KR0152016B1 - Encoding and decoding system using variable bit allocation - Google Patents

Abstract

The present invention relates to an encoding and decoding system of an audio signal. The present invention relates to an audio signal encoding and decoding system, and analyzes frequency components of an input signal, performs a spectral analysis of a signal by combining an auditory characteristic and an input signal, and uses spectral analysis results to quantize data. A variable number of bits are variably assigned, the signals of each frequency component are quantized according to the allocated number of bits, and a buffer occupancy rate is generated, which is the extent to which the quantized data is stored, so that the bit allocation is adaptively converted according to the buffer occupancy rate. An encoding system and a decoding system for inversely quantizing a signal of each frequency component encoded by such an encoding system, synthesizing and reproducing a signal as an original signal.

Description

Encoding and decoding system using variable bit allocation.

1 is a block diagram showing an example of a coding apparatus using a conventional bit allocation.

2 is a waveform diagram showing the results of spectral analysis in the apparatus of FIG.

3 is a block diagram showing an example of a decoding apparatus using a conventional bit allocation.

4 is a schematic diagram showing a data state by a conventional bit allocation method.

5 is a block diagram showing an example of an encoding apparatus using variable bit allocation according to the present invention.

6 is a block diagram showing an example of a decoding apparatus using variable bit allocation according to the present invention.

7 is a schematic diagram showing a data state by the bit allocation method of the present invention.

* Explanation of symbols for main parts of the drawings

12: frequency analysis unit 13: auditory characteristics storage unit

15,33: bit allocation unit 14: spectrum analysis unit

16: quantization unit 17: formatting unit

31: de-formatting unit 32: inverse quantization unit

34: sub-reverse synthesis unit 50, 60: buffer

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal encoding and decoding system, and more particularly, to compresses data into a state useful for recording and transmission by adaptively converting the number of data bits of an input signal allocated for encoding, and a signal closer to the original state. The present invention relates to an encoding and decoding system that can be reproduced with a.

In general, human hearing has an audible range of frequency components up to about 20 kHz. Conventional digital audio products such as CD and DAT use audio characteristics to sample and record and reproduce audio signals with 16-bit resolution at a sampling frequency of 40 kHz or higher. In this case, it is possible to reproduce a wide range audio signal with a dynamic range of 96dB, which is an operating range capable of processing the signal without distortion. When recording and reproducing or transmitting and receiving the sampled digital audio signal, it takes 700kbit / sec or more per channel. Depending on the application, the required data amount is excessive, which may be difficult to implement. In order to solve this problem, various methods for performing data compression of four times or more while maintaining the sound quality of a CD have been proposed.

1 shows an example of a conventional audio signal encoding apparatus. The encoder of FIG. 1 stores an A / D converter 11 for converting an analog audio signal into a digital signal, a frequency analyzer 12 for analyzing frequency components of the digital audio signal, and stores human auditory characteristic data. Each of the frequency components divided by the auditory characteristic storage unit 13, the spectrum analyzer 14 which calculates a threshold level at which the noise due to the auditory characteristics cannot be detected by spectrum analysis of the digital audio signal, and the frequency analyzer 12 The bit analyzer 15 allocates sufficient data bits within the limit not exceeding the threshold level calculated by the spectrum analyzer 14, and the frequency analyzer uses the bits allocated by the bit assigner 15. A quantization unit 16 for quantizing the signal of each frequency component divided in (12), and a formatting unit 17 for formatting and transmitting the quantized data into a data packet of an appropriate format. Is done.

In FIG. 1, an audio signal having a frequency bandwidth of about 20 kHz is converted into a digital signal through the A / D converter 11, and this digital signal is generally about 40 kHz or more with a sampling frequency (f _s ) and about 16 bits. Has the resolution of. The digital audio signal converted as described above is analyzed by the frequency analyzer 12 for the frequency component of the signal, and thus is divided into frequency components of subbands. The spectrum analyzer 14 performs spectral analysis on the audio signal output from the A / D converter 11. As shown in FIG. 2, the spectrum analyzer 14 receives spectral analysis (S _SPT ) of the digital audio signal and data about the human auditory characteristics from the auditory characteristic storage unit 13, thereby reducing noise in the auditory characteristics. Obtain the Masking Threshold (S _MT ), which is the maximum level that cannot be detected even if present. Then, the torsion allocator 15 refers to the analysis result of the spectrum analyzer 14 to determine each frequency as long as the noise S _QN existing when quantized does not exceed the _masking threshold value S _MT . Allocate enough bits to the band. The quantization unit 16 quantizes the audio signal of each frequency band by using the bits allocated to each frequency band as described above. The quantized audio signal is supplied to the formatting unit 17 and configured in an appropriate format including additional information. This additional information includes bit allocation information for the number of bits allocated for a predetermined amount of unit data required for quantization in the quantization unit 16. The encoded audio signal S _CD output from the formatting unit 17 is recorded or transmitted. As described above, the spectral analysis of the digital audio signal and the bit allocation for each frequency band are performed at constant time intervals in the encoding apparatus. Hereinafter, each data bundle of audio signals divided by such a predetermined time interval is called a frame.

