KR100895100B1 - Method and device for decoding digital audio data - Google Patents

Abstract

The present invention relates to a method and apparatus for decoding digital audio data, wherein an inverse discrete cosine transformer (IMDCT) of a filter bank having the longest clock occupancy time and a clock occupancy time (that is, a clock requirement) of one frame period for decoding are one. Parallelize the Huffman decoder that has not been settled. It also parallelizes the window / overlap adder and dequantizer of the filterbank. In addition, the present invention implements, in hardware, a computing unit shared by a plurality of memories operating in parallel, a dequantizer, and a filter bank. The data processing time of the entire decoding apparatus is shortened, and the performance of the entire decoding system can be improved due to the efficient distribution and use of the clock and can be applied to the portable apparatus.

Description

A method for decoding digital audio data and a device for decoding digital audio data {METHOD AND DEVICE FOR DECODING DIGITAL AUDIO DATA}

1 is a schematic block diagram of a conventional digital audio data decoding apparatus.

2 is an operation timing diagram of a conventional digital audio data decoding apparatus.

3 is a schematic block diagram of a digital audio data decoding apparatus according to an embodiment of the present invention.

4 is a flowchart illustrating a digital audio data decoding method according to an embodiment of the present invention.

5 is an operation timing diagram of a digital audio data decoding apparatus according to an embodiment of the present invention.

<Explanation of symbols for the main parts of the drawings>

110, 210: Input buffer 120, 220: Deformatter

130, 230: Huffman decoder 140, 240: dequantizer

150, 250: filter bank 260: control unit

270: storage unit 280: arithmetic unit (ALU)

The present invention relates to a method for decoding digital audio data and a device for decoding digital audio data, and more particularly, to a method and apparatus for decoding digital audio data applicable to a portable terminal.

Recently, only digital audio capable of playing digital audio such as MPEG1 Audio Layer-3 (MP3), Dolby AC-3 (Audio Coding-3), Ogg Vorvis, WMA (abbreviated as "WMA") Portable terminals equipped with devices or digital audio playback functions have been rapidly spread, allowing users to listen to music conveniently at any time and place.

Digital audio has the advantage of being able to reduce the file size to less than 1/10 while minimizing the degradation of audio quality by compressing using a perceptual coding method based on human cognitive ability.

Among digital audio, WMA is an audio compression format developed by Microsoft that can support streaming and reduce file size to half MP3 while maintaining audio quality equivalent to MP3.

In general, files in WMA format are decoded by a software method through a program called Windows Media Player, also produced by Microsoft Corporation, and output as sound.

1 is a schematic block diagram of a conventional digital audio data decoding apparatus. Referring to FIG. 1, a conventional digital audio data decoding apparatus 100 includes an input buffer 110, a de-formatter 120, a Huffman decoder 130, and an inverse quantizer 140. ), A filter bank (150), a control unit (160), and a storage unit (170) including an Inverse Modified Cosine Transform (IMDCT) 151 and a Window / Overlap-Add (155).

The storage unit 170 includes a ROM 1 171, a ROM 2 172, and a ROM 3 173. The conventional digital audio data decoding apparatus 100 receives an audio bit stream and temporarily stores the audio bit stream in the input buffer 110. The audio bitstream stored in the input buffer 110 is transmitted to the formatter 120. The deformatter 120 decodes and analyzes a header having information necessary for decoding the audio data received from the audio bitstream, and extracts the encoded data.

Meanwhile, the encoded data is transferred to the Huffman decoder 130 and decoded using the Huffman table, and then passed through a dequantizer 140, a filter bank 150, and a digital-to-analog converter (DAC, not shown). These are converted into audible audio signals and output.

ROM 1 171 stores a Huffman table, ROM 2 172 stores quantization coefficients, and ROM 3 173 stores window coefficients.

Conventionally, the input audio bit stream is decoded while sequentially passing through the Huffman decoder 130, the dequantizer 140, the IMDCT 151, and the window / overlap adder 155.

2 is a timing diagram of the conventional digital audio data decoding apparatus that performs the sequential processing.

