prediction filter is calculated by performing frequency linear prediction on the low frequency band signal on which the MDCT is performed, and a value corresponding to the coefficient is expressed as a vector index. Step (f) And encoding the encoded bitplane, the encoded bandwidth extension information, and the vector index as an encoding result of the input signal. A computer-readable recording medium having recorded thereon a program for executing the method of encoding an audio signal according to any one of claims 6 to 8. (a) converting an input signal from time domain to frequency domain; (b) dividing the converted input signal into a high frequency band signal and a low frequency band signal; (c) quantizing the low frequency band signal and encoding the bitplane based on a context; (d) generating and encoding bandwidth extension information representing the characteristics of the high frequency band signal using the low frequency band signal; And and (e) outputting the encoded bitplane and the encoded bandwidth extension information as an encoding result of the input signal. The method of claim 10, Step (a) is And performing an MDCT on the input signal to convert the input signal from a time domain to a frequency domain. The method of claim 10, (f) performing frequency linear prediction on the low frequency band signal and filtering the result; And (g) further analyzing at least one of the low frequency band signals by multi-resolution, Step (c) is And encoding the filtered signal or the signal analyzed by the multi-resolution by bit quantization based on a context. The method of claim 12, Step (f) Calculating a coefficient of a linear prediction filter by performing frequency linear prediction on the low frequency band signal, and expressing a value corresponding to the coefficient as a vector index, Step (e) is And encoding the encoded bitplane, the encoded bandwidth extension information, and the vector index as an encoding result of the input signal. A computer-readable recording medium having recorded thereon a program for executing an audio signal encoding method according to any one of claims 10 to 13. (a) receiving an encoding result of an audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; (d) performing an Inverse Modified Discrete Cosine Transform (MDCT) on each of the low frequency band signal and the high frequency band signal to convert from the frequency domain to the time domain; And (e) synthesizing the converted low frequency band signal and the converted high frequency band signal. The method of claim 15, Step (b) is Synthesizing the dequantized signal into a multi-resolution; And And synthesizing a result of the frequency linear prediction performed by the encoding end into the dequantized signal by using the vector index included in the encoding result. A computer-readable recording medium having recorded thereon a program for executing a method for decoding an audio signal according to any one of claims 15 and 16. (a) receiving an encoding result of an audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) performing inverse MDCT on the low frequency band signal and converting from the frequency domain to the time domain; (d) converting the low frequency band signal subjected to the inverse MDCT into a frequency domain or a time / frequency domain; (e) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal converted into the frequency domain or the time / frequency domain using the decoded bandwidth extension information; (f) inversely converting the high frequency band signal into a time domain; And and (g) synthesizing the inverse transformed high frequency band signal and the transformed low frequency band signal. The method of claim 18, Step (b) is Synthesizing the dequantized signal into a multi-resolution; And And synthesizing a result of performing frequency linear prediction at the encoding end into the dequantized signal by using the vector index included in the decoding result. A computer-readable recording medium having recorded thereon a program for executing the method for decoding an audio signal according to any one of claims 18 and 19. (a) receiving an encoding result of an audio signal; (b) generating a low frequency band signal by decoding and inverse quantizing the encoded bitplane included in the encoding result based on a context; (c) decoding the encoded bandwidth extension information included in the encoding result and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; (d) synthesizing the low frequency band signal and the high frequency band signal; And (e) performing an inverse MDCT on the synthesized signal and converting from the frequency domain to the time domain. The method of claim 21, Step (b) is Synthesizing the dequantized signal into a multi-resolution; And And synthesizing a result of the frequency linear prediction performed by the encoding end into the dequantized signal by using the vector index included in the encoding result. A computer-readable recording medium having recorded thereon a program for executing the method for decoding an audio signal according to any one of claims 21 and 22. A band dividing unit dividing the input signal into a high frequency band signal and a low frequency band signal; A converter for converting the high frequency band signal and the low frequency band signal from a time domain to a frequency domain, respectively; A low frequency band encoder which quantizes the converted low frequency band signal and encodes the bit frequency into a bit plane based on a context; And And a bandwidth extension encoder configured to generate and encode bandwidth extension information indicating a characteristic of the converted high frequency band signal by using the converted low frequency band signal. The method of claim 24, The conversion unit A first MDCT performing unit performing an MDCT on the high frequency band signal and converting from the time domain to the frequency domain; And And a second MDCT performing unit performing an MDCT on the low frequency band signal and converting from the time domain to the frequency domain. The method of claim 24, The low frequency band encoder A frequency linear prediction performing unit performing frequency linear prediction on the converted low frequency band signal to filter the converted low frequency band signal; And Further comprising at least one of a multi-resolution analysis unit for analyzing the converted low frequency band signal by a multi-resolution, And encoding the filtered signal or the signal analyzed by the multi-resolution to a bit plane based on a context. The method of claim 26, The frequency linear prediction performing unit performs frequency linear prediction on the transformed low frequency band signal to calculate coefficients of a linear prediction filter, and expresses a value corresponding