KR20060049980A - Apparatus for encoding and decoding multichannel audio signal and method thereof - Google Patents

Abstract

The present invention relates to a method and apparatus for encoding / decoding a multichannel audio signal, and more particularly, to a method and apparatus for multichannel parametric encoding / decoding for encoding and decoding a multichannel audio signal based on channel correlation. will be.

An apparatus for encoding a multichannel audio signal according to the present invention includes: signal dividing means for dividing a multichannel audio signal in a frequency domain; Downmixing means for downmixing the divided signals into a single channel and encoding the divided signals; And parameter analyzing means for measuring a channel correlation degree in units of frames with respect to the divided signals, and re-dividing the divided signals according to the channel correlation degree in units of frames. If the correlation is less than or equal to the first predetermined value, it is a queue parameter for decoding, and the channel ID and the subband unit channel correlation for the subband of the subdivided signal are extracted and encoded.

The present invention provides an effect of restoring and reproducing a multichannel audio signal from a single channel audio signal even in an environment where a small bandwidth is allocated by optimizing a cue parameter required for decoding a multichannel audio signal.

Audio Coding, Cue Parameters, Multichannel, Channel Correlation

Description

Apparatus for encoding and decoding multichannel audio signal and method

1 is a block diagram of a multi-channel audio signal encoding / decoding apparatus according to the present invention;

FIG. 2 is a detailed block diagram of a multi-channel parametric encoder of FIG. 1; FIG.

3 is a detailed block diagram of a parameter analyzer of FIG. 2;

4 is a detailed block diagram of a multi-channel parametric decoder of FIG. 1;

FIG. 5 is a detailed block diagram of a parameter-based multichannel generator of FIG. 4; FIG.

6 is a flowchart illustrating a process of encoding a multichannel audio signal according to the present invention; And

7 is a flowchart illustrating a process of decoding a multichannel audio signal according to the present invention.

The present invention relates to a method and apparatus for encoding / decoding a multichannel audio signal, and more particularly, to a method and apparatus for multichannel parametric encoding / decoding for encoding and decoding a multichannel audio signal based on channel correlation. will be.

The technology of encoding wideband audio signals has grown tremendously in recent years. With the development of such coding technology, a multichannel audio system capable of reproducing multichannel audio at home or in a moving vehicle has been commercialized. However, due to the bandwidth problem in the transmission of the multi-channel audio signal, it has not yet reached a wider commercialization stage.

Recently, the introduction of Spatial Audio Coding (SAC) technology enables efficient transmission and presentation of multi-channel audio signals in a limited bandwidth. In particular, BAC (Binaural cue coding) SAC is a technique that has attracted much attention recently as an algorithm capable of reproducing high-quality multichannel audio signals with low bandwidth.

The basic concept of BCC is to generate an audio signal using cue parameters that can be expressed based on the fact that two ears generate a sense of space with which humans can feel. BCC is a human-recognized cue parameter that recovers multichannel audio signals from downmixed single channel audio signals using differences in loudness and sound delay. More details on BCC are described in the article "Binaural Cue Coding-part II: schemes and application" (IEEE Trans.on speech and audio Proc. Vol. 11. No. 6. Nov. 2003).

However, the BCC method has a problem in that the bandwidth for transmitting the extracted queue parameter also increases linearly as the number of channels increases. For example, assuming that the cue parameter required to represent a stereo signal takes 4 kbps, it takes 20 kbps for 5 channels, which takes up a considerable amount of bandwidth.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and an object thereof is to provide an audio reproduction quality similar to that of a BCC by using a small amount of cue parameters by extracting an optimal cue parameter according to channel correlation.

Those skilled in the art to which the present invention pertains will readily recognize other objects and advantages of the present invention from the drawings, the description of the invention, and the claims.

An apparatus for encoding a multichannel audio signal according to the present invention includes: signal dividing means for dividing a multichannel audio signal in a frequency domain; Downmixing means for downmixing the divided signals into a single channel and encoding the divided signals; And parameter analyzing means for measuring a channel correlation degree in units of frames with respect to the divided signals, and re-dividing the divided signals according to the channel correlation degree in units of frames. If the correlation is less than or equal to the first predetermined value, it is a queue parameter for decoding, and the channel ID and the subband unit channel correlation for the subband of the subdivided signal are extracted and encoded.

