KR0183778B1 - Bit allocation compatible with dolby a-3 encoding algorithm - Google Patents

Abstract

The present invention discloses a bit allocation method compatible with the Dolby AC-3 encoding algorithm.

The present invention provides a method of performing bit allocation by comparing a masking value with values of parameters of a plurality of channels (csnroffst, 1fesnroffst, fesnroffst, and cp1snroffst), and ending the bit allocation when the flag End_Flag is '1'; If the flag (End_Flag) for terminating bit allocation is not '1', the first parameter is changed after changing the parameter value (csnroffst) using the parameter value (csnroffst) of the channel with a large data size and the maximum value (MAXCSNROFFST) of the system. A second process of performing a process; And if the bit remaining after the bit allocation is performed in the second process, the parameter (fsnroffst) is changed by using the parameter (fsnroffst) of the channel having the small data size and the maximum value (MAXFSNROFFST) of the system. It characterized in that it comprises a third process of performing.

Accordingly, the present invention can perform optimal bit allocation by changing a parameter value of an encoder according to a channel type using a masking curve value in a bit allocation apparatus of a Dolby AC-3 audio algorithm.

Description

Bit allocation method compatible with Dolby AC-3 encoding algorithm

1 is a flow diagram of a Dolby AC-3 encoding algorithm.

FIG. 2 is a detailed flowchart of Core Bit Allocation shown in FIG. 1. FIG.

3a and b are flowcharts showing bit allocation compatible with the Dolby AC-3 encoding algorithm according to the present invention.

The present invention relates to a bit allocation method compatible with the Dolby AC-3 encoding algorithm. In particular, in the bit allocation apparatus for the Dolby AC-3 audio algorithm, the parameter value of the encoder is changed using a masking curve value to obtain an optimal bit allocation. A bit allocation method compatible with Dolby AC-3 encoding algorithm is achieved.

The AC3 audio algorithm, announced by Dolby Laboratories, is used in digital video discs (DVDs) and ATVs, and up to six channels can be used in conventional methods, with much less than the amount of bits required by one channel of the digital audio signal. Audio signal can be transmitted. Currently, the AC-3 audio algorithm is gradually expanding its application, and will be adopted in many media such as digital video disc (DVD) and ATV.

Codecs, including MPEG, perform bit allocation on the encoder and then transmit the result so that the decoder does not perform any calculation for bit allocation. In contrast, in Dolby AC-3, both the encoder and the decoder perform calculation for bit allocation, and parameters necessary for the calculation for bit allocation are determined by the encoder and transmitted to the decoder. However, in the AC-3 algorithm, only the decoder is disclosed, and the bit allocation routine of the encoder is not disclosed.

Accordingly, the present invention provides a bit allocation method compatible with Dolby AC-3 encoding algorithm that performs optimal bit allocation while changing parameter values of an encoder using masking curve values in a bit allocation apparatus for Dolby AC-3 audio algorithm. The purpose is to provide.

In order to achieve the above object, a bit allocation method compatible with the Dolby AC-3 encoding algorithm according to the present invention performs bit allocation by comparing a masking value with a parameter value of a plurality of channels (csnroffst, 1fesnroffst, fesnroffst, cp1snroffst), Terminating the bit allocation if the flag End_Flag is '1'; If the flag (End_Flag) for terminating bit allocation is not '1', the first parameter is changed after changing the parameter value (csnroffst) using the parameter value (csnroffst) of the channel with a large data size and the maximum value (MAXCSNROFFST) of the system. A second process of performing a process; And if the bit remaining after the bit allocation is performed in the second process, the parameter (fsnroffst) is changed by using the parameter (fsnroffst) of the channel having the small data size and the maximum value (MAXFSNROFFST) of the system. It characterized in that it comprises a third process of performing.

A bit allocation method compatible with the Dolby AC-3 encoding algorithm, the first process comprising: a first step of performing bit allocation by comparing a predetermined parameter value and a masking value; A second step of calculating the total amount of bits allocated in the first step; And a third step of determining whether the flag End_Flag for terminating bit allocation is '1'.

