KR0124384B1 - Variable bit assigning device in audio - Google Patents

Abstract

The present invention provides a variable bit allocating system according to kinds of audio broadcasting which restricts the compression ratio of the audio data and the band according to the kinds of audio broadcasting and the tone quality, and allocates the excess of the remaining bit number to video data. The inventive system includes a high speed fourier transformation and threshold computing part computing a masking threshold that does not sense noise according to the broadcasting mode selection information; a bit allocating part which allocates bits with respect to each divided band signal within a range where quantizing noise does not exceed the masking threshold; a first quantizing part quantizing divided band signals in response to the bit allocating part's output; a second quantizing part quantizing adaptively video data; a fullness computing part computing fullness, the degree of storing video data in a buffer; a multiplexing part varying and multiplexing the bit stream of the audio data and video data processed according to the broadcasting mode.

Description

Variable bit allocation device according to audio broadcasting type

1 is a graph showing bandwidth according to a sound source according to a conventional variable bit allocation coding scheme.

2 is a block diagram of a variable bit allocation device according to the audio broadcasting type of the present invention.

FIG. 3 is a bandwidth state diagram showing split characteristics of the split band filter group in FIG.

4 is a graph showing an example of a variable masking threshold according to the broadcast mode selection of the present invention, (a) a masker for each frequency band, (b) a masking threshold curve corresponding to mode 1, and (c) Is a masking threshold curve corresponding to mode 2, and (d) is a masking threshold curve corresponding to mode 3.

5 is a state diagram showing the amount of data in a buffer that changes according to the number of bits saved by the present invention.

6 is a configuration diagram of a frame unit of audio data encoded according to the present invention.

* Explanation of symbols for the main parts of the drawings

10: split band filter group 30: bit allocation unit

20: fast Fourier transform and threshold calculation unit 40,50: quantization unit,

60: buffer 70: fullness calculation unit

80: multiplexer.

According to the present invention, the compression rate of audio data and the band to which bits are allocated are limited according to the types of audio broadcasts, thereby limiting the number of bits to which audio data is allocated, and additionally assigning an extra amount of extra bits to video data, thereby reproducing video quality. The present invention relates to a variable bit allocation device according to one type of audio broadcasting.

In general, the human auditory characteristics can hear sound waves with frequency components up to about 20KH _Z. As a result, existing digital audio products such as Compact Disk Player (CDP) or Digital Audio Tape (DAT) recorders can use analog audio signals using a sampling frequency of 40KH _Z and 16-bit resolution. After sampling, it is encoded and converted to a digital audio signal for recording or transmission on a medium. When the digital audio signal recorded or transmitted on the medium is reproduced as described above, the audio signal of 96dB dynamic range and wideband can be reproduced, which is an operating range capable of processing the signal without distortion compared to the analog method.

However, recording and reproducing or transmitting and receiving sampled digital audio signals requires 700 Kbps or more per channel. Depending on the application, the amount of data required is excessive, which makes it difficult to actually adapt the digital method to the product. In order to solve this problem, various coding schemes have been proposed for performing data compression more than four times while maintaining the sound quality of an audio signal. One of them is a variable bit allocation encryption scheme.

FIG. 1 is a graph showing bandwidths of sound sources according to a conventional variable bit allocation coding scheme, and bandwidths are variably allocated according to specific gravity and sources of sound quality. According to this graph, the human speech is assigned the band to 3.5KH _Z, and the case of AM broadcast signal in the case of _{Z 7KH,} FM broadcast signal, it gives each allocated the bands to 15KH _Z, Digital Audio Products In this case, the 20KH _Z band is guaranteed for the reproduction of high quality audio signals.

As such, the final frequency band to be allocated is selectively determined, and when the bit is allocated below the final frequency band, a masking threshold, which is an upper limit level that is not detectable by humans even if noise is present due to auditory characteristics, is calculated. Bits are allocated to each subband within a range in which noise existing when quantizing the digital audio signal does not exceed the masking threshold. The data bit allocation method calculates a mask to noise ratio (MNR) of any number of divided bands, allocates bits from the divided band having the lowest MNR value among them, and calculates the MNR to obtain the lowest MNR. It is a traversal method of finding a divided band and allocating bits in the divided band, which continues until all the bits allocated to the frame are exhausted. In this way, the number of bits allocated to each divided band is determined.

In the case of a system which compresses, records, reproduces, transmits, and receives audio and video signals, for example, multimedia, digital VCR, and digital broad-casting system (DBS), In this case, a much wider frequency band is allocated to the video signal than the audio signal. However, since the frequency band assigned to the video signal is determined, there is a limit to the improvement of the picture quality when the digital video signal recorded or transmitted on the medium is reproduced.