The parameter analyzing means re-segments the divided signal into an increasing number of subbands as the frame-by-frame channel correlation decreases. The frame correlation is obtained from channel correlation in subbands for the divided signals. The frame unit channel correlation is determined by the subband unit channel correlation of the low frequency band rather than the subband unit channel correlation of the high frequency band. The channel ID for the subdivided subband b is an index of a channel at which the power of the subdivided subband b is maximized.

The parameter analyzing means is a cue parameter when the frame unit channel correlation is equal to or greater than the first predetermined value, and includes an inter-channel level difference and a subband unit channel for each subband of the subdivided signal. The correlation is extracted and encoded. The parameter analyzing means additionally extracts an inter-channel level difference for each subband of the subdivided signal as the cue parameter.

The parameter analyzing means may repartition the divided signal into a number greater than the number of subbands of the divided signal when the frame unit channel correlation is less than or equal to the first predetermined value, and the frame unit channel correlation may be determined by the method. When the first predetermined value is greater than or equal to a second predetermined value, the divided signal is re-divided into a number less than the number of subbands of the divided signal, and the channel unit channel correlation is equal to the first predetermined value and the first value. If it is between two predetermined values, no repartitioning is performed.

The parameter analyzing means is a cue parameter when the frame unit channel correlation is between the first predetermined value and the second predetermined value, and the frequency magnitude difference between channels for each subband of the divided signal (Inter-Channel). Level Difference) and subband unit channel correlation are extracted and encoded.

 An apparatus for decoding a multichannel audio signal according to the present invention downmixes a multichannel audio signal divided in a frequency domain into a single channel, encodes the multichannel audio signal, and divides the multichannel audio signal according to a channel correlation degree in units of frames for the divided multichannel audio signal. 1. A multichannel audio signal decoding apparatus for decoding an encoded audio signal into a multichannel audio signal by re-dividing a signal, comprising: single channel decoding means for decoding the encoded single channel audio signal; And parameter-based multi-channel generating means for dividing the decoded single-channel audio signal in a frequency domain in the same manner as the sub-band interval used in the re-division according to the frame unit channel correlation. The generating means decodes the single channel audio signal using the channel ID and the subband unit channel correlation for the subband of the subdivided signal when the frame unit channel correlation is less than or equal to a first predetermined value. do.

When the frame-based channel correlation is greater than or equal to the first predetermined value, the parameter-based channel generating unit decodes the subband-based channel magnitude and the subband-specific channel correlation between the subdivided signals. The parameter-based channel generating means additionally decodes the time delay difference between the subbands of the subdivided signals.

In the parameter-based channel generating means, when the channel ID for the subband b is c '(1≤c'≤C, C is the number of channels), the subband b signal of the c' channel is the subband b of the single channel audio signal. Assign it as a signal. In addition, the parameter-based channel generating means allocates the remaining subband signal to the single channel audio signal by reflecting components related to the spatial feeling of the sound source generated between channels.

The multi-channel audio signal encoding method according to the present invention comprises the steps of: dividing the multi-channel audio signal in the frequency domain; Down-mixing the divided signal into a single channel and encoding the divided signal; Measuring a channel correlation in units of frames with respect to the divided signal; Repartitioning the divided signal according to the channel correlation in the frame unit; And extracting and encoding a channel ID and a subband unit channel correlation for the subbands of the subdivided signal when the frame unit channel correlation is less than or equal to a first predetermined value.

In the multi-channel audio signal decoding method according to the present invention, the multi-channel audio signal divided in the frequency domain is downmixed and encoded into a single channel, and the division is performed according to the channel correlation in the frame unit for the divided multi-channel audio signal. CLAIMS 1. A multichannel audio signal decoding method for decoding an encoded audio signal into a multichannel audio signal by re-dividing an encoded signal, comprising: decoding the encoded single channel audio signal; Dividing the decoded single channel audio signal in a frequency domain in the same manner as the subband interval used in the re-division according to the frame-by-frame channel correlation; And decoding the single channel audio signal using a channel ID for the subband of the subdivided signal and the subband unit channel correlation when the frame unit channel correlation is less than or equal to a first predetermined value. It features.