3 shows an example of a conventional audio signal decoding apparatus. The apparatus of FIG. 3 is a deformatter 31 which decomposes the encoded data S _CD transmitted from the encoder of FIG. 1 and separates pure audio data components and additional information, and data of each frequency band obtained by deformatting. An inverse quantization unit 32 for inverse quantization, a bit allocation unit 33 for providing bit allocation information necessary for inverse quantization from the additional information, and a unit for generating a digital synthesized sound by receiving an audio signal of each dequantized frequency band An inverse synthesizer 34 and a D / A converter 35 for converting the digital synthesized sound into an analog signal and outputting the final audio signal S _A.

In such a conventional encoding and decoding system, when it is determined how much to compress transmission data, the number of bits allocated to one frame is determined, and the limited bits are appropriately allocated to each frequency band.

4 shows a mask to noise ratio (MNR) of each frequency band according to a conventional bit allocation method. As shown in (a) of FIG. 4, bit allocation sequentially allocates bits in order from the lowest portion of the MNR so that the overall MNR rises almost evenly. For reference, when the data bit is increased by 1 bit when representing the sample data in digital signal processing, the signal to noise ratio (SNR) is known to increase by about 6 dB.

Thus, the final MNR generated by bit allocation is as shown in FIG. 4 (b) or (c). (B) of FIG. 4 is a case where a bit is sufficiently allocated relative to the MNR, and the sound close to the original sound can be reproduced by raising the MNR to 0 dB or more over the entire frequency band.

On the other hand, (c) of FIG. 4 is a case in which bits are insufficiently allocated relative to MNR, and in some frequency bands, the MNR is smaller than 0 dB even when all the allocated bits are exhausted. As such, when the MNR is less than 0 dB, an audible sound is heard when listening.

In the conventional encoding and decoding apparatus, in allocating the number of bits necessary to quantize the divided audio signals of each frequency band, the MNR is less than 0 dB even if all the allocated bits are used by allocating a constant number of bits for each frame. There will be a band. This frequency band causes annoying sound when listening.

Accordingly, the present invention provides a new bit allocation scheme to solve such a conventional problem, and also encodes using variable bit allocation that can be reproduced as a signal closer to the original signal by variably allocating the number of bits allocated to each frame. And a decoding system.

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

5 shows an embodiment of an apparatus for encoding an audio signal according to the present invention. The encoder of FIG. 5 further includes a buffer 50 in the encoder of FIG. In FIG. 5, the same components as those of the apparatus of FIG. 1 are denoted by the same reference numerals, and the description of the operation of the same components will be omitted. In FIG. 5, the buffer 50 is a storage means for storing data input at a variable rate from the formatting unit 17 and outputting at a fixed rate. Since the audio signal to be encoded may have insufficient or remaining data in each frame, it is necessary to save the bits in the remaining frames and allocate them to the frames having insufficient bits, where the buffer 50 needs information for such variable bit allocation. Serves to provide. To this end, the buffer 50 supplies the buffer occupancy, which is the amount of data stored in the buffer, to the bit allocator 15 so that the bit allocation is variably made according to the buffer occupancy. The bit allocation unit 15 performs bit allocation in three modes according to the occupancy of the buffer 50. In the first mode, when the buffer occupancy is high (more than the second predetermined reference value), that is, when the buffer 50 is almost filled with the storage data, the bit allocation unit 15 performs the quantization unit 16 according to the first fixed ratio. Provides bit allocation information in. In this case, the first fixed ratio allows the maximum number of bits allocated to one frame to be the average number of bits per frame so that no overflow occurs. When a bit is allocated by the first fixed ratio in which the maximum number of bits is limited in this manner, a frame requiring only a little bit may exist, and thus the occupancy of the buffer 50 gradually decreases. If the buffer occupancy is between the predetermined reference values of the first and second, the bit allocator 15 enters the second mode state. The second mode performs bit allocation with a variable allocation ratio for each frame. In other words, the number of bits required in each frame is allocated so that the MNR of each frequency band is 0 dB or more over the entire frequency band. Data quantized by such a variable allocation ratio has a variable data amount for each frame. In addition, when the buffer 50 is almost empty and the occupancy rate is lowered below the first predetermined reference value, the bit allocation unit 15 is converted to the third mode so that the bit allocation is performed by the second fixed ratio to the quantization unit 16. Provides bit allocation information. The second fixed ratio allows the minimum number of bits allocated to one frame to be the average number of bits per frame so that underflow does not occur. Allocate these bits so that the MNR of each frequency band rises even if bits remain.