to the coefficients as a vector index. Chi. The method of claim 27, And a multiplexer which multiplexes the encoded bitplane, the encoded bandwidth extension information, and the vector index. A band dividing unit dividing the input signal into a high frequency band signal and a low frequency band signal; An MDCT application unit performing an MDCT on the low frequency band signal and converting the time domain from the time domain to the frequency domain; A low frequency band encoder for quantizing the output of the MDCT applier and encoding the bitplane based on a context; A converter for converting the high frequency band signal and the low frequency band signal from a time domain into a frequency domain or a time / frequency domain; And And a bandwidth extension encoder configured to generate and encode bandwidth extension information indicating a characteristic of the converted high frequency band signal by using the converted low frequency band signal. The method of claim 29, The low frequency band encoder A frequency linear prediction performing unit performing frequency linear prediction on the output of the MDCT application unit and filtering the output; And Further comprising at least one of a multi-resolution analysis unit for analyzing the output of the MDCT application unit with a multi-resolution, And encoding the filtered signal or the signal analyzed by the multi-resolution to a bit plane based on a context. The method of claim 30, The frequency linear prediction performing unit calculates a coefficient of the linear prediction filter by performing frequency linear prediction on the low frequency band signal on which the MDCT is performed, and expresses a value corresponding to the coefficient as a vector index. Encoding device. The method of claim 31, wherein And encoding multiplexing the encoded bitplane, the encoded bandwidth extension information, and the vector index. A converter for converting an input signal from a time domain to a frequency domain; A band dividing unit dividing the converted input signal into a high frequency band signal and a low frequency band signal; A low frequency band encoder for quantizing the low frequency band signal and encoding the bit frequency based on a context; And And a bandwidth extension encoder configured to generate and encode bandwidth extension information indicating a characteristic of the high frequency band signal by using the low frequency band signal. The method of claim 33, wherein And the conversion unit performs an MDCT on the input signal to convert the input signal from the time domain to the frequency domain. The method of claim 33, wherein The low frequency band encoder A frequency linear prediction performing unit performing filtering on the low frequency band signal by performing frequency linear prediction; And Further comprising at least one of a multi-resolution analysis unit for analyzing the low-frequency band signal in a multi-resolution, And encoding the filtered signal or the signal analyzed by the multi-resolution to a bit plane based on a context. 36. The method of claim 35 wherein The frequency linear prediction performing unit performs frequency linear prediction on the low frequency band signal to calculate coefficients of a linear prediction filter, and expresses a value corresponding to the coefficients as a vector index. The method of claim 36, And encoding multiplexing the encoded bitplane, the encoded bandwidth extension information, and the vector index. A low frequency band decoder configured to decode and dequantize the encoded bitplane based on a context to generate a low frequency band signal; A bandwidth extension decoder for decoding encoded bandwidth extension information and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; An inverse MDCT performing unit performing inverse MDCT on each of the low frequency band signal and the high frequency band signal and converting from the frequency domain to the time domain; And And a band synthesizer for synthesizing the converted low frequency band signal and the converted high frequency band signal. The method of claim 38, The low frequency band decoder A multi-resolution synthesis unit for synthesizing the dequantized signal into a multi-resolution; And And at least one of an inverse frequency linear prediction performing unit configured to synthesize the result of the frequency linear prediction performed by the encoding end using the vector index to the dequantized signal. A low frequency band decoder configured to decode and dequantize the encoded bitplane based on a context to generate a low frequency band signal; An inverse MDCT application unit performing inverse MDCT on the low frequency band signal and converting from the frequency domain to the time domain; A converter for converting the low frequency band signal subjected to the inverse MDCT to a frequency domain or a time / frequency domain; A bandwidth extension decoding unit for decoding encoded bandwidth extension information and generating a high frequency band signal from the low frequency band signal converted into the frequency domain or the time / frequency domain using the decoded bandwidth extension information; An inverse transformer for inversely converting the generated high frequency band signal into a time domain; And And a band synthesizer configured to synthesize the inverse transformed high frequency band signal and the converted low frequency band signal. The method of claim 40, The low frequency band decoder A multi-resolution synthesis unit for synthesizing the dequantized signal into a multi-resolution; And And at least one of an inverse frequency linear prediction performing unit for synthesizing a result of the frequency linear prediction performed by the encoding end using the vector index to the dequantized signal. A low frequency band decoder configured to decode and dequantize the encoded bitplane based on a context to generate a low frequency band signal; A bandwidth extension decoder for decoding encoded bandwidth extension information and generating a high frequency band signal from the low frequency band signal using the decoded bandwidth extension information; A band synthesizer for synthesizing the low frequency band signal and the high frequency band signal; And And an inverse MDCT applying unit for performing inverse MDCT on the synthesized signal and converting the frequency signal from the frequency domain to the time domain. The method of claim 42, wherein The low frequency band decoder A multi-resolution synthesis unit for synthesizing the dequantized signal into a multi-resolution; And And at least one of an inverse frequency linear prediction performing unit for synthesizing a result of the frequency linear prediction performed by the encoding end using the vector index to the dequantized signal.

Images