In the example of FIG. 2, 'A' indicates a clock period occupied by the deformatter 120, and 'B' indicates a clock interval occupied by the Huffman decoder 130, and 'C' indicates inverse quantization. The instrument 140 represents a clock section occupied when the filter 140 operates, and 'D' represents a clock segment occupied by the filter bank 150.

Referring to FIG. 2, in the conventional digital audio data decoding apparatus, a clock included in one frame includes a deformatter A, a Huffman decoder B, an inverse quantizer C, and a filter bank D. Occupies sequentially. The fact that each deformatter (A), Huffman decoder (B), inverse quantizer (C) and filter bank (D) occupy a clock period wakes up at a predetermined time to wake up the predetermined clock period (t1, t2, t3, t4). ) Means to perform the operation. Therefore, referring to FIG. 2, the deformatter A, the Huffman decoder B, the inverse quantizer C, and the filter bank D included in the conventional digital audio data decoding apparatus have one operation. It can be seen that the following operation is performed in a sequential manner.

However, the Huffman decoder 130 does not have a constant clock requirement. This is because, in the case of Huffman decoding, the time to find a symbol depends on the length of the codeword of the input bitstream, and the codeword used for Huffman decoding is different every frame.

As such, since the clock requirement of the Huffman decoder 130 is not constant, the conventional digital audio decoder has a problem that it is difficult to fix the entire clock speed.

Therefore, the conventional digital audio decoder has a disadvantage in that it is difficult to predict the total clock cycle of the system.

In addition, in the related art, the algorithm has a complicated problem by implementing the process of FIG. 1 and FIG. 2 in software. That is, in order to output sound after decoding a file in WMA format using software, the sound output device equipped with the software should have a large storage space and maintain the CPU performance of a home computer. In order to satisfy the CPU performance and the storage space required to support the operation of the software, the size of the device for decoding the file in the WMA format must be increased.

Therefore, it is inappropriate to use the software to decrypt files in WMA format for portable devices (eg, mobile phones, etc.) that are following the miniaturization trend.

That is, decoding WMA format files using conventional software has a problem that it is difficult to apply to portable devices despite the superior decoding performance.

Accordingly, a first object of the present invention is to provide an apparatus for decoding digital audio data, which makes it possible to apply an audio file of WMA format to a portable device.

It is also a second object of the present invention to provide a method for decoding digital audio data, which makes it possible to apply an audio file of WMA format to a portable device.

According to an aspect of the present invention, a method of decoding digital audio data according to an aspect of the present invention is to perform digital processing of a next frame in parallel while performing inverse discrete cosine transform on dequantized data of a current frame. A first parallel processing step of extracting header information and coded data for audio data and decoding the coded data of the next frame, and the next frame in parallel while performing window / overlap-add of the current frame And a second parallel processing step of dequantizing the decoded data of. The first and second parallel processing steps may be repeatedly performed until the decoding of the digital audio data ends. The decoding may be Huffman decoding. The decoding method of the digital audio data includes initializing extracting header information and encoded data from the digital audio data of the first frame, decoding the encoded data of the first frame, and inversely quantizing the decoded data of the first frame. It may further comprise a step. The initialization may write the encoded data of the first frame to a first memory. The initializing step reads the encoded data of the first frame from the first memory, performs inverse quantization on the encoded data of the first frame, and then transfers the dequantized data of the first frame to a second memory. Can record The first parallel processing step may read inverse quantized data of the current frame from the second memory and perform inverse discrete cosine transform on the dequantized data of the current frame. In the first parallel processing step, the decoded data of the next frame may be recorded in the first memory. In the second parallel processing step, the windowed data of the previous frame is written to a third memory when the window / overlap-add is performed, and the windowed data of the previous frame is read from the third memory to read the current frame. It can overlap-add with the windowed data of. The digital audio data may be data encoded by window media audio.