On the other hand, the bandwidth required for the audio signal can be varied according to the type of sound. For example, the announcement of the announcer in the news of the TV does not have high frequency components and does not require high sound quality, thereby reducing the band allocated to the bit at the time of beat. In addition, the actor's dialogue in the TV drama has more high frequency components than the announcement of the announcer, but since it does not require high sound quality, it is possible to limit the band to which the beat is allocated like the announcement of the announcer. In the case of music, high quality is required, so as in the past, the band to which the beat is allocated is set wide. In this way, the bandwidth to which bits are allocated can be appropriately reduced according to the type of audio signal and the specific gravity of sound quality, so that an extra portion of the number of bits remaining here can be allocated to the video signal.

Accordingly, an object of the present invention is to reduce the number of bits to which audio data is allocated according to the type of audio broadcasting and the specific gravity of sound quality, thereby improving the image quality of the video signal by additionally assigning an extra bit to the video data. The present invention provides a variable bit allocation device according to an audio broadcast type.

The variable bit allocation apparatus of the present invention includes a split band filter group for dividing the sampled audio data into a predetermined number of split band signals, and the split band filter group includes a fast Fourier transform (FFT) and a threshold calculation unit. Connected. The fast Fourier transform and threshold calculation unit receives broadcasting mode selection information, sampled audio data, and divided band signals of the divided band filter group, and performs fast spectrum Fourier transform on the audio data to perform spectrum analysis to detect noise due to auditory characteristics. Calculate. The bit allocator calculates bits to which bits are allocated according to broadcast mode selection information, and the first quantization unit quantizes each divided band signal according to the output of the bit allocator. In addition, the second quantization unit quantizes the video data according to the calculated fullness, and the buffer connected to the second quantization unit stores the video data input at a variable ratio and outputs it at a fixed ratio. The fullness calculator connected between the second quantizer and the buffer calculates the fullness, which is a degree of data stored in the buffer, according to the broadcast mode selection information, and applies it to the second quantizer. The multiplexer multiplexes the audio data output from the first quantizer and the video data output from the buffer according to the broadcast mode selection information and the output signal of the bit assignment unit into an appropriate data format.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to FIGS. 2 to 6 as follows.

2 is a block diagram of a variable bit allocation apparatus according to the type of audio broadcasting of the present invention. As shown in FIG. 2, the variable bit allocation apparatus of the present invention includes a divided band filter group 10 for dividing the sampled audio data into N divided band signals. A fast Fourier transform and threshold calculation unit 20 and a first quantizer 40 are connected to the output of the split band filter group 10. The fast Fourier transform and threshold calculation unit 20 receives the broadcast mode selection information, the original sampled audio data, and the divided band signals output from the split band filter group 10, and performs spectral analysis by fast Fourier transforming the audio data. Calculate a masking threshold that cannot detect noise due to auditory characteristics. A bit allocator 30 is connected to the fast Fourier transform and threshold calculating unit 20. The bit allocator 30 is assigned to each divided band divided by the divided band filter group 10 according to broadcast mode selection information. On the other hand, the data bits are allocated such that the masking threshold value output from the fast Fourier transform and threshold calculation unit 20 does not exceed the quantization noise as much as possible. The output of the bit allocator 30 is applied to the first quantizer 40 and the multiplexer 80, and the first quantizer 40 is divided using the bits allocated by the bit allocator 30. The divided band signals output from the band filter group 10 are quantized. On the other hand, the variable bit allocation apparatus of the present invention includes a second quantizer 50 for adaptively quantizing the video data according to the fullness output from the fullness calculator 70. A buffer 60 is connected to the output terminal of the second quantizer 50 to store video data input at a variable rate from the second quantizer 50 and output the same at a fixed ratio. The fullness calculator 70 connected between the buffer 60 and the second quantizer 50 receives the broadcast mode selection information and the output signal of the buffer 60 to calculate the fullness which is the data stored in the buffer. And output to the second quantization unit 50. In addition, the apparatus of the present invention is suitable for the audio data output from the first quantizer 40 and the video data output from the buffer 60 according to the broadcast mode selection information and the output signal of the bit allocation unit 30. Also provided with a multiplexer 80 for multiplexing and outputting in a format.

The operation of the variable bit allocation apparatus according to the type of audio broadcasting of the present invention configured as described above will be described with reference to FIGS. 3 to 6 as follows.

When an analog audio signal for recording or transmitting to a medium is input to a coding apparatus of a variable bit allocation method, the analog audio signal is sampled at a predetermined sampling frequency (Fs), converted into a digital audio signal, and then divided into a split band filter group ( 10) and fast Fourier transform and threshold calculation unit 20, respectively. In the split band filter group 20, the input digital audio signal is divided into a predetermined number N of split band signals. At this time, each divided bandwidth is set constant in the existing Moving Picture Expert Group (MPEG) algorithm, but the present invention is not limited.