The above objects and features will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, whereby those skilled in the art to which the present invention pertains may easily implement the technical idea of the present invention. . In addition, in describing the present invention, when it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of a multi-channel audio signal encoding / decoding apparatus according to the present invention. The multichannel audio signal having C channels is input to the multichannel parametric encoder 101. The multi-channel parametric encoder 101 down-mixes the input multi-channel audio signal and outputs a single channel audio signal in the form of raw data. In addition, the multichannel parametric encoder 101 extracts and outputs cue parameter information for decoding into a multichannel signal based on channel correlation.

The cue parameter information includes Inter-Channel Level Difference (ICLD) or Channel ID (IDentification) according to channel correlation. More specifically, ICLD is used when Inter-Chanenel Correalation (ICC) is greater than or equal to a predetermined value. In this case, an inter-channel time difference (ICTD) may be additionally used. In addition, the channel ID is used when the channel correlation is less than or equal to a predetermined value. At this time, ICTD may additionally be used.

The single channel audio signal output from the multichannel parametric encoder 101 is encoded by the general audio encoder 103 into a specific file format. The general audio encoder 103 is a general audio encoder generally known as an mp3 encoder.

The general audio decoder 105 corresponds to the general audio encoder 103, and decodes a voice file having a specific file format and outputs a single channel audio signal having a raw data format. The multichannel parametric decoder 107 reconstructs and outputs the single channel audio signal input to the multichannel audio signal using the cue parameter information generated by the multichannel parametric encoder 101. More specifically, when channel correlation is less than or equal to a predetermined value, ICLD is used to restore a single channel audio signal to a multichannel audio signal. In addition, when the channel correlation is greater than or equal to a predetermined value, the channel ID is used to restore the single channel audio signal to the multichannel audio signal.

FIG. 2 is a detailed block diagram of the multichannel parametric encoder 101 of FIG. 1. Each channel signal is frequency-transformed, for example, by Fast Fourier Transform (FFT), on a frame-by-frame basis by a time frequency (TF) converter 201. Each FFT-converted channel signal is divided into B subbands by the signal splitter 203. The b (1≤b≤B) th subband consists of a split boundary value, an FFT spectral coefficient between A _b-1 and A _b -1.

For example, each FFT-converted channel signal of 1024 points may be divided into 19 subbands by Table 1 in the signal splitter 203. In this case, the first subband will consist of 0, 1, 2nd frequency coefficients, and the second subband will consist of 3, 4, 5th frequency coefficients. Here, only the 513th frequency coefficient is considered due to the symmetry of the frequency coefficient.

Equivalent Rectangular Band (ERB) is used similarly to the critical band division so that the auditory characteristics are reflected in the subband division method. The divided frequency domain subband signals are mixed downmix by the downmix unit 205 in subband units, thereby generating a single channel frame signal in the frequency domain.

The downmixed single channel frame signal is inverse fast Fourier transform (Inverse FFT) by the inverse time frequency (TF) converter 207 to generate a single channel audio signal that can be used as an input wave signal of a general audio decoder.

Meanwhile, the frequency domain subband signals divided into 19 bands are also input to the parameter analyzer 209. The parameter analyzer 209 measures channel correlation. In addition, the parameter analyzer 209 extracts cue parameter information necessary to restore the multichannel audio signal from the single channel audio signal based on the measured channel correlation. A more detailed description of the parameter analyzer 209 will be made with reference to FIG. 3.

3 is a detailed block diagram of the parameter analyzer 209 of FIG. 2. The channel correlation measurer 301 measures a channel correlation between subbands based on the divided frequency domain subband signals. The subband unit channel correlation is calculated using a normalized coherence function as shown in Equation 1 below.

는 i번째 채널 신호의 b번째 서브밴드 신호와 j번째 채널 신호의 b번째 서브밴드 신호간의 상관도이다. x^f _i,b는 i번째 채널 신호의 b번째 서브밴드의 주파수 계수이다. 위첨자 f는 주파수 영역임을 표현한 것이며 *는 해당 복소수의 공액복소수 변환을 나타낸다. i는 기준채널로서 고정되어 있는 것으로 간주한다. 그러므로 채널 수가 C개일 때, 각 서브밴드별로 C-1개의 상관도가 측정된다. 기준채널로 1번째 채널 신호가 사용되는 경우, 서브밴드 b에 대한 채널 상관도 계수 Γ_b는 수학식 2로 표현된다. 기준채널로서는 오디오 신호의 주요 전력을 제 공하는 센터(cneter) 채널 신호가 사용된다.here,

Is a correlation between the b th subband signal of the i th channel signal and the b th subband signal of the j th channel signal. x ^f _{i, b} is the frequency coefficient of the b th subband of the i th channel signal. The superscript f denotes the frequency domain, and * denotes a conjugate complex conversion of the corresponding complex number. i is considered to be fixed as a reference channel. Therefore, when the number of channels is C, C-1 correlations are measured for each subband. When the first channel signal is used as the reference channel, the channel correlation coefficient Γ _b for the subband _b is expressed by Equation 2. As the reference channel, a center channel signal that provides the main power of the audio signal is used.