In addition, the second process includes a first step of determining whether the total amount of allocated bits is greater than the total amount of bits that can be used; Performing a first process after increasing C_End_Flag if C_End_Flag is not '1' and the csnroffst value and the MAXCSNROFFST value are not equal to the total amount of allocated bits, if the total amount of allocated bits is not greater than the total amount of available bits; A third step of performing the third process after setting C_End_Flag to '1' if C_End_Flag is not '1' and the csnroffst value and the MAXDSNROFFST value are the same if the total amount of allocated bits is not greater than the total amount of available bits; A fourth step of performing the third process if C_End_Flag is '1' when the total amount of allocated bits is not greater than the total amount of available bits; If the total amount of allocated bits is greater than the total number of available bits, if the C_End_Flag is not '1', reducing the csnroffst value, setting C_End_Flag to '1', and then performing the third process; And when the total amount of allocated bits is greater than the total number of available bits, if C_End_Flag is '1', a sixth step of performing a first process by decreasing the last increased fsnroffst sequence value and setting _End_Flag to '1'; It is preferable to include.

In addition, the third process is a first step of performing a first process after increasing the fsnroffst value if the 1fesnroffst value is not equal to the MAXSNROFFST value and not greater than the maximum value of the fsnroffst value when there is an LFE channel; If there is no LFE channel and there is no coupling channel, bit allocation is terminated when all fsnroffst values are equal to MAXFSNROFFST value.If all fsnroffst values are not equal to MAXFSNROFFST value, the fsnroffst value of the channel with the smallest index value among the smallest fsnroffst values is increased. A second step of performing a first process afterwards; If there is no LFE channel and there is a coupling channel, if the cp1snroffst value is smaller than the minimum value of the fsnroffst value, it is preferable to increase the cp1snroffst value and then perform the first process.

Hereinafter, with reference to the accompanying drawings will be described in detail with respect to the present invention.

Figure 1 is a flow diagram of the Dolby AC-3 encoding algorithm, which is described as follows.

First, when PCM data of multi-channel is input, it analyzes whether there is a drastic change in the input signal in the TRANSIENT DETECTION, and then changes the transformer size in the FORWARD TRANSFORM according to the result. In FORWARD TRANSFORM, multi-channel signals in the time domain are converted into multi-channel signals in the frequency domain. In COUPLING STRATEGE, similar signals are tied together in the FORM COUPLING CHANNEL by examining the components of each channel in the frequency domain. In addition, REMATRIX mixes the given signals as needed, and EXTRACT EXPONENTS separates only the EXPONENT part from the result of REMATRIX. EXPONENT STRATEGE looks at the distribution of separated EXPONENT values and determines how to code EXPONENT values. DITHER STRATEGE adds DITHER as needed. Using the above result, the part coding the EXPONENT is NORMALIZE MANTISSA.

In CORE BIT ALLOCATION, if you use EXPONENT ENCODING and code MANTISSA using the remaining bits, it determines how many bits are best used for each frequency band of each channel. Using this result, the MANTISSA is ENCODING in the QUANTIZE MANTISSA, and the EXPONENT values and other information coded above are combined to form an AC-3 frame. The present invention proposes a method for optimally performing bit allocation in CORE BIT ALLOCATION.

In Figure 1, there is a part that calculates an exponent value. This part divides the value of the frequency domain passing through the filter by mantissa and exponent and then removes the exponent. Looking at the exponent value, it is possible to know the size distribution of each value, and bit allocation is performed as shown in FIG. 2 by using the exponent value.

FIG. 2 is a detailed flowchart of Core Bit Allocation shown in FIG. 1 and will be described below.