In addition, the apparatus for decoding digital audio data according to an aspect of the present invention for achieving the second object of the present invention is a deformatting operation for extracting the header information and the encoded data from the first and second memories, the digital audio data A deformatter for performing a decoding operation, a decoder for decoding the encoded data, and storing the decoded data in the first memory, and reading and inversely quantizing the decoded data from the first memory. An inverse quantizer for performing inverse quantization operation stored in the second memory, an inverse discrete cosine converter for performing inverse discrete cosine transform operation by reading the inverse quantized data from the second memory, and performing window / overlap-add Filter banks that contain window / overlap-adds, and the inverse quantized data of the current frame. Perform the deformatting operation and the decoding operation on the digital audio data of the next frame in parallel while performing the inverse discrete cosine transform of the filter bank, and performing parallel operation during the window / overlap-add operation of the current frame. And a controller configured to perform the inverse quantization operation on the decoded data of the next frame. The apparatus may further include an operation unit configured to perform a complex / multiply / add operation for the inverse quantization and the inverse discrete cosine transform and shared by the inverse quantizer and the inverse discrete cosine transformer. The method may further include a third memory configured to store windowed data of a previous frame for window / overlap-add of the filter bank. The second memory may store windowed data of a previous frame for window overlap add of the filter bank. The decoder may be a Huffman decoder. The apparatus for decoding digital audio data may be used in a portable terminal.

In addition, the apparatus for decoding digital audio data according to another aspect of the present invention for achieving the second object of the present invention includes a deformatter for performing a deformatting operation for extracting the header information and the encoded data from the digital audio data; A decoder for performing a decoding operation for decoding coded data, an inverse quantizer for performing inverse quantization for inverse quantization of the decoded data, and an inverse discrete cosine transform operation for inverse discrete transforming the dequantized data. A filter bank comprising an inverse discrete cosine converter and a window / overlap-add that performs window / overlap-add, and in parallel during inverse discrete cosine transform of the filter bank on the dequantized data of the current frame. The deformatting operation and the decoding operation on the digital audio data of a frame. And to the control carried out, the window / overlap of said current frame with respect to the decoded data of the next frame in parallel for performing add operation and a control unit for controlling to perform the inverse quantization operation.

As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component. The term and / or includes a combination of a plurality of related items or any item of a plurality of related items.

The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting of the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof.

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 are not 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. Hereinafter, the same reference numerals are used for the same components in the drawings, and duplicate descriptions of the same components are omitted.

3 is a schematic block diagram of a digital audio data decoding apparatus according to an embodiment of the present invention. Referring to FIG. 3, the digital audio data decoding apparatus 200 according to an embodiment of the present invention includes an input buffer 210, a deformatter 220, a Huffman decoder 230, a dequantizer 240, and a filter bank. Reference numeral 250 includes an arithmetic-logic unit (ALU) 280, a controller 260, and a storage 270.

The input buffer 210 temporarily receives the input decoded data. The data to be decoded may be, for example, digital audio data having a bit stream form.

The controller 260 allows the bit stream type data (hereinafter, referred to as 'bit stream data') stored in the input buffer 210 to be transmitted to the deformatter 220 and / or the Huffman decoder 230.

The deformatter 220 performs a deformatting operation. Here, the deformatting operation means parsing a header having encoded information necessary for decoding from bit stream data transmitted through the input buffer 210 and encoded data (eg, Huffman encoded data). At this time, the deformatter 220 includes a first deformatter for processing an ASF (Advanced Stream Format) header, which holds information for decoding WMA data, and a second for processing a packet information header included in each packet. It may include a deformatter. In addition, the deformatter 220 extracts coding information (eg, lossy coding or lossless coding) and Huffman table number information of the bit stream data from the header and transfers the Huffman table number information to the Huffman decoder 230.

The Huffman decoder 230 RLC encodes coded data in the bit stream data transferred through the input buffer 210 based on the coding information transmitted from the deformatter 220 and the Huffman table number information, for example, Huffman data. (Run-Length decoding) or VLC (Variable-Length decoding) decoding. The memory ROM1 271 stores a plurality of Huffman tables, and the Huffman decoder 230 performs decoding by referring to the Huffman table corresponding to the Huffman table number information from ROM1 (Read Only Memory 1, 271). do. The Huffman decoder 230 stores the decoding result in random access memory 1 (274).

