KR20100090962A - Multi-channel audio decoder, transceiver comprising the same decoder, and method for decoding multi-channel audio - Google Patents

Abstract

PURPOSE: A multichannel audio decoder for increasing the speed and memory size of the decoder is provided to reduce the operation quantity about the decoder in a multi channel audio system using a conversion coding method. CONSTITUTION: A frequency domain signal generator(310) generates the data signal of frequency domain from an encoded bitstream data signal. A down mixer performs the down-mixing of the frequency domain about data signal. An IFFT(Inverse Fast Fourier Transform) unit(350) transforms down-mixed data signal into the data signal of a temporal area. A window unit(370) generates PCM(Pulse Code Modulation) output signal by overlap add operation.

Description

Multi-channel audio decoder, transceiver including the decoder and multi-channel audio decoding method {Multi-channel audio decoder, transceiver comprising the same decoder, and method for decoding multi-channel audio}

TECHNICAL FIELD The present invention relates to signal processing, and in particular, to multichannel audio decoding and methods using transform coding schemes.

Generally, a digital audio coder or codec refers to a device capable of converting an analog audio signal into a digital form and vice versa. The conversion from the analog signal to the digital signal is performed at the encoder stage of the audio coder. If we have a digital signal represented as a series of numbers through this transformation, we can store it, process it, and transmit it. We can also play this sound later and listen again. This requires converting the digital signal back into an analog signal for the human ear to hear. The inverse conversion from digital signals to analog signals is done at the decoder stage of the coder.

The advantage of digital audio is that it can process signals with little change in sound quality compared to analog. Therefore, the introduction of digital audio has provided an environment in which the original sound can be heard with little loss and transmitted elsewhere. In addition, digital audio is highly compatible with other media such as movies and computers, and its applications continue to expand.

On the other hand, high-quality digital audio signals obtained by sampling analog audio signals require a large amount of storage media due to a large amount of data, and occupy a very wide frequency band during transmission. Therefore, there is a need for a method of compressing an audio signal effectively without loss of sound quality. Currently, the main coding technique for audio signals is to use signal characteristics in the frequency domain using time-to-frequency mapping to increase the extrusion rate. Most audio signals are better described as a function of frequency than time. The reason is that when we perceive sound, it depends mainly on the tone component of the sound, and because it can express the sound much more concisely in frequency than in time.

As described above, in a multi-channel audio decoder using audio coding using time-frequency conversion, a signal in a frequency domain of an input channel is converted into a signal in a time domain, and down-sized according to the number of output channels. Down-Mixing The main purpose of such multichannel coding is to reduce the data error rate by eliminating the redundancy between the channels and the independence of the signal space representation while maintaining the original spatial properties and basic sound quality of the multichannel audio signal.

Today, multichannel audio is becoming a viable alternative for a wide range of audio playback systems due to the advances in digital technology, transmission bandwidth and storage space. However, the listener of an audio recording or film mix can have an audio system that supports fewer channels than the number of channels from which the audio recording or film mix was created, thereby downgrading the multichannel signal depending on the number of channels supported. You have to mix. However, the current down-mixing method requires a large amount of computation since the time-frequency conversion needs to be preceded by the number of input channels, thereby causing a problem of high memory consumption.

The problem to be solved by the present invention is to perform down-mixing according to the number of output channels in advance, and time-frequency inverse conversion of the mixed coefficients to generate a time domain signal, thereby reducing the amount of computation in the conversion process, The present invention provides a multichannel audio decoder, a transceiver including the decoder, and a multichannel audio decoding method capable of reducing use.

In order to solve the above problems, the present invention provides a frequency domain signal generation unit for generating a frequency domain (Frequency Domain) data signal from the encoded bitstream data signal; A down mixer for down-mixing the data signal in the frequency domain according to the number of output channels; An Inverse Fast Fourier Transform (IFFT) unit for converting down-mixed data signals in the frequency domain into data signals in the time domain; And a window unit generating a pulse code modulation (PCM) output signal by performing windowing and overlap add operations on the data signal in the time domain. .

In the present invention, the IFFT unit may perform time-frequency inverse transformation by the number of output channels. For example, when the data signal of the frequency domain is input to the down mixer through M input channels, the IFFT unit is down-mixed by the down mixer into N output channels smaller than M by the data of the frequency domain. The signal may be converted into a data signal of the time domain.