After initializing the parameters of each channel (1), the exponent value is used to calculate the PSD value to determine what size structure the values of each frequency domain have. (2) After that, considering that the human ear listens based on a certain band, each PSD value is summed to obtain a PSD value in units of bands. (3) The exit function is calculated from the PSD value of the band unit. (4) The exitation function superimposes the prototype masking curve on the PSD value of the current band unit. The masking curve in the range is obtained and the final masking curve is obtained by comparing the result with the hearing threshold of the person. (5) Then, optimal bit allocation is performed while changing the snroffst value using the final masking curve value. (6,7)

3A and 3B are a flowchart of a bit allocation method compatible with the Dolby AC-3 encoding algorithm according to the present invention, in which bit allocation is performed by comparing a masking value with a plurality of channel parameter values (csnroffst, 1fesnroffst, fesnroffst, and cp1snroffst). Performing a step of terminating the bit allocation if the flag End_Flag is '1'; If the flag (End_Flag) for terminating bit allocation is not '1', the first parameter is changed after changing the parameter value (csnroffst) using the parameter value (csnroffst) of the channel with a large data size and the maximum value (MAXCSNROFFST) of the system. A second process of performing a process; And if the bit remaining after the bit allocation is performed in the second process, the parameter (fsnroffst) is changed by using the parameter (fsnroffst) of the channel having the small data size and the maximum value (MAXFSNROFFST) of the system. It includes a third process of performing.

Here, the coupling channel and the LFE channel are different from the general independent channel, and the Fsnroffst value used for the coupling channel to treat them separately from each other is called the cp1snroffst value, and the Fsnroffst value used for the LFE channel is called 1fesnroffst. .

In addition, C_End_Flag is a flag indicating that the adjustment of the Csnroffst value is finished, and the MAXCSNROFFST value indicates the maximum value of snroffst supported by AC-3 Syntax, and the MAXFSNROFFST value indicates the maximum value of Fsnroffst supported by AC-3 Syntax.

Referring to the present invention in more detail, as follows.

Bit allocation in the Dolby AC-3 audio algorithm is determined by the masking values and the snroffst of each channel. The basis of this bit allocation algorithm is to make the noise level caused by quantization smaller than the masking level. And the noise level caused by quantization is closely related to snroffst value. In other words, if the snroffst value of the corresponding channel is large, detailed coding should be performed, and eventually, a lot of bits should be used to reduce the noise level caused by quantization.

Even though the audio signal comes from several channels, each signal is related to each other, and the parameters of each channel (hereinafter referred to as snroffst) have a common parameter (hereinafter referred to as Csnroffst) and the unique parameters of each channel. Can be divided into values (hereinafter referred to as Fsnroffst). Here, the Csnroffst value is larger in width than the Fsnroffst value of each channel.

In other words, when adjusting the snroffst value, first adjust the overall level of the multi-channel by adjusting the Csnroffst value, and then adjust the Fsnroffst value for each channel. The adjustment of Csnroffst value can be seen as COARSE-snroffst control, and the adjustment of Fsnroffst value can be seen as FINE-snroffst control.

In the third process, the Fsnroffst value processing routine (①) is executed. Even if one Fsnroffst value is increased, the influence on the snroffst value is smaller than the csnroffst value. The value of fsnroffst is increased in order of 1fesnroffst value, fsnroffst value, and cp1snroffst value.

First, determine if there is an LFE channel (1D), if there is an LFE channel, if the 1fesnroffst value is not equal to the MAXFSNROFFST value (2D), and if the 1fesnroffst value is not greater than the maximum value of fsnroffst (3D), then increase the 1fesnroffst value by one (4D). Perform the first step. (②)

The reason why the value of the LFE channel is increased first is that the data of the channel due to the coding noise of the channel may be buried because the size of the data of the LFE channel is larger than that of other channels.