Inverse quantizer 240 dequantizes the data decoded by Huffman decoder 230. ROM2 272 stores exponential values used to perform inverse quantization. In particular, the inverse quantizer 240 reads the decoding result of the Huffman decoder 230 from the RAM1 274 under the control of the controller 260 and receives the number of entropy information and quantization coefficients provided from the deformatter 220. Inverse quantization using a and stores the result in the RAM2 (275).

The inverse quantizer 240 receives the exponential calculation result necessary for performing the inverse quantization from the calculator 280.

The operation unit 280 is shared by the inverse quantizer 240 and the IMDCT 251, and processes the exponential operation of the inverse quantizer 240 and the complex number / multiplication / add operation of the IMDCT 251 and dequantizes the result. To each of the groups 240 and IMDCT 251. The arithmetic units ALU 280 may be separated into separate hardware blocks, and the arithmetic operations performed by the inverse quantizer 240 and the IMDCT 251 may be performed by the arithmetic units ALU 280. That is, by separating the operation unit (ALU) 280 into a separate hardware block it is possible to simplify the processing algorithms of the dequantizer 250 and the IMDCT 251 and to increase the processing speed.

Filter bank 250 includes inverse discrete cosine transform (IMDCT) 251 and window / overlap-add 255. The filter bank 250 performs inverse discrete cosine conversion and window / overlap-add by using the processing result of the inverse quantizer 240 stored in the RAM2275.

The IMDCT 251 converts the dequantized data from the dequantizer 240 into a signal in the time domain. In detail, the signal output in the frequency domain is converted into the signal in the time domain by reading the data output through the inverse quantizer stored in the RAM 275 and performing inverse modified discrete cosine transformation. In this case, the IMDCT 251 receives the processing result of the inverse quantizer 240 from the RAM2 275, performs inverse discrete cosine transformation, and stores the result in the RAM2 275. In addition, the IMDCT 251 may receive a complex / multiply / add operation result required for inverse discrete cosine transform from the calculator 280. In other words, by separating the computing unit (ALU) 280 into a separate hardware block, the processing algorithm of the IMDCT 251 can be simplified and the processing speed can be increased.

Meanwhile, the window / overlap adder 255 is a window processing part for each subband in order to soften the output sound. In general, the windowed 2N data of the previous frame and the 2N data of the current frame are windowed. Add to overlap. For this purpose, the window / overlap-adder 255 reads the window coefficients for window processing from the ROM3 273, writes the windowed data of the previous frame into the RAM3 276, and the window / overlap from the RAM3 276. The windowed data of the previous frame for the overlap-add is read and the overlap-add operation is performed with the windowed data of the current frame.

3 illustrates a case where the previous frame data is stored in RAM3 276 which is a separate memory.

4 is a flowchart illustrating a digital audio data decoding method according to an embodiment of the present invention. 5 is an operation timing diagram of a digital audio data decoding apparatus for performing parallel processing according to an embodiment of the present invention. Referring to FIG. 5, 'A' represents a clock period occupied by the deformatter 220, 'B' represents a clock interval occupied by the Huffman decoder 230, and 'C' is an inverse. The quantizer 240 indicates the clock period occupied when the operation is performed, 'Dm' indicates the clock period occupied by the IMDCT 251 when the operation is performed, and the 'Dw' indicates the occupied time when the window / overlap-adder 255 operates. Represents a clock period.

Hereinafter, a digital audio data decoding method according to an embodiment of the present invention will be described with reference to FIGS. 3 to 5. 3 to 5, first, to perform window / overlap-add after inverse discrete cosine conversion, the memories RAM3 and 276 storing the windowed data of the previous frame are initialized (steps S200 and 5). Of 500).

When digital audio data is input through the input buffer 210, the controller 260 controls the deformatter 220 to extract header information and encoded data from the digital audio data of the first frame (step S205, 501 of FIG. 5). The Huffman decoder 230 is controlled to decode the encoded data of the first frame (S210, 503 of FIG. 5). Here, the encoded data may be Huffman encoded data. The controller 260 controls the inverse quantizer 240 to perform inverse quantization on the decoded data of the current frame (step S215, 503 of FIG. 5).