In addition, the frequency domain signal generator generates a data signal in the frequency domain according to each frequency band of the input channels. The multi-channel audio decoder of the present invention determines whether the block sizes of the input channels are the same. The apparatus may further include a block switching unit configured to input the data signal in the frequency domain to the down mixer or the IFFT unit. If the block sizes are the same, data signals of the frequency domain are input and down-mixed to the frequency domain down mixer, and then converted into data signals of the time domain through the IFFT unit, and if the block sizes are different, the IFFT unit is used. After the data signal in the frequency domain is input and converted into a time domain data signal, the down mixer may be down-mixed in the time domain.

The present invention also provides a transceiver for transmitting and receiving an encoded bitstream (Bitstream) to achieve the above object; A multichannel audio encoder for encoding Pulse Code Modulation (PCM) into a bitstream; A multi-channel audio decoder having a frequency domain signal generator, a down mixer, an IFFT unit, and a window unit to decode the PCM signal; An input / output unit for inputting a voice signal to the multichannel audio encoder or outputting a PCM signal from the multichannel audio decoder; And a control unit for controlling the transceiver, the multichannel encoder, the decoder, and the input / output unit, wherein the down mixer of the multichannel audio decoder performs down-mixing in a frequency domain. .

The present invention further provides a method for generating the data signal in the frequency domain from an encoded bitstream data signal to achieve the above object; Down-mixing the frequency-domain data signal according to the number of output channels; Converting the down-mixed data signal in the frequency domain into a time domain data signal through an IFFT; And outputting a PCM signal by performing windowing and overlap-add operations on the data signal in the time domain.

In the down-mixing step, in the down-mixing step, when there are M input channels and N output channels, down-mixing on the frequency domain or the time domain using N × M downmix matrix G (N, M) Can be performed. In addition, when the data signal of the input channel in the frequency domain is represented by the M × 1 F _M (x) column matrix, and the data signal of the output channel in the frequency domain is represented by the N × 1 F _N '(x) column matrix. In the down-mixing step, multiplying G (N, M) by the F _M (x) column matrix to calculate the F _N ′ (x) column matrix, and in the converting step, F _N ′ IFFT may be performed at (x) to generate a T _N '(x) column matrix of 1 × N, which is a data signal of an output channel in the time domain.

On the other hand, to achieve the above object, the present invention comprises the steps of: generating a data signal in the frequency domain according to each frequency band of the M input channels from the encoded bitstream data signal; Determining whether the block sizes of the input channels are the same; Changing the order of down-mixing and performing IFFT according to whether the block size is the same; And outputting a PCM signal by performing windowing and overlap-add operations on the data signal in the time domain.

The multichannel audio decoder, the transmitting and receiving apparatus including the decoder, and the multichannel audio decoding method according to the present invention have the same output while reducing the number of time-frequency inverse transforms in a multichannel audio system using transform coding. It can reduce the amount of computation used in. As a result, a decoder having improved performance in terms of speed and memory size can be implemented, and the total number of operating clocks can be reduced or lowered, thereby enabling a low power design of the decoder.

The multi-channel audio decoder and decoding method of the present invention downloads data having M multiple channels to N (N <M) channels in an audio or voice decoder having a structure of converting and decoding from other frequency domains to time domains. Applicable to all decoders that mix and output.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, when a component is described as being connected to another component, it may be directly connected to another component, but a third component may be interposed therebetween. In addition, in the drawings, the structure or size of each component is exaggerated for convenience and clarity of explanation, and parts irrelevant to the description are omitted. Like numbers refer to like elements in the figures. It is to be understood that the terminology used is for the purpose of describing the present invention only and is not used to limit the scope of the present invention.

1 is a block diagram of a transmission and reception apparatus according to an embodiment of the present invention.

Referring to FIG. 1, a transceiver of the present invention includes a transceiver 100, a multichannel audio encoder 200, a multichannel audio decoder 300, an input / output unit 400, and a controller 500.

The transceiver 100 transmits and receives an encoded bitstream through an encoder. The multichannel audio encoder 200 encodes an audio Pulse Code Modulation (PCM) data signal into a bitstream for transmission. That is, windowing is performed on the audio PCM data signal, converted into a frequency domain data signal through time-frequency conversion (IFFT), and then converted into a bitstream through compression.