If the 1fesnroffst value is not equal to the MAXFSNROFFST value in step (2D) or if the 1fesnroffst value is not greater than the maximum value of fsnroffst in step (3D), the coupling channel is compared. (1E)

After increasing the value of 1fesnroffst (4D) and performing bit allocation, if there are available bits left, it is determined whether there is a coupling channel (1E) .If all fsnroffst values are equal to the MAXFSNROFFST value, bit allocation ends. If the fsnroffst value is not equal to the MAXFSNROFFST value (2E), and all fsnroffst values are not equal (3E), increase the fsnroffst value of the channel with the smallest index value among the smallest fsnroffst values (4E), and then proceed to Step 1. Perform. If all fsnroffst values are the same in step (3E), the first process is performed after increasing the fsnroffst value of channel 0 by one (4E).

The reason for increasing the value of fsnroffst of the smallest channel in step 4E is that the Dolby AC-3 encoding algorithm allocates the most important channel to the channel with the smallest index. In this way, if the fsnroffst value of all indexes is increased, then the cp1snroffst value is increased.

That is, if there is a coupling channel in step (1E), if cp1snroffst is not less than the minimum value of fsnroffst, perform step (2E) .If cp1snroffst is smaller than the minimum value of fsnroffst, increase cp1snroffst and then perform bit allocation. do.

The reason why the coupling channel is out of sequence with other channels is that independent channels that are not included in the coupling have different notes from other channels, and these notes are easily picked up by the ear.

Increasing the value of Fsnroffst in this way allocates more bits than the total number of bits available at any given time. At this time, decrease the last increased fsnroffst value by one (4C), then set End_Flag to 1 ( 5C) Final bit allocation. (2A)

The present invention achieves optimal bit allocation while varying the parameter values of the encoder. That is, the final product of the present invention is an optimal parameter value and bit allocation accordingly, and the decoder receives this parameter value to obtain bit allocation information of the result as in the encoder.

As described above, the bit allocation method compatible with the Dolby AC-3 encoding algorithm according to the present invention is optimized by changing the parameter values of the encoder in the bit allocation apparatus for the Dolby AC-3 audio algorithm according to the type of channel. Bit allocation can be performed.

Claims (4)

Performing a bit allocation by comparing the masking value with the values of the parameters of the plurality of channels (csnroffst, 1fesnroffst, fesnroffst, cp1snroffst), and ending the bit allocation when the flag End_Flag to terminate bit allocation is '1'; If the flag (End_Flag) for terminating bit allocation is not '1', the parameter (csnroffst) is changed by using the parameter value (csnroffst) of the channel having a large data size and the maximum value (MAXCSNROFFST) of the system. A second process of performing a first process; And changing the parameter value fsnroffst using the parameter value fsnroffst of the channel with the small data size and the maximum value MAXFSNROFFST if there is a bit remaining after the bit allocation in the second step. A bit allocation method compatible with Dolby AC-3 encoding algorithm comprising a third step of performing one step. The method of claim 1, wherein the second process comprises: a first step of performing the first process after increasing the fsnroffst value if the 1fesnroffst value is not equal to the MAXFSNROFFST value and not greater than the maximum value of the fsnroffst value when there is an LFE channel; If there is no LFE channel and there is no coupling channel, bit allocation is terminated when all fsnroffst values are equal to MAXFSNROFFST value.If all fsnroffst values are not equal to MAXFSNROFFST value, the fsnroffst value of the channel with the smallest index value among the smallest fsnroffst values is increased. A second step of performing the first process thereafter; And if there is no LFE channel and there is a coupling channel, if the cp1snroffst value is smaller than the minimum value of the fsnroffst value, a third step of performing the first step after increasing the cp1snroffst value is compatible with the Dolby AC-3 encoding algorithm. Bit allocation method. 3. The bit allocation method of claim 2, wherein the first step determines whether there is a coupling channel when the value of 1fesnroffst is equal to the MAXFSNROFFST value or greater than the maximum value of the fsnroffst value. 3. The method of claim 2, wherein the second step is compatible with a Dolby AC-3 encoding algorithm that increases the fsnroffst value of channel 0 if all fsnroffst values are equal.