As such, the header information and the coded data extraction, decoding, and dequantization processes (steps S205, S210, and S215) of one frame are performed only at the beginning of the digital audio data decoding process according to an embodiment of the present invention. Steps S205, S210, and S215 are defined as an initialization process below.

The controller 260 controls the filter bank 250 to perform an inverse discrete cosine transform Dm1 (step S220, 504 of FIG. 5) with respect to the dequantized data of the current frame.

In addition, the controller 260 controls the deformatter 220 and the Huffman decoder 230 in parallel with performing the inverse discrete cosine transform 504 of the filter bank 250 to the digital audio data of the next frame. A process of extracting header information and coded data (step S225, 511 of FIG. 5) and a process of decoding coded data of the next frame (step S230, 512 of FIG. 5) are performed.

Next, the controller 260 controls the filter bank 250 to perform a window / overlap-add operation (step S222, 505 of FIG. 5). That is, the controller 260 controls the window / overlap-add operation 505 of the filter bank 250 to start as the inverse discrete cosine transform 504 operation of the filter bank 250 ends.

In addition, the controller 260 controls the inverse quantizer 240 in parallel with performing the window overlap add of the filter bank 250 to perform inverse quantization on the decoded data of the next frame. S232, 513 in FIG. 5).

When the window / overlap-add operation 505 of the filter bank 250 ends, the decoding of the current frame is completed and the dequantization operation 513 of the next frame is performed.

The controller 260 controls to execute the deformatting operation and the decoding operation on the data of the next frame in parallel during the inverse discrete cosine transform operation of the filter bank 250 on the data of the current frame. In addition, the controller 260 controls the inverse quantization operation on the data of the next frame to be executed in parallel during the window / overlap-add operation of the filter bank 250 on the data of the current frame.

Table 1 is a table showing the amount of calculation for each module of the WMA decoding system of FIG.

TABLE 1

Average amount of computation (clock cycles required) Deformatter 5% Huffman Decryption 25% Dequantization 10% Filter bank IMDCT 45% Widow / Overlap 15% Sum 100%

Referring to Table 1, the sum of the clock requirements of the deformatter operation 511 and the Huffman decoding operation 512 occupies about 30% and the IMDCT operation 504 of the filter bank occupies about 45%. While the IMDCT operation 504 is processed, the deformatter operation 511 and the Huffman decoding operation 512 of the next frame can be sufficiently processed in parallel. In addition, since the clock requirement of the dequantization operation 513 occupies 10% and the window / overlap add operation 505 of the filter bank occupies about 15%, the window / overlap add operation 505 of the filter bank is processed in the current frame. The inverse quantization operation 513 of the next frame can be fully processed in parallel.

In addition, the deformatter 220 and the Huffman decoder 230 may be processed in parallel because the IMDCT 251 of the filter bank does not share any memory and arithmetic unit.

In addition, the dequantizer 240 and the window / overlap adder 255 of the filter bank share the memory RAM2 275 when the operation is performed, but the window / overlap of the filter bank is terminated when the operation of the IMDCT 251 of the filter bank is terminated. Since the adder 255 performs a read operation from the memory RAM2 275, and the dequantizer 240 performs a write operation to the memory RAM2 275, there is no problem in accessing the memory in parallel processing.

Since the operation of the filter bank occupies more than half (45% + 15% = 60%) of the decoding processing period of one frame, other operations that do not share memory with the filter bank 250 during that time (eg, deformatting operation). 511 and the decoding operation 512 and inverse quantization operation 513 can be operated together to shorten the decoding processing cycle of one frame.

In addition, since the clock occupancy (about 60%) of the filter bank 250 is more than twice as large as the clock occupancy (about 25%) of the Huffman decoder 230, the system due to the variation of the clock occupancy of the Huffman decoder 230 is increased. Cycle changes can be minimized.

