KR0144841B1 - Apparatus for adaptive encoding and decoding of sound signals - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for encoding or decoding an audio signal for transmission or storage, and more particularly, to decode bit allocation according to the importance of information included in each subband of an audio signal, that is, the energy distribution of the audio signal. Priority is set and bit allocation is most closely performed on the human auditory characteristics by the psychoacoustic model, thereby adaptively allocating a large number of bits to subbands containing information important for encoding and decoding the acoustic signal. The present invention relates to an adaptive encoding and decoding apparatus for an acoustic signal capable of minimizing a quantization error caused by encoding with a fixed number of bits and reproducing an acoustic signal close to a human auditory characteristic.

Description

Apparatus for adaptive encoding and decoding of sound signals

1 (a) is a block diagram showing an example of a conventional sound signal encoding apparatus,

1 (b) is a block diagram showing an example of a conventional sound signal decoding apparatus,

2 (a) is a block diagram showing an embodiment of an adaptive encoding apparatus for an acoustic signal according to the present invention;

2 (b) is a block diagram showing an embodiment of an adaptive decoding apparatus for an acoustic signal according to the present invention;

* Explanation of symbols for main parts of the drawings

21: split band decomposition unit 22, 26: energy analysis unit

23: psychoacoustic model unit 24: quantization unit

25: dequantization unit 27: split band synthesis unit

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for encoding or decoding an audio signal for transmission or storage, and to decode the audio signal. The present invention relates to an adaptive encoding and decoding apparatus for an acoustic signal for minimizing quantization error.

In recent years, as the development of high-definition television (HDTV) and digital video tape recorders has been actively developed, video and audio signals are digitally encoded and stored or transmitted in a system for transmitting and receiving video and audio signals. An encoding / decoding system that decodes and reproduces the original signal is mainly used. For the industrial use of this digital encoding / decoding system, the International Committee for International Telegraphy and Telephony (CCITT), the International Radiocommunication Consultative Committee (CCIR), and the International Organization for Standardization for Standardization (ISO), etc., is conducting standardization activities to determine the appropriate standard coding scheme for each application. Among them, a typical method widely used for encoding and decoding an acoustic signal is shown in FIG.

FIG. 1 (a) is a block diagram illustrating an example of a conventional audio signal encoding apparatus. In the apparatus of FIG. 1 (a), an acoustic signal is applied to the divided band resolution 11 and the psychoacoustic model unit 12, respectively, and the output end of the divided band resolver 11 is the quantization 13 and the psychoacoustic model unit. It is connected to the input terminal of (12). The input terminal of the quantization unit 13 is connected to the output terminal of the psychoacoustic model unit 12 and configured to output quantized sound data.

In the apparatus of FIG. 1 configured as described above, first, the split band decomposition unit 11 divides an input signal into a predetermined amount of signal and performs subband analysis to compress an applied sound signal. In general, a weighted overlap-add filter bank or wavelet transform is mainly used as a susband analysis method. By this subband analysis, the acoustic signal is divided into a plurality of subbands in the frequency domain. Meanwhile, the psychoacoustic model unit 12 receives an acoustic signal and allocates the number of bits to each of the divided subbands to the quantization unit 13 by the psychoacoustic model that models a person's auditory characteristics. . At this time, the psychoacoustic model unit 12 allocates fewer bits to subband groups corresponding to high frequencies among the frequencies (100 to 10,000 kHz) that humans can detect mainly for data compression within a limited number of bits. . This is because human hearing is more sensitive to low frequencies than to high frequencies. The quantization unit 13 quantizes the subbands applied from the split band decomposition unit 11 according to the number of bits designated by the psychoacoustic model unit 12 and outputs an encoded sound signal.

The data encoded and compressed by the acoustic signal encoding apparatus is transmitted to the acoustic signal decoding apparatus as shown in FIG. 1 (b). The transmitted data is then reproduced as data close to the state before the subbands were synthesized and encoded by the subband synthesizer 15 via the inverse quantizer 14.

However, the conventional audio signal encoding / decoding apparatus encodes with a fixed number of bits even when a large number of important signals are included on the high frequency side, so that a large quantization error occurs, resulting in negative distortion after synthesis in the decoding apparatus. There was this. In addition, even if the important signal is not included in the low frequency side, there is a problem in that the transmission data efficiency is lowered because high bit allocation is performed.

Accordingly, an object of the present invention is to provide an apparatus for adaptively encoding an acoustic signal for minimizing a quantization error by adaptively allocating and encoding a large number of bits in a subband including important information in encoding an acoustic signal.

Another object of the present invention is to provide an adaptive decoding apparatus of an acoustic signal for decoding the acoustic data encoded by the aforementioned adaptive encoding apparatus of the acoustic signal.