3 is a bandwidth state diagram showing the split characteristics of the divided band filter group, and shows an example of variable bandwidth allocation according to the broadcast mode selection of the present invention. In the example of FIG. 3, three broadcast modes are defined to allocate bandwidth for each mode, but the number of broadcast modes in the present invention can be set to any number. Mode 1 shows the bandwidth corresponding to the announcement of the announcer in the news. Mode 2 shows the bandwidth corresponding to the actor's dialogue in the drama, and Mode 3 shows the bandwidth corresponding to the music. Here, each bandwidth is determined in division band units, and represents the division bands to which bits are allocated when bit allocation.

The split band singular codes output from the split band filter group 10 are applied to the fast Fourier transform and the threshold calculator 20 and the first quantizer 40, respectively. The fast Fourier transform and threshold calculation unit 20 receives the original sampled audio data and the divided band signals, performs spectrum analysis by fast Fourier transform, and uses broadcast mode selection information and human auditory characteristics. The masking threshold, which is the upper limit level that humans cannot detect, even if noise is present due to auditory characteristics, is calculated. Here, the broadcast mode selection information is information about an audio broadcast mode arbitrarily set as in the example of FIG. 3 and is provided by a control unit included in the encoding apparatus of the present invention.

4 is a graph showing an example of a variable masking threshold of each frequency band according to the broadcast mode selection of the present invention. (A) of FIG. 4 shows the characteristics of the divided band filter group and a masker for each frequency band of the spectral line. Here, the term "mask" is the same as that expressed in MPEG, and is calculated by the method proposed by MPEG. (B)-(d) of FIG. 4 shows a masking threshold curve according to the broadcast mode selection of the present invention, (b) is a masking threshold curve corresponding to mode 1 in FIG. 3, and (c) is mode 2 , (D) is a masking threshold curve corresponding to mode 3.

As shown in FIG. 4, the calculation of the masking threshold should be similarly restricted according to the broadcast mode, i.e., by limiting the divided band for bit allocation. The maskers considered in the masking threshold calculation by the masker should be those present in the divided bands to which the bits are allocated. If the masker is considered over the entire divided band without such limitation, the total masking threshold is affected by the masker of the split band that is not actually transmitted, and thus an incorrect MNR value is calculated when calculating the MNR. Bit allocation is performed.

The masking threshold calculated by the fast Fourier transform and threshold calculation unit 20 is applied to the bit allocator 30. When the bit allocator 30 is quantized with reference to broadcast mode selection information applied from the control unit, The bits allocated to each divided band are calculated within the range in which the existing noise does not exceed the masking threshold. The first quantizer 40 uses the number of bits allocated to each divided band provided by the bit allocator 30 to output audio signals of the frequency bands output from the divided band filter group 10, that is, divided band signals. Quantize. The quantized audio signal is supplied to the multiplexer 80.

On the other hand, the analog video signal for recording or transmitting to the medium, like the audio signal is sampled and converted into a digital video signal, coded in any manner and applied to the second quantizer 50. The second quantizer 50 adaptively quantizes the input video signal according to the fullness of the buffer applied from the fullness calculator 70 and outputs the video signal to the buffer 60. The buffer 60 is a storage means for storing the video data input at a variable rate from the second quantizer 50 and outputting at a fixed rate. The fullness calculator 70 connected between the buffer 60 and the second quantizer 50 serves to provide information necessary for such bit allocation. To this end, the fullness calculation unit 70 receives the broadcast mode selection information and the output signal of the buffer 60 from the control unit and calculates the buffer fullness which is the degree of video data stored in the buffer. In addition, the calculated value is supplied to the second quantization unit 50 so that quantization is adaptively performed according to the fullness.

5 shows the data amount of a buffer to be converted according to the number of bits saved by the application of the variable bit allocation method by selecting the broadcast mode of the audio signal of the present invention. As described above, the divided bands to which bits are not allocated are determined according to the broadcast mode, and accordingly, the number of bits allocated per frame is determined according to the amount of bits to be saved. The amount of bits saved in this way determines the number of bits allocated per frame. The amount of bits thus saved is denoted by M (depending on the mode) in FIG. In FIG. 5, areas A and B are actually video bit streams stored in a buffer. If the divided band to which bits are allocated in accordance with the broadcast mode is variably limited, an audio signal band corresponding to the number of bits saved can be allocated to the video signal band. Therefore, the fullness of the buffer actually applied in the fullness calculator 70, that is, the virtual buffer fullness is calculated. At this time, the virtual buffer fullness B 'becomes C'-M.