Windowing refers to generating a windowed data signal by applying a window function, and generally uses four window types to accommodate variable time segmentation. Also, here windowing is used with the concept of including an overlap-add operation to maintain signal continuity.

Current compression techniques use a model of the human perceptual system to remove information about the source signal and nothing. For example, in the multi-channel audio encoder of the present embodiment, the compression is performed by appropriate quantization and coding of frequency lines to generate an output signal of a prescribed bitstream format. With this quantization and coding, a psycho-acoustic model A bit-allocation algorithm adjusted by the model is used. Since the multichannel audio encoder is far from the main features of the present invention, a detailed description thereof will be omitted.

The multichannel audio decoder 300 decodes the bitstream received through the transceiver 100 into an original PCM signal. The multi-channel audio decoder 300 includes a frequency domain signal generator, a down mixer, an IFFT unit, and a window unit to process the received bitstream signal in the reverse order in the multi-channel audio encoder 200 to generate a PCM signal. do. The multichannel audio decoder 300 in this embodiment includes a down mixer that down-mixes the input channel to the output channel to match the number of channels of the user's audio output system. It is characterized by performing down-mixing in the region. A more detailed description of the multichannel audio decoder 300 is provided in the description of FIGS. 2 and 3.

The input / output unit 400 functions to input a voice signal to the multi-channel audio encoder 200 or to output a PCM signal from the multi-channel audio decoder 300. The control unit 500 is a component of the transmission / reception apparatus, that is, The transceiver 100, the multi-channel encoder and decoders 200 and 300, and the input / output unit 400 are generally controlled. Although not shown, the transmission and reception apparatus of the present embodiment may include a storage unit for storing various data signals such as video and audio data signals.

The transmitting and receiving device of the present embodiment can reduce the number of time-frequency conversions by performing down-mixing in the frequency domain in a multichannel audio system using transform coding, and accordingly, the amount of computation used in the decoder while having the same output as before. Can be reduced. In addition, due to the reduction of the calculation amount, the performance of the audio decoder can be improved in terms of speed and memory size, and the number of operating clocks can be reduced or lowered, thereby enabling the low-power design of the decoder, thereby making the size and reception of the transmitting and receiving device It can improve performance.

FIG. 2 is a block diagram illustrating in detail the multichannel audio decoder of FIG. 1.

Referring to FIG. 2, the multi-channel audio decoder of the present embodiment includes a frequency domain signal generator 310, a frequency domain down mixer 330, an inverse fast fourier transform (IFFT) unit 350, and a window unit 370. do.

The frequency domain signal generator 310 generates a data signal of the frequency domain from the encoded bitstream data signal. The frequency domain signal generation is performed through decoding and inverse quantization in response to quantization and coding of the multichannel audio encoder. The data signal of the frequency domain is extracted according to each frequency band of the input channels. The frequency domain down mixer 330 performs down-mixing of the frequency domain data signal according to the number of output channels. Such frequency domain down-mixing is described in more detail in the description of FIGS. 4 and 5 as compared to time domain down-mixing.

The IFFT unit 350 converts the down-mixed data signal of the frequency domain into a data signal of the time domain through time-frequency inverse transform. The window unit 370 outputs a PCM signal by applying a window function to a data signal in the time domain to take a window wing. Here, the window unit 370 also performs the overlap-add operation as described above in the multi-channel audio encoder.

The multi-channel audio decoder of this embodiment can reduce the number of time-frequency conversions by performing down-mixing in the frequency domain in advance, thereby reducing the amount of computation used in the decoder while having the same output as before. .

3 is a block diagram illustrating a modification of the multichannel audio decoder of FIG. 2.

Referring to FIG. 3, the multi-channel audio decoder of the present embodiment is similar to the multi-channel audio decoder of FIG. 2, but further includes a block switching unit 320 after the frequency domain signal generator 310.