The channel correlation coefficient Γ for one frame may be obtained from Equation 3.

As N ≒ M / 2, when M is 19, N is determined to be 10. α is a smoothing factor, and preferably has a small value of 0.5 or less so that the initial N low-pass signals mainly influence the channel correlation coefficient per frame.

The channel correlation measurer 301 transmits the frame correlation coefficient Γ as a cue parameter to the multi-channel parametric decoder 107.

According to the channel correlation coefficient Γ, the type of cue parameter required for reconstructing the multichannel audio signal is determined differently. Further, according to the channel correlation coefficient Γ, the number of subbands for extracting the cue parameter is determined differently.

The relationship between the channel correlation coefficient Γ and the number of subbands for extracting the cue parameter may be determined as shown in Table 2.

That is, if the channel correlation coefficient Γ is large, the cue parameter is extracted from each subband of the subband split signal divided into a small number. On the contrary, if the channel correlation coefficient Γ is small, the cue parameter is extracted from each subband of the multiple subband signal. Extract the parameters.

For example, when the channel correlation coefficient Γ is 0.9, the 10 subband parameter analyzer 303 repartitions frequency coefficients according to Table 3 with respect to the input 19 subband split signals. That is, the divided signal having 19 subbands is transformed into a divided signal having 10 subbands by the 10 subband parameter analyzer 303.

Thereafter, the 10 subband parameter analyzer 303 extracts the queue parameters to be transmitted to the multichannel parametric decoder 107 for each of the 10 subbands. The channel size difference (ICLD) is used as the cue parameter. ICLD is a parameter that indicates the power ratio of a subband to a reference channel, which is typically a center channel. In addition, the inter-channel time difference (ICTD) can additionally be used as a cue parameter. ICTD is a parameter representing a time delay difference between a reference channel in each subband, and is obtained so that maximum correlation between signals is provided from a correlation function between the reference channel and another channel. ICTD can be selectively used as a cue parameter because it does not significantly affect the audio quality.

ICLD is obtained by equations (4) and (5).

ΔL _{c −1, b} is an ICLD value for the b th subband in the c th channel, and the first channel is used as the reference channel. S _{c, n} is the _n -point frequency coefficient in the c-th channel, P _{c , b} is the power of the b-th subband in the c-channel. If the number of channels is C, C-1 ICLDs are extracted for one subband.

The extraction process of ICLD and ICTD is well known as published in the article "Binaural Cue Coding-part II: schemes and application" (IEEE Trans. On speech and audio Proc. Vol. 11. No.6. Nov. 2003). Description is omitted.

The 10 subband parameter analyzer 303 re-extracts the channel correlation coefficient Γ _b for the 10 subbands according to the subband intervals of Table 3 and transmits the channel correlation coefficient Γ _b to the multichannel parametric decoder 107 as a cue parameter.

When the channel correlation coefficient Γ is between 0.5 and 0.8, the 19 subband parameter analyzer 305 extracts a cue parameter for each of 19 subbands for the 19 subband split signals. As in the case where the channel correlation coefficient Γ is between 0.8 and 1.0, the inter-channel frequency magnitude difference (ICLD) is used as the cue parameter. In addition, the inter-channel time difference (ICTD) can additionally be used as a cue parameter. In this case, the subband unit channel correlation coefficient equal to the subband unit channel correlation coefficient calculated by the channel correlation measurer 301 is transmitted to the multichannel parametric decoder 107.

When the channel correlation coefficient Γ is between 0 and 0.5, the 32 subband parameter analyzer 303 repartitions frequency coefficients according to Table 4 with respect to the input 19 subband split signals. That is, the divided signal having 19 subbands is transformed into a divided signal having 32 subbands by the 32 subband parameter analyzer 303.