3 to 5, the controller 260 determines whether the operation of the inverse quantizer of the next frame (513 of FIG. 5) is terminated (S235). It is determined whether the processing of the filter bank 250 (504 and 505 of FIG. 5) has ended (step S240), and the filter on the data of the current frame until the processing of the filter bank 250 is terminated for the data of the current frame. Only the process of the bank 250 is repeatedly performed (step S220). This is because, as shown in Table 1, the clock occupancy (about 60%) for the operation of the filter bank 250 is generally larger than the sum of the clock occupancy of the deformatter 220 and the Huffman decoder 230.

As a result of the determination of step S240, when the processing of the filter bank 250 for the data of the current frame ends, the controller 260 moves to the next frame (step S245) and repeats the process after step S215. At this time, the repetition of the operations after the dequantization process (step S215) continues until the decoding operation ends (step S250).

On the other hand, the movement to the next frame in step S245 does not mean physically moving the frame but means logical movement. Since the processing of the current frame is completed, the next frame is changed to the current frame in the description, and then the next frame. To change to the next frame.

In the description of FIG. 4, the decoding of steps S210 and S230 is preferably Huffman decoding. However, the decoding method is not limited to Huffman decoding, and may include all other decoding methods having a constant clock occupancy rate.

3, the memory RAM 1 274 connected to the Huffman decoder 230 and the memory RAM 2 275 connected to the IMDCT 251 and the window / overlap-adder 255 are separated. S220 (504 of FIG. 5), step S225 (511 of FIG. 5), and step S230 (512 of FIG. 5) may be performed in parallel.

In addition, since the window / overlap-adder 255 does not use the operation unit 280 as illustrated in FIG. 3, the inverse quantizer 240 is operated at a timing at which the window / overlap-adder 255 starts operation. The operation of the window / overlap-adder 255 and the operation of the dequantizer 240 may be processed in parallel pipeline. Therefore, while the filter bank of the current frame is being performed, the deformatter 220, the Huffman decoder 230, and the dequantizer 240 of the next frame are operated in advance so that the Huffman decoder having an uneven decoding period ( Pipeline processing is possible without affecting the system clock.

Referring to FIG. 5, except for the first frame, the digital audio data decoding apparatus of the present invention includes the clock occupancy time A of the deformatter, the clock occupancy time region B of the Huffman decoder, and the clock occupancy time of the dequantizer ( C) does not exist on its own. That is, the deformatter 220, the Huffman decoder 230, and the inverse quantizer 240 occupy the clock independently only in the 't0' and 't1' intervals, after which the clock is shared with the filter bank. It can be seen. In other words, in the first and subsequent frames, the deformatter and the Huffman decoder are processing the frame during the clock occupancy time t2 of the IMDCT 251 of the filter bank which is processing the previous frame, and processes the previous frame. During the clock occupancy time t3 of the window / overlap adder 255 of the filter bank, the inverse quantizer 240 performs processing on the corresponding frame, and it can be seen that the sound is negative.

Therefore, referring to Tables 1 and 5, the present invention requires only 40% of the occupancy time in the initialization process, and the time for processing data of one frame occupies in the filter bank in the first and subsequent frames. (60%). That is, only 60% of each frame is required. In the conventional digital audio data decoding apparatus of FIG. 1, this means that the time occupied by the deformatter (5%), the time occupied by the Huffman decoder (25%), the time occupied by the dequantizer (10%), and the filter bank. Compared to 100% of each frame, which takes up 60% of the time occupied by, the clock cycles per frame are reduced by 40%, which reduces the overall data processing time.

The present invention relates to a method and apparatus for decoding digital audio data, wherein a deformatter, a Huffman decoder, and an inverse quantizer of a next frame are operated in advance while a filter bank of a current frame is being performed. Accordingly, the filter bank, deformatter, and Huffman decoder having the longest clock occupancy time in one frame period while preventing the Huffman decoder 230 whose clock occupancy time (that is, clock requirement) do not affect the system clock. In addition, since the inverse quantizer can be processed in parallel, the data processing time of the entire decoding system can be shortened, and the performance of the overall decoding system can be improved due to the efficient distribution and use of the clock.