The block switching unit 320 determines whether the block sizes of the input channels with respect to the data signal input from the frequency domain signal generator 310 are the same, and performs down-mixing of the frequency domain or downs the time domain. Determine whether to perform mixing. That is, if the block sizes of the input channels are the same, down-mixing is first performed in the frequency domain down mixer 330 to reduce the number of time-frequency inverse transforms, and then time-frequency inverse transform is performed through the IFFT unit 350. Do it. On the other hand, if the block sizes of the input channels are different, the time-frequency inverse transform is first performed through the IFFT unit 350, and then down-mixing is performed through the time domain down mixer 330a. Here, although reference numerals are different for the frequency domain down mixer 330 and the time domain down mixer 330a, the down-mixing of the frequency domain or the down-mixing of the time domain is selectively performed through one down mixer. Of course you can.

When the block size of the input channel is different, the time-frequency inverse transform through the IFFT unit 350 varies depending on the channel. Thus, the result of down-mixing the time domain after the time-frequency inverse transform is after the down-mixing of the time domain. It is different from the result of the time-frequency inverse transform, so that down-mixing of the frequency domain cannot be performed first. However, when the block sizes of the input channels are the same, the result value of the time-domain down-mixing after the time-frequency inverse transform is the same as the result value of the time-frequency inverse transformation after the down-mixing of the time domain. Therefore, it is advantageous to first perform down-mixing of the frequency domain to reduce the number of time-frequency inverse transforms. 4 and 5, when the block sizes of the input channels are the same, the result of down-mixing the time domain after time-frequency inverse transform is the same as the result of time-frequency inverse transform after down-mixing of time domain. Explain.

4 and 5 are block diagrams illustrating a down-mixing process in a time domain and a down-mixing process in a frequency domain in a multichannel audio decoding method of the present invention. FIG. 4 shows the down-mixing process after performing the time-frequency inverse transform first, and FIG. 5 shows the time-frequency inverse transformation after the down-mixing. Here, it is assumed that the block size of the input channel is the same, and it will be described with reference to FIGS. 2 and 3 together for convenience of understanding.

Referring to FIG. 4, after the data signal of the frequency domain is extracted through bit allocation (dequantization) and decoding in the frequency domain signal generator 310, a time-frequency inverse transform is performed in the IFFT 350. At this time, the data signal of the frequency domain is extracted according to each frequency band of the input channel of the down mixer. Meanwhile, the IFFT 350 performs time-frequency inverse transform for each channel. If the number of input channels is M, the IFFT 350 performs time-frequency inverse transform as much as M loops. In this way, the time-frequency inverse transformed signal is input to M input channels of the down mixer 330a to perform down-mixing. For example, when the output channel is N (N <M), the data signals of the M input channels are down-mixed with the data signals of the N output channels. The down-mixed signal is then windowed through window portion 370 and converted to a PCM signal.

Referring to FIG. 5, after the data signal of the frequency domain is extracted by bit allocation and decoding in the frequency domain signal generator 310, the down mixer 330 performs down-mixing of the frequency domain. That is, when the number of input channels of the down mixer 330 and the number of output channels is N, the data signal of the frequency domain corresponding to the M channels is the data signal of the frequency domain corresponding to the N channels through the down mixer 330. Are down-mixed. In the down-mixed frequency domain data signal, time-frequency inverse transform is performed in the IFFT 350. Since the data signal in the frequency domain that is input has N channels, time-frequency inverse transform is performed only by N loops. The data signal of the area is converted. The data signal of the time domain is also window-winged through the window unit 370 and converted into a PCM signal.

Comparing FIG. 4 and FIG. 5, it can be seen that the time-frequency inverse transform can be reduced if the time-frequency inverse transform is performed after down-mixing in the frequency domain. That is, in case of performing down-mixing in the time domain, time-frequency inverse transform should be performed by the number of input channels M times, but in case of performing down-mixing in the frequency domain, only time N of output channels should be used. Since the frequency inverse transform is performed, the number of time-frequency inverse transforms can be reduced by the difference between the number of input channels and the number of output channels.

Quantitatively calculating the amount of calculation reduced in the IFFT unit 350 can be expressed by the following equations.