Thereafter, the 32 subband parameter analyzer 307 extracts the queue parameters to be transmitted to the multichannel parametric decoder 107 for each of the 32 subbands. As a cue parameter, a channel ID assigned to each subband is used instead of ICLD. Here, the channel ID I _{b for} the subband refers to the index of the channel having the maximum power for any subband b. If this is expressed as an equation, Equation 6 is obtained.

By using the channel ID instead of ICLD as the queue parameter to be transmitted to the multi-channel parametric decoding unit 107, the amount of queue parameter transmission is greatly reduced. For example, in case of a 5-channel signal, a total of 32 required cue parameters are required for each subband. If ICLD is used, it will take 128 ICs (= 4 x 32) since 4 ICLDs are required for each subband. However, even when the number of subbands is small, if the channel ID is used as the cue parameter, the signal deterioration becomes large. Therefore, when the number of subbands is large enough, such as 32, the channel as a queue parameter to be transmitted to the multi-channel parametric decoder 107 is used. It is preferable to use ID.

Of course, the inter-channel time difference (ICTD) can additionally be used as a cue parameter.

The 32 subband parameter analyzer 307 re-extracts the channel correlation coefficient Γ _b for the 32 subbands according to the subband intervals of Table 4 and transmits the channel correlation coefficient Γ _b to the multichannel parametric decoder 107 as a cue parameter.

4 is a detailed block diagram of the multi-channel parametric decoding unit 107 of FIG. 1.

The single channel audio signal in raw data form is FFT-converted by the TF converter 401 to become a single channel audio signal in a frequency domain. The parameter-based multi-channel generator 403 generates a multi-channel audio signal in the frequency domain from the single channel audio signal in the frequency domain using the cue parameter information transmitted from the parameter analyzer 209. The multi-channel audio signal in the frequency domain is inverse FFT transformed by the inverse TF converter 405 for each channel to generate a multi-channel audio signal in the time domain.

FIG. 5 is a detailed block diagram of the parameter-based multichannel generator 403 of FIG. 4.

The correlation measurer 501 selects a corresponding subband parameter based generator 503, 505, 507 by comparing the frame unit channel correlation coefficient Γ transmitted from the parameter analyzer 209 with Table 1.

For example, when the channel correlation coefficient Γ is 0.9, the 10 subband parameter generator 503 divides the input single channel audio signal into 10 subbands according to Table 3. The 10 subband parameter generator 503 generates a multichannel audio signal with respect to the divided signal using the ICLD transmitted from the 10 subband parameter analyzer 303 and the subband unit channel correlation coefficient Γ _b . In this case, Equations 7,8, and 9 may be used.

는 c번째 채널 b번째 서브밴드의 n포인트 주파수 계수이다.

는 전력 이득 인자(power gain factor)로서 수학식 8로 표현된다.Where S _n ^b is the n point frequency coefficient of the b th subband of the downmixed single channel audio signal,

Is the n-point frequency coefficient of the c-th channel b-th subband.

Is a power gain factor represented by Equation (8).

r _{i, n} is a random variable to reflect the spatial feeling of the sound source generated between the channels. That is, although the channel signals to be restored must maintain a sense of space between each other, all channel signals are recovered from the downmix signal, thereby reducing the sense of space. In order to save the sense of space, a wider sense of space can be obtained by inserting random variable values into the process of calculating the power gain factor and adding randomness to the reconstructed channel signal.

r _{i, n} may be expressed as in Equation (9).

은 ±5 dB의 크기변화를 갖는 랜덤신호이다.

은 모든 채널과 서브밴드에 적용되며 동일한 분산과 제로 평균을 갖는다. here,

Is a random signal with a magnitude change of ± 5 dB.

Applies to all channels and subbands and has the same variance and zero mean.

The method for generating a multi-channel audio signal through the above Equations 7, 8, and 9 is well known as described in the above paper, and thus, further description thereof will be omitted.

The 19 subband parameter based generator 505 operates when the frame correlation coefficient Γ is 0.5 and 0.8, and generates a multichannel audio signal in the same manner as the 10 subband parameter based generator 503. .

The 32 subband parameter based generation unit 507 operates when the frame correlation coefficient Γ is 0 and 0.5. The 32 subband parameter based generation unit 507 divides the single channel audio signal into 32 subbands according to Table 4. The 32 subband parameter generator 507 generates a multichannel audio signal with respect to the divided signal using the channel ID transmitted from the 32 subband parameter analyzer 307 and the subband unit channel correlation coefficient Γ _b . .