In addition, the present invention can simplify the decoding processing algorithm and increase the processing speed by implementing a plurality of memories operating in parallel, and an operation unit (ALU) shared by the IMDCT of the inverse quantizer and the filter bank in hardware. As a result, there is an advantage that makes it possible to apply an audio file in WMA format to a portable device.

Although described with reference to the embodiments above, those skilled in the art will understand that the present invention can be variously modified and changed without departing from the spirit and scope of the invention as set forth in the claims below. Could be.

Claims (18)

Extracts header information and encoded data from the digital audio data of the first frame, decodes the encoded data of the first frame to write to the first memory, and reads the decoded data of the first frame from the first memory. An initialization step of indenting and dequantizing; Firstly extracting header information and encoded data for digital audio data of a next frame in parallel while performing inverse discrete cosine transformation on the dequantized data of the current frame and decoding the encoded data of the next frame Parallel processing step; And And a second parallel processing step of dequantizing the decoded data of the next frame in parallel while performing the window / overlap-add of the current frame. The method of claim 1, wherein the first and second parallel processing steps And decoding the digital audio data repeatedly until the decoding of the digital audio data ends. The method of claim 1, wherein the decoding is Huffman decoding. delete delete The method of claim 1, wherein the initialization step Read the decoded data of the first frame from the first memory to perform inverse quantization on the decoded data of the first frame, and then write the dequantized data of the first frame to a second memory A decoding method of digital audio data. The method of claim 6, wherein the first parallel processing step And reading inversely quantized data of the current frame from the second memory and performing inverse discrete cosine transform on the inversely quantized data of the current frame. 8. The method of claim 7, wherein the first parallel processing step And decoding the decoded coded data of the next frame into the first memory. The method of claim 1, wherein the second parallel processing step When the window / overlap-add is performed, windowed data of a previous frame is written to a third memory, and windowed data of the previous frame is read from the third memory to overlap with windowed data of the current frame. -Adding method for decoding digital audio data. The method of claim 1, wherein the digital audio data is A method of decoding digital audio data, characterized in that it is data encoded by Window Media Audio. First and second memories; A deformatter for performing a deformatting operation of extracting header information and encoded data from digital audio data; A decoder configured to decode the encoded data and store the decoded data in the first memory; An inverse quantizer configured to read inversely quantize the decoded data from the first memory and to perform inverse quantization to store dequantized data in the second memory; A filter bank including an inverse discrete cosine converter for reading inverse quantized data from the second memory and performing an inverse discrete cosine transform operation and a window / overlap-add for performing window / overlap-add; And While the inverse discrete cosine transform of the filter bank is performed on the dequantized data of the current frame, the deformatting operation and the decoding operation are performed on the digital audio data of the next frame in parallel. And a control unit configured to perform the inverse quantization operation on the decoded data of the next frame in parallel during the window / overlap-add operation. The method of claim 11, And performing a complex number / multiplication / add operation for the inverse quantization and the inverse discrete cosine transform and being shared by the inverse quantizer and the inverse discrete cosine transformer. The method of claim 11, And a third memory for storing windowed data of a previous frame for window / overlap-add of the filter bank. The method of claim 11, wherein the second memory is And windowed data of a previous frame for window overlap add of the filter bank. The method of claim 11, wherein the decoder And a Huffman decoder. The method of claim 11, wherein the digital audio data is A device for decoding digital audio data, characterized in that it is data encoded by Window Media Audio. 12. The apparatus of claim 11, wherein the apparatus for decoding digital audio data is used in a portable terminal. A deformatter for performing a deformatting operation of extracting header information and encoded data from digital audio data; A decoder configured to perform a decoding operation of decoding the encoded data; An inverse quantizer for inversely quantizing the decoded data; A filter bank including an inverse discrete cosine transformer for performing inverse discrete cosine transform operation for inverse discrete conversion of the dequantized data and a window / overlap-add for performing window / overlap-add; And While the inverse discrete cosine transform of the filter bank is performed on the dequantized data of the current frame, the deformatting operation and the decoding operation are performed on the digital audio data of the next frame in parallel. And a control unit configured to perform the inverse quantization operation on the decoded data of the next frame in parallel during the window / overlap-add operation.

Images