Calculation amount of IFFT per channel = Q Mips ..... (1)

IFFT computed amount for time-domain down-mixing = M * Q Mips ...... (2)

Frequency-domain down-mixing IFFT computation = N * Q Mips ............... Equation (3)

Reduced IFFT computation = (M-N) * Q Mips, (where, N <M) ......... Equation (4)

For example, in the case of down-mixing a bitstream data signal encoded in a 5.1 channel (corresponding to six channels) which is frequently used, into two channels by applying a frequency domain down-mixing method, 4 * The amount of computation in Q Mips can be reduced. Considering that the amount of time taken by the time-frequency inverse transform operation in the decoder using the transform coding occupies about 30% of the total amount of computation, it can be seen that a considerable amount of computation can be reduced.

On the other hand, in order to be able to use frequency domain down-mixing instead of time domain down-mixing, it should be ensured that the final result should be the same. This final result identity can be explained by the linear property of the time-frequency transform below.

First, when down-mixing the signals of the M input channels in the time domain with the signals of the N output channels, the relationship between the signal of the input channel and the output signal is expressed using Equation (5).

...............식(5)

............... Equation (5)

Time domain signals f _{1 (} x), f ₂ (x),... Of m input channels using an arbitrary downmix matrix G (n * m) for down-mixing as represented in equation (5). If we down-mix f _m (x), the result is (6).

here,

n <m,

f (x) = {f1 (x), f2 (x), ...., fm (x)}: signals of m input channels,

f '(x) = {f1´ (x), f1´ (x), ...., fn´ (x)}: signals from n output channels,

G = {{A ₁₁ , B ₁₂ , ...., K _1m }, {A ₂₁ , B ₂₂ , ...., K _2m }, ...., {A _n1 , B _n2 , ... , Knm}}: downmix coefficient of the mth input channel to the nth output channel

f ₁ '(x) = A ₁₁ * f ₁ (x) + B ₁₂ * f ₂ (x) + .... + K _1m * f _m (x),

f ₂ '(x) = A ₂₁ * f ₁ (x) + B ₂₂ * f ₂ (x) + .... + K _2m * f _m (x),

.

.

.

f _n '(x) = A _n1 * f ₁ (x) + B _n2 * f ₂ (x) + .... + K _nm * f _m (x) ............ Equation (6)

Down-mixing the coefficients in this time domain yields the same result as Equation (7).

F ₁ '(X) = A ₁₁ * F ₁ (X) + B ₁₂ * F ₂ (X) + .... + K _1m * F _m (X),

F ₂ '(X) = A ₂₁ * F ₁ (X) + _B22 * F ₂ (X) + .... + K _{2 m} * F _m (X),

.

.

.

F _n '(X) = A _n1 * F ₁ (X) + B _n2 * F ₂ (X) + .... + K _nm * F _m (X) ............ . (7)

Here, each signal in the time domain is called x ₁ (x), x ₂ (x), and the signals in the frequency domain in which the time-frequency conversion of the signals are transformed as in Equations (8) and (9) are X _1. In the case of (X) and X ₂ (X), equation (10) is established by the linearity characteristic of the time-frequency conversion.

x ₁ (x) ↔ X ₁ (X) ..................... Formula (8)

x ₂ (x) ↔ X ₂ (X) ..................... Formula (9)

A * x ₁ (x) + B * x ₂ (x) ↔ A * X ₁ (X) + B * X ₂ (X) ... )

By the same principle, equation (6) can be obtained by inversely transforming equation (7) due to the linearity of the time-frequency conversion. Therefore, the final output signal which performed the down-mixing of the frequency domain and the final output signal which performs the down-mixing of the time domain may be the same.

6 is a flowchart illustrating a multichannel audio decoding method according to another embodiment of the present invention.

Referring to FIG. 6, in the multichannel audio decoding method of the present embodiment, first, a bitstream encoded by a multichannel decoder is input (S110). Next, a data signal of a frequency domain is generated from the encoded bitstream data signal through decoding and inverse quantization (S130). Thereafter, the data signal in the frequency domain is down-mixed to match the number of output channels through down-mixing in the frequency domain (S150). The data signal of the down-mixed frequency domain is transformed into a data signal of the time domain by performing time-frequency inverse transformation, that is, IFFT. Finally, the data signal in the time domain is converted into a PCM signal through the widowing (S190).