If the channel ID for the subband b is c '(1≤c'≤C), the subband b signal of the c' channel is allocated by Equation 10.

S ^b _{c ', n} = S ^b _n

The other subband signals are generated by Equations 7, 11 and 12.

Here, r _{i, n} is a random variable for reflecting the spatial sense of the sound source generated between the channels.

은 ±5 dB의 크기변화를 갖는 랜덤신호이다.

은 모든 채널과 서브밴드에 적용되며 동일한 분산과 제로 평균을 갖는다. here,

Is a random signal with a magnitude change of ± 5 dB.

Applies to all channels and subbands and has the same variance and zero mean.

Where r _{i, n} is defined as

The multi-channel audio signal in the frequency domain generated by the subband parameter based generators 503, 505, and 507 is inversely FFT-converted by the inverse TF converter 405 for each channel to restore the multi-channel audio signal in the time domain.

The multi-channel parametric encoding and decoding process according to the present invention described above is summarized with reference to FIGS. 6 and 7 as follows. 6 is a flowchart illustrating a multi-channel parametric coding procedure according to the present invention.

The multi-channel audio signal is divided into 19 subbands according to Table 1 after FFT conversion (S601). The channel correlation coefficient Γ _b in units of subbands is measured with respect to the 19 subband split signal, and the channel correlation coefficient Γ in units of frames is measured (S603).

If the frame-by-frame channel correlation coefficient Γ is greater than 0.8, the 19 subband split signal is subdivided into 10 subband signals according to Table 3, and the channel correlation coefficient Γ _b in ICLD and subband units for 10 subbands is It extracts (S605, S607, S609). Subsequently, the cue parameter including the ICLD and subband unit channel correlation coefficient Γ _b extracted in step S609 and the frame unit channel correlation coefficient Γ measured in step S603 is transmitted to the multi-channel parametric decoder (S611).

When the frame unit correlation coefficient Γ is between 0.5 and 0.8, ICLDs are extracted for 19 subbands of the 19 subband split signals (S605, S607, and S613). Subsequently, the queue parameter including the ICLD extracted in step S613 and the subband unit channel correlation coefficient Γ _b and the frame unit channel correlation coefficient Γ measured in step S603 are transmitted to the multichannel parametric decoder (S615).

If the frame-by-frame channel correlation coefficient Γ is between 0 and 0.5, the 19 subband split signal is subdivided into 32 subband signals by Table 4, and the channel correlation coefficient of the channel ID and the subband unit for the 32 subbands is shown. Γ _b is extracted (S605, S617). Subsequently, the queue parameter including the channel ID and the subband unit channel correlation coefficient Γ _b extracted in step S617 and the frame unit channel correlation coefficient Γ measured in step S603 is transmitted to the multichannel parametric decoder (S619). ).

7 is a flowchart illustrating a multichannel parametric decoding procedure according to the present invention.

The single channel audio signal is FFT converted (S701). The frame unit channel correlation coefficient Γ transmitted from the multichannel parametric encoder 101 is observed (S703).

If the frame-by-frame channel correlation coefficient Γ is greater than 0.8, the FFT-converted signal is divided into 10 subband signals by Table 3, and the signal is transmitted by subband units by the transmitted ICLD and sub-band unity channel correlation coefficient Γ _b . Is decoded (S705, S707, S709).

If the frame-by-frame channel correlation coefficient Γ is between 0.5 and 0.8, the FFT-converted signal is divided into 19 subband signals by Table 1, and the subband-by-band unit by the transmitted ICLD and subband unit channel correlation coefficients Γ _b . The signals are decoded for each other (S705, S707, S711).

When the frame unit correlation coefficient Γ is between 0 and 0.5, the FFT transformed signal is divided into 32 subband signals by Table 4, and the subbands are transmitted by the transmitted channel ID and the subband unit channel correlation coefficient Γ _b . The signal is decoded in units (S705 and S713).

The method of the present invention as described above may be implemented in a program and stored in a recording medium in a computer-readable form.

 As described above, although the present invention has been described by means of a limited embodiment and drawings, the present invention is not limited by this and the technical spirit of the present invention and the following by those skilled in the art to which the present invention pertains. Various modifications and variations are possible without departing from the scope of the claims to be described in the following.