6 shows an embodiment of a decoding apparatus according to the present invention. The decoding apparatus of FIG. 6 further includes a buffer 60 in the decoding apparatus of FIG. Here, the buffer 60 stores the encoded audio signal transmitted from the encoding apparatus of FIG. Deformatting unit 31 decodes the audio signal stored in the buffer 60, and decodes the encoded signal as the bit allocation information using the number of bits allocated for the pure audio data component and a predetermined amount of unit data necessary for quantization during encoding. Separate into additional information that contains. The audio data of each frequency band output from the deformatting unit 31 is dequantized by the dequantization unit 32 and the additional information is supplied to the bit allocation unit 33. At this time, the dequantization unit 32 allocates a predetermined number of bits to each frequency band to be dequantized according to the bit allocation information supplied from the bit allocation unit 33. The digital audio signal of each dequantized frequency band as described above is synthesized by the sub-reverse combining unit 34. The synthesized digital audio signal is converted into an analog audio signal S _A by the D / A converter 35 and output.

7 shows MNR of each frequency band according to the bit allocation scheme of the present invention. As shown in (a) of FIG. 7, in the present invention, bits are allocated from low frequency components until MNR is sequentially raised to 0 dB or more. That is, the basic principle is to allocate bits on the low frequency side as much as possible when the bits are insufficient. In FIG. 7B, when the bits are relatively sufficient, the MNR of the entire frequency band becomes 0 dB or more, and the original sound is faithfully reproduced. On the other hand, (c) of FIG. 7 shows a case in which bits are relatively inadequate, and bits are allocated to low frequencies to maintain an MNR of 0 dB or more , but some frequency bands of the highest frequencies are not allocated bits. Signals in the unassigned frequency band X are not transmitted. This in turn has the same effect as low pass filtering. As such, although some of the highest frequency components may be omitted in the bit allocation method of the present invention, it does not feel rejection due to human hearing characteristics, and MNR is more than 0 dB while allocating bits for the entire frequency band in the conventional bit allocation method. It offers much better sound quality than when low frequencies exist.

As described above, the encoding and decoding system according to the present invention, in quantizing the divided audio signal of each frequency band, saves the bits in the frame in which the allocated bits remain and allocates them to the frames lacking bits, and bit allocation is performed at low frequency. By sufficiently assigning components, the MNR of each frequency band is set to 0 dB or more, so that an audio signal can be encoded with a limited data amount and reproduced with a sound closer to the original sound during decoding.

In the above, only the encoding and decoding of the audio signal is described as an embodiment of the present invention. However, the encoding and decoding system using the variable bit allocation according to the present invention can be applied not only to the audio signal but also to the video signal. When used for processing video signals, it is possible to perform effective data compression while maintaining a smoother screen.