In the multi-channel audio decoding method of the present embodiment, down-mixing of the frequency domain is performed through down-mixing before performing time-frequency inverse transform, thereby reducing the number of time-frequency inverse transform performed in the IFFT unit later. That is, when there are M input channels and N output channels, the number of time-frequency inverse transforms performed by M-N can be reduced. Accordingly, considering that the amount of time occupied by the time-frequency inverse transform operation in the multichannel decoder occupies about 30% of the total amount of operations, a considerable amount of computation can be reduced.

7 is a flowchart illustrating a multichannel audio decoding method according to another embodiment of the present invention.

Audio signals can be roughly divided into stationary signals and non-stationary signals. Since the normal signal and the abnormal signal have different characteristics, the coding method of the signal is also different. Block switching can be used as a method of distinguishing between normal signal and abnormal signal, and this block switching enables to identify the characteristics of the signal more accurately by coding the normal signal and the abnormal signal in different block sizes. . Here, the normal signal is a signal whose characteristic does not change with time, and is coded with the same block size, and the abnormal signal is a signal whose characteristic varies with time, and coding while adjusting the block size according to a situation. Therefore, in an abnormal signal, it means that the block size may vary for each channel.

Referring to FIG. 7, in the multi-channel audio decoding method according to the present embodiment, similarly to FIG. 6, a bitstream encoded by a multichannel decoder is input (S210), and a decoding and dequantization process is performed from the encoded bitstream data signal. The data signal of the frequency domain is generated through the display (S230). Thereafter, it is determined whether the block sizes of all input channels are the same through the block switching unit (S230).

If the block sizes of the input channels are the same, the process as shown in FIG. 6 is performed. That is, after down-mixing of the frequency domain is performed (S270), a time-frequency inverse transform is performed (S250), and finally, a PCM signal is output through windowing (S290).

On the other hand, when the block sizes of the input channels are different, as described above, since the time-frequency inverse transform through the IFFT unit varies depending on the channel, the result of down-mixing the time domain after the time-frequency inverse transform is down in the frequency domain. After mixing, it is different from the result of the time-frequency inverse transform, so that down-mixing of the frequency domain cannot be performed first. Accordingly, when the block sizes of the input channels are different, unlike in FIG. 6, time-frequency inverse transform is first performed 250, and then down-mixing of the time domain is performed (S270a). Finally, the PCM signal is output through windowing (S290).

The multi-channel audio decoding method of the present embodiment can efficiently perform decoding on various audio signals by selectively performing a decoding method according to whether the block sizes are identical in consideration of the block sizes of the input channels.

So far, the present invention has been described with reference to the embodiments shown in the drawings, which are merely exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. . Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

1 is a block diagram of a transmission and reception apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating in detail the multichannel audio decoder of FIG. 1.

3 is a block diagram illustrating a modification of the multichannel audio decoder of FIG. 2.

4 and 5 are block diagrams illustrating a down-mixing process in a time domain and a down-mixing process in a frequency domain in a multichannel audio decoding method of the present invention.

6 is a flowchart illustrating a multichannel audio decoding method according to another embodiment of the present invention.

7 is a flowchart illustrating a multichannel audio decoding method according to another embodiment of the present invention.

Claims (18)