The present invention provides an effect of restoring and reproducing a multichannel audio signal from a single channel audio signal even in an environment where a small bandwidth is allocated by optimizing a cue parameter required for decoding a multichannel audio signal.

Claims (32)

Signal dividing means for dividing the multi-channel audio signal in the frequency domain; Downmixing means for downmixing the divided signals into a single channel and encoding the divided signals; And A parameter analysis means for measuring a channel correlation in units of frames with respect to the divided signal, and re-dividing the divided signals according to the channel correlation in units of frames; The parameter analysis means, A channel parameter for decoding when the frame unit channel correlation is less than or equal to a first predetermined value, and extracts and encodes a channel ID and a subband unit channel correlation for the subbands of the subdivided signal. Multi-channel audio signal encoding apparatus. The method of claim 1, The parameter analyzing means Re-segmenting the divided signal into an increasing number of subbands as the frame-by-frame channel correlation decreases. Multi-channel audio signal encoding apparatus. The method of claim 1, The frame correlation is obtained from the channel correlation in subbands for the divided signals. Multi-channel audio signal encoding apparatus. The method of claim 3, wherein The frame unit channel correlation is determined by the subband unit channel correlation of the low frequency band rather than the subband unit channel correlation of the high frequency band. Multi-channel audio signal encoding apparatus. The method of claim 1, The channel ID for the subdivided subband b is Is the index of the channel at which the power of the subdivided subband b is maximum Multi-channel audio signal encoding apparatus. The method of claim 1, The parameter analyzing means When the frame unit channel correlation is greater than or equal to the first predetermined value, an inter-channel level difference and a subband unit channel correlation are extracted for each subband of the subdivided signal. Coded Multi-channel audio signal encoding apparatus. The method according to claim 1 or 6, The parameter analyzing means As the cue parameter, an inter-channel level difference is additionally extracted for each subband of the subdivided signal. Multi-channel audio signal encoding apparatus. The method of claim 1, The parameter analyzing means When the frame correlation is less than or equal to the first predetermined value, the divided signal is re-divided into a number greater than the number of subbands of the divided signal, When the frame unit channel correlation is greater than or equal to the second predetermined value greater than the first predetermined value, the divided signal is re-divided into a number smaller than the number of subbands of the divided signal, When the frame unit channel correlation is between the first predetermined value and the second predetermined value, no subdivision is performed. Multi-channel audio signal encoding apparatus. The method of claim 8, The parameter analyzing means When the frame unit channel correlation is between the first predetermined value and the second predetermined value, an inter-channel level difference and a sub-channel frequency difference for each subband of the divided signal as the cue parameter. Extracting and coding channel correlation Multi-channel audio signal encoding apparatus. The multi-channel audio signal divided in the frequency domain is downmixed into a single channel and encoded, and the divided audio signal is re-divided according to the channel correlation in the frame unit for the divided multi-channel audio signal. In the multi-channel audio signal decoding apparatus for decoding into an audio signal, Single channel decoding means for decoding the encoded single channel audio signal; And And parameter-based multichannel generation means for dividing the decoded single channel audio signal in a frequency domain in the same manner as the subband interval used in the re-division, according to the frame unit channel correlation. The parameter-based multi-channel generating means, When the frame unit channel correlation is less than or equal to a first predetermined value, the signal is decoded into the single channel audio signal using a channel ID and a subband unit channel correlation of the subband of the re-divided signal. Multi-channel audio signal decoding device. The method of claim 10, The parameter based channel generating means Dividing the decoded single channel audio signal into an increasing number of subbands as the frame unit channel correlation decreases. Multi-channel audio signal decoding device. The method of claim 10, The channel ID for the subdivided subband b is Is the index of the channel at which the power of the subdivided subband b is maximum Multi-channel audio signal decoding device. The method of claim 10, The parameter based channel generating means When the frame unit channel correlation is greater than or equal to the first predetermined value, decoding is performed by using a subband-specific channel correlation and a subband-based channel correlation between the subdivided signals. Multi-channel audio signal decoding device. The method of claim 10 or 13, The parameter based channel generating means Decoding by additionally using a time delay difference between channels of each subband of the re-divided signal Multi-channel audio signal decoding device. The method of claim 12, The parameter based channel generating means If the channel ID for the subband b is c '(1≤c'≤C, C is the number of channels), the subband b signal of the c' channel is allocated to the subband b signal of the single channel audio signal. Multi-channel audio signal decoding device. The method of claim 15, The parameter based channel generating means The single channel audio signal is allocated to the remaining subband signals by reflecting components related to the spatial feeling of the sound source generated between channels. Multi-channel audio signal decoding device. Dividing the multichannel audio signal in a frequency domain; Down-mixing the divided signal into a single channel and encoding the divided signal; Measuring a channel correlation in units of frames with respect to the divided signal; Repartitioning the divided signal according to the channel correlation in the frame unit; And Extracting and encoding a channel ID and a subband unit channel correlation for the subbands of the re-divided signal when the frame unit channel correlation is less than or equal to a first predetermined value. Multi-channel audio signal coding method. The method of claim 17, The repartition stage Re-segmenting the divided signal into an increasing number of subbands as the frame-by-frame channel correlation decreases. Multi-channel audio signal coding method. The method of claim 17, The frame correlation is obtained from the channel correlation in subbands for the divided signals. Multi-channel audio signal coding method. The method of claim 19, The frame unit channel correlation is determined by the subband unit channel correlation of the low frequency band rather than the subband unit channel correlation of the high frequency band. Multi-channel audio signal coding method. The method of claim 17, The channel ID for the subdivided subband b is Is the index of the channel at which the power of the subdivided subband b is maximum Multi-channel audio signal coding method. The method of claim 17, Extracting and encoding an inter-channel level difference and a subband unit channel correlation for each subband of the subdivided signal when the frame unit channel correlation is greater than or equal to the first predetermined value. More containing Multi-channel audio signal coding method. The method of claim 17 or 23, Extracting an inter-channel level difference for each subband of the re-divided signal; Multi-channel audio signal coding method. The method of claim 17, The repartitioning step Subdividing the divided signal into a number greater than the number of subbands of the divided signal when the frame unit channel correlation is less than or equal to the first predetermined value; Subdividing the divided signal into fewer than the number of subbands of the divided signal when the frame unit channel correlation is greater than or equal to the second predetermined value greater than the first predetermined value; And If the frame-by-frame channel correlation is between the first predetermined value and the second predetermined value, not re-dividing. Multi-channel audio signal coding method. The method of claim 24, If the frame unit channel correlation is between the first predetermined value and the second predetermined value, extracting and encoding an inter-channel frequency magnitude difference and a subband unit channel correlation for each subband of the divided signal; Containing more Multi-channel audio signal coding method. The multi-channel audio signal divided in the frequency domain is downmixed into a single channel and encoded, and the divided audio signal is re-divided according to the channel correlation in the frame unit for the divided multi-channel audio signal. In the multi-channel audio signal decoding method for decoding into an audio signal, Decoding the encoded single channel audio signal; Dividing the decoded single channel audio signal in a frequency domain in the same manner as the subband interval used in the re-division according to the frame-by-frame channel correlation; And Decoding the single channel audio signal using a channel ID and a subband unit channel correlation for the subband of the subdivided signal when the frame unit channel correlation is less than or equal to a first predetermined value. Multi-channel audio signal decoding method. The method of claim 26, The dividing step Dividing the decoded single channel audio signal into an increasing number of subbands as the frame unit channel correlation decreases. Multi-channel audio signal decoding method. The method of claim 26, The channel ID for the subdivided subband b is Is the index of the channel at which the power of the subdivided subband b is maximum Multi-channel audio signal decoding method. The method of claim 26, If the frame unit channel correlation is greater than or equal to the first predetermined value, decoding the subdivided signal using the sub-channel frequency magnitude difference between the subbands and the subband unit channel correlation. Multi-channel audio signal decoding method. The method of claim 26 or 29, The method may further include decoding using a time delay difference between channels of each subband of the re-divided signal. Multi-channel audio signal decoding method. The method of claim 28, If the channel ID for the subband b is c '(1≤c'≤C, C is the number of channels), the subband b signal of the c' channel is allocated to the subband b signal of the single channel audio signal. Multi-channel audio signal decoding method. The method of claim 31, wherein The remaining subband signals are assigned to the single channel audio signal by reflecting components of the spatial sense of the sound source generated between channels. Multi-channel audio signal decoding method.

Images