Claims (18)

Analyzing a frequency component of the input signal, spectroscopically analyzing the input signal to calculate a predetermined threshold level at which the noise cannot be detected due to auditory characteristics, and using the threshold level to determine signals of the divided frequency bands. A coding method comprising the step of quantizing, comprising the steps of storing the data quantized and formatted in an appropriate format, generating occupancy which is the extent to which the data is stored in the data storage step, and the threshold level obtained from the spectral analysis. And adaptively allocating the number of bits necessary for quantization according to the occupancy rate, and quantizing the signals of the divided frequency bands according to the allocated number of bits. The method of claim 1, wherein the bit allocation step comprises allocating bits at a predetermined fixed ratio when the data occupancy is less than a predetermined reference value or more than another predetermined reference value, and when the data occupancy is within the two predetermined reference values. And allocating bits at a rate. The encoding method according to claim 2, wherein the fixed ratio bit allocation step is characterized in that, when the data occupancy is equal to or less than a predetermined reference value, the number of bits allocated to a predetermined amount of unit data to be quantized is equal to or greater than the average number of bits. 3. The encoding method according to claim 2, wherein the fixed ratio bit allocation step is characterized in that the number of bits allocated to the unit data of a predetermined amount to be quantized is equal to or less than the average number of bits when the data occupancy is equal to or greater than another predetermined reference value. The encoding method of claim 2, wherein the variable bit allocation step saves bits in unit data in which bits remain for a predetermined amount of unit data to be quantized, and allocates the saved bits to unit data that lacks bits. Way. The method of claim 1 or 2, wherein the bit allocation step includes allocating bits sequentially from a signal of a low frequency band to a signal to be quantized, and so that a mask to noise ratio (MNR) of a signal of each frequency band is 0 dB or more. And encoding the bits. A method for decoding an encoded signal, comprising the steps of: inputting a signal encoded by the method of claim 1, deformatting the encoded signal to separate the signal and additional information of pure data components; Inversely quantizing a signal of a predetermined data component, receiving the separated additional information, providing bit allocation information necessary for the inverse quantization, and synthesizing a signal of each dequantized frequency component Decoding method, characterized in that. 8. The decoding method according to claim 7, wherein the bit allocation information is information included in additional information in the encoded signal as the number of bits allocated for a predetermined amount of unit data in the quantization step during encoding. A coding apparatus comprising means for analyzing a frequency component of an input signal, and means for spectral analyzing the input signal to calculate a predetermined threshold level at which the noise cannot be detected due to auditory characteristics. A buffer for generating occupancy indicative of a stored degree, and means for adaptively allocating the number of bits required for quantization according to the threshold level calculated by the spectrum analysis means and the occupancy of the buffer; And quantization means for quantizing a signal of each frequency component output from the frequency analysis means and supplying the signal to the buffer according to the bit allocation information supplied from the bit assignment means. 10. The apparatus of claim 9, wherein the buffer stores data input at a variable rate and outputs the stored data at a fixed rate. 10. The encoding apparatus according to claim 9, wherein the bit allocation means includes means for comparing the buffer occupancy rate with a predetermined reference value, and means for allocating bits at a variable rate according to the comparison result. 12. The encoding method according to claim 11, wherein the comparing means sets first and second reference values having a predetermined size for buffer occupancy, and compares the buffer occupancy supplied from the buffer with the first and second reference values. Device. The encoding device according to claim 9 or 11, wherein the bit allocation means has a number of bits allocated for a predetermined amount of unit data to be quantized when the buffer occupancy is equal to or less than the first reference value. 12. The encoding apparatus according to claim 9 or 11, wherein the bit allocation means has a number of bits allocated for a predetermined amount of unit data to be quantized when the buffer occupancy is equal to or greater than the second reference value or less. 12. The apparatus of claim 9 or 11, wherein the bit allocation means saves bits in unit data in which bits remain for a predetermined amount of unit data to be quantized when the buffer occupancy is between the first reference value and the second reference value. And assigning the saved bits to unit data that lacks bits. 10. The encoding apparatus according to claim 9, wherein the bit allocation means allocates bits in order from a signal of a low frequency band to a signal output from the frequency analysis means. The encoding device according to claim 9 or 16, wherein the bit allocation means allocates bits so that MNR of a signal of each frequency component output from the frequency analysis means is equal to or greater than 0 dB. An apparatus for receiving and decoding a signal encoded by an encoding apparatus, the apparatus comprising: an input terminal to which a signal encoded by the encoding apparatus of claim 9 is input, and a signal of pure data components by deforming the encoded signal applied to the input terminal; Means for separating the number of bits allocated for a predetermined amount of unit data necessary for quantization during encoding into additional information including as bit allocation information; Means for inversely quantizing a signal of the separated data component; Means for receiving the separated additional information and providing bit allocation information in the additional information to the dequantization means; And means for synthesizing a signal of each dequantized frequency component.