A frequency domain signal generator configured to generate a frequency domain data signal from the encoded bitstream data signal; A down mixer for down-mixing the data signal in the frequency domain according to the number of output channels; An Inverse Fast Fourier Transform (IFFT) unit for converting down-mixed data signals in the frequency domain into data signals in the time domain; And And a window unit configured to generate a pulse code modulation (PCM) output signal by performing windowing and overlap add operations on the data signal of the time domain. According to claim 1, And the IFFT unit performs time-frequency inverse transformation by the number of output channels. According to claim 1, The frequency domain signal generator generates a data signal of the frequency domain according to each frequency band of input channels. And a block switching unit configured to determine whether the block sizes of the input channels are the same and to input the data signal in the frequency domain to the down mixer or the IFFT unit. The method of claim 3, If the block sizes are the same, the data signal of the frequency domain is input to the frequency domain down mixer, down-mixed, and then converted into a time domain data signal through the IFFT unit. If the block size is different, the multi-channel audio decoder characterized in that the data signal of the frequency domain is input to the IFFT unit and converted into a time domain data signal, and then down-mixing of the time domain through the down mixer. According to claim 1, The down mixer has M input channels and N output channels, And performing down-mixing on a frequency domain or a time domain using a downmix matrix composed of a gain value of N × M. 6. The method of claim 5, The time-frequency inverse transform performed by the IFFT unit has linearity, The value of performing the time-frequency inverse transform after down-mixing the frequency domain using the downmix matrix and the value of the down-mixing of the time domain using the downmix matrix after the time-frequency inverse transform are the same. Multichannel audio decoder. A transceiver for transmitting and receiving an encoded bitstream; A multichannel audio encoder for encoding Pulse Code Modulation (PCM) into a bitstream; A multi-channel audio decoder having a frequency domain signal generator, a down mixer, an IFFT unit, and a window unit to decode the PCM signal; An input / output unit for inputting a voice signal to the multichannel audio encoder or outputting a PCM signal from the multichannel audio decoder; And And a controller for controlling the transceiver, the multichannel encoder, the decoder, and the input / output unit. And the down mixer of the multichannel audio decoder performs down-mixing in a frequency domain. The method of claim 7, wherein The frequency domain signal generator generates a frequency domain data signal from the encoded bitstream data signal, The down mixer performs down-mixing of the data signal in the frequency domain in the frequency domain according to the number of output channels, The IFFT unit converts the down-mixed data signal of the frequency domain into a data signal of the time domain, and performs time-frequency inverse conversion by the number of output channels. And the window unit generates a pulse code modulation (PCM) output signal by performing windowing and overlap add operations on the data signal in the time domain. The method of claim 7, wherein The data signal of the frequency domain is input to the down mixer through M input channels, And the IFFT unit converts the data signal in the frequency domain down-mixed into N output channels smaller than M by the down mixer into a data signal in the time domain. The method of claim 7, wherein The frequency domain signal generator generates a data signal of the frequency domain according to each frequency band of input channels. And a block switching unit which determines whether the block sizes of the input channels are the same, and inputs the data signal in the frequency domain to the down mixer or the IFFT unit. Generating a data signal in a frequency domain from the encoded bitstream data signal; Down-mixing the frequency-domain data signal according to the number of output channels; Converting the down-mixed data signal in the frequency domain into a time domain data signal through an IFFT; And Outputting a PCM signal by performing windowing and overlap-add operations on the data signal in the time domain. 12. The method of claim 11, And inversely converting time-frequency inverse number by the number of output channels in the converting step. 12. The method of claim 11, In the down-mixing step, when there are M input channels and N output channels, A down-mixing method in a frequency domain or a time domain is performed by using a downmix matrix composed of a gain value of N × M. The method of claim 13, When the data signal of the input channel in the frequency domain is represented by the M × 1 F _M (x) column matrix, and the data signal of the output channel in the frequency domain is represented by the N × 1 F _N '(x) column matrix, In the down-mixing step, multiply G (N, M) by the F _M (x) column matrix to calculate the F _N ′ (x) column matrix, In the converting, the IFN is performed on the F _N '(x) to generate a T _N ' (x) column matrix of 1 × N that is a data signal of an output channel in a time domain. Way. 15. The method of claim 14, The T _N '(x) column matrix is equal to the value representing the data signal of the output channel of the time domain down-mixed using G (N, M) by performing IFFT on the data signal in the frequency domain. Multi-channel audio decoding method, characterized in that. Generating a data signal in a frequency domain according to each frequency band of the M input channels from the encoded bitstream data signal; Determining whether the block sizes of the input channels are the same; Changing the order of down-mixing and performing IFFT according to whether the block size is the same; And Outputting a PCM signal by performing windowing and overlap-add operations on the data signal in the time domain. The method of claim 16, If the block size is the same, Down-mixing the data signal of the frequency domain according to the number of output channels according to the number of output channels, and then converting the down-mixed data signal of the frequency domain into a data signal of the time domain through an IFFT, If the block size is different, Converting the data signal in the frequency domain into a data signal in the time domain through an IFFT, and then performing down-mixing of the time domain data signal according to the number of output channels. . 18. The method of claim 17, The data signal of the time domain converted through the IFFT when the block sizes are the same, After converting the data signal in the frequency domain to the data signal in the time domain through an IFFT, the multi-value of the time domain data signal is the same as the result of down-mixing the time domain according to the number of output channels Channel audio decoding method.

Images