The multiplexer 80 multiplexes the audio data output from the first quantizer 50 and the video data output from the buffer 60 into a data format of an appropriate type. In this case, since the length of one frame of the audio data varies depending on the broadcasting mode, the amount of video data transmitted is similarly changed. Therefore, the multiplexer 80 multiplexes the broadcast mode selection information applied from the control unit, the bit allocation information of the bit allocation unit 30, the bit streams of the audio data and the video data, and outputs the multiplexed data having a fixed length. .

6 is a configuration diagram of a frame unit of audio data encoded by the apparatus of the present invention. One frame includes a header, an ALLOC (Allocation) indicating bit information of each frequency band, a scale factor, a sample (Sample), which is audio data of the frequency band, and an ANC, which is additional information. Broadcast mode selection information is assigned to a header of one frame so that the audio data can be properly decoded by the decoding apparatus. In this case, for example, up to four broadcast modes can be determined by using 2-bit information. As shown in FIG. 6, the header portion of one frame is fixed, but the remaining portion varies depending on the broadcast mode, so the length of one frame also changes.

As described above, the present invention saves the number of bits by varying the compression rate of audio data and the number of bits to which data is allocated according to the type of audio broadcasting and the specific gravity of sound quality when encoding an audio signal. Since an extra amount of spare bits is allocated to the video data, the picture quality can be improved when the video signal is reproduced.

Claims (10)

An apparatus for adaptively encoding an audio signal sampled at a predetermined frequency according to a broadcast mode, wherein the sampled audio signal is divided into divided bands and a bit is allocated according to a broadcast mode in order to reduce the number of bits to which audio data is allocated. Audio signal encoding means for quantizing the divided band signals by calculating the bits of each divided band so as to limit the band so that the masking threshold calculated according to the broadcast mode in the divided band does not exceed the quantization noise as much as possible; Video signal encoding means for calculating the fullness of the buffer in consideration of the number of bits remaining by the audio signal encoding means in accordance with the broadcast mode, and adaptively quantizing the video signal according to this value; An audio broadcasting type comprising multiplexing means for multiplexing the broadcasting mode information, the bit allocation information output from the audio signal encoding means, and the audio data and the video data output from the video signal encoding means into multiple data formats. Variable bit allocation device according to the present invention. 2. The apparatus of claim 1, wherein the audio signal encoding means comprises: a division band filter group for dividing the sampled audio signal into a predetermined number of division band signals; A fast Fourier transform and a threshold calculation unit for analyzing the sampled audio signal by performing Fast Fourier transform and calculating a masking threshold which cannot detect noise due to auditory characteristics according to the split band signal outputted from the split band filter group and the broadcast mode; A bit allocation unit for allocating bits for each divided band within a range not exceeding the masking threshold value output from the fast Fourier transform and the threshold calculating unit according to the broadcasting mode; And a first quantizer for quantizing the divided band signals outputted from the divided band filter group according to the bit allocation information outputted from the bit allocation unit. The apparatus of claim 2, wherein a bandwidth to which an audio signal output from the split band filter group is allocated is changed in units of a split band according to an audio broadcasting mode. 3. The method of claim 2, wherein the fast Fourier transform and threshold calculation unit calculates a masking threshold in consideration of only the maskers present in the bit-assigned split bands by limiting the divided band to be allocated by the broadcast mode. Variable bit allocation device according to the type of audio broadcast. The video signal encoding means of claim 1, further comprising: a second quantizer for adaptively quantizing the sampled video data according to the fullness value applied from the fullness calculator; A buffer for storing video data input at a variable rate from the second quantizer and outputting the fixed data at a fixed rate; Variable bit allocation apparatus according to the type of audio broadcast, characterized in that it comprises a fullness calculator for calculating the degree of fullness of the data stored in the buffer according to the output signal and the broadcast mode of the buffer and outputs this value to the second quantization unit. . The method of claim 5, wherein the fullness calculating unit calculates the fullness of the buffer by subtracting the amount of data saved by the bandwidth limit from the amount of data stored in the buffer when the bandwidth is not limited according to the broadcasting mode. Variable bit allocation device according to the type of audio broadcast. The apparatus of claim 1, wherein the audio data output from the multiplexing means is configured by adding broadcast mode selection information to a header portion of each frame. 8. The variable bit allocation according to the audio broadcast type according to claim 7, wherein the broadcast mode selection information is information for selecting one of audio broadcast modes set to an arbitrary number according to the type of audio broadcast and the weight of sound quality. Device. 10. The audio broadcasting method according to claim 8, wherein the multiplexing means adds an audio signal band corresponding to the number of bits saved at the time of bit allocation according to the broadcasting mode selection information to multiplex the data. Variable Bit Allocation. 10. The variable bit allocation according to the audio broadcasting type according to claim 9, wherein the multiplexing means outputs data having a fixed length by multiplexing while varying bit streams of audio data and video data according to broadcast mode selection information. Device.