According to an aspect of the present invention, there is provided an apparatus for encoding an acoustic signal. The apparatus for encoding an acoustic signal includes a split band decomposition unit for dividing an input sound signal into predetermined frequency bands. It includes an energy analysis unit for calculating the energy for each sound signal divided by the unit. In addition, the psychoacoustic model unit for receiving the acoustic signal and modeling according to the human auditory characteristics, and assigning the coding bits according to the energy distribution of each divided acoustic signal calculated from the energy analysis unit to output the final coded bit allocation information, and And a quantization unit for allocating and quantizing the sound signal applied through the energy analysis unit according to the encoding bit allocation information applied from the psychoacoustic model unit.

An apparatus for adaptively decoding an acoustic signal according to the present invention for achieving the above-described object includes a dequantization unit for inversely quantizing the acoustic data, in the apparatus for decoding an acoustic signal for reproducing encoded sound data. The apparatus may include an energy analyzer configured to calculate energy for each subband of the acoustic data, and a split band synthesizer configured to band synthesize the dequantized acoustic signal according to the energy distribution calculated by the energy analyzer.

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

2 (a) is a block diagram showing an embodiment of an apparatus for adaptive encoding an acoustic signal according to the present invention. In the apparatus of FIG. 2 (a), an acoustic signal is applied to the split band resolver 21 and the psychoacoustic model 23, and the output end of the split band resolver 21 is connected to the input of the energy analyzer 22. do. The output of the energy analysis unit 22 is applied to the psychoacoustic model unit 23 and the quantization unit 24, and the output of the psychoacoustic model unit 23 is applied to the quantization unit 24, so that the quantization unit 24 The encoded sound data is configured to be output.

In the apparatus of FIG. 2 (a) configured as described above, first, the division band decomposition unit 21 divides the applied acoustic signal into a predetermined amount of signal, and performs subband analysis on the divided signals. The description of the subband analysis method is the same as that described in the apparatus of FIG. For the subbands divided by bands, the energy analyzer 22 calculates the energy of each subband as shown in Equation (1).

In Equation (1), X (k) represents energy at the kth sampling point in one subband, N represents the number of sampling points in one subband, and x (n) represents a given acoustic signal in the time domain. Indicates. First, the acoustic signal x (n) in the time domain is Fourier transformed into a frequency domain, and the energy of the frequency domain is multiplied to obtain a square power. The logarithm is taken in the above equation to convert to decibel (dB).

In this way, equation (1) is applied to all sampling points in one subband to obtain the energy value of each sampling point, and the energy values of these sampling points are summed as shown in the following equation (2), and the total energy S _ij [X (k)] is obtained.

In Eq. (2), i to j represent a range of subbands for obtaining energy.

The energy analysis unit 22 prioritizes the subbands in order of the total energy obtained as described above, and outputs the priority to the sub-band. In this case, the psychoacoustic model unit 23 receives a sound signal and has a psychoacoustic model that models the hearing characteristics of a person. Here, it is known that the human auditory characteristics have an effect of masking adjacent frequencies of a small level with frequency components of a large level. The concept of a critical band is used as a measure to specifically illustrate this. In each hearing, when a different component near a frequency component is moved away from the frequency axis, the masking effect is weakened and sounds loud, despite the same level. The band at this time is defined as a critical band. It is decomposed into 20 bands in a band of 20 Hz. By this property, the perceptible noise level (noise masking threshold) is obtained for each band, and modeled in combination with the hearing limit (absolute threshold). The psychoacoustic model unit 23 allocates bits to each subband by synthesizing the energy analysis results for the respective subbands applied from the psychoacoustic model and the energy analyzer 22. The quantization unit 24 quantizes the sound signal applied from the energy analyzer 22 according to the bit allocation information applied from the psychoacoustic model unit 23.

At this time, one of the following two methods is used to allocate bits to each subband. The first is a method of allocating bits in the energy analyzer 22 through energy analysis for all subbands. This method applies the same method to the low frequency region which is very sensitive to the human ear and allocates a small bit if there is no important signal in the region. Second, a predetermined bit set by the psychoacoustic model unit 23 is allocated to the low frequency region sensitive to the human ear, and the bit allocation given as a result of the energy analysis of the energy analyzer 22 is allocated to the remaining subbands. to be.

By applying one of the above methods to the acoustic signal, the data encoded by adaptively bit allocation according to the energy distribution is transmitted to the apparatus as shown in FIG.

FIG. 2 (b) is a block diagram showing an embodiment of an adaptive decoding apparatus according to the present invention. The inverse quantization unit 25 dequantizes sound data encoded by the apparatus as shown in FIG. 2 (a). Applied to the split band synthesis section 27, and outputs the information analyzed the energy of the acoustic data in the energy analysis section 26 to the split band synthesis section 27, thereby splitting into subbands in the split band synthesis section 27 The synthesized sound signal is synthesized into a continuous sound signal by the energy analysis result, and the decoded sound signal is output.

As described above, the adaptive encoding and decoding apparatus for the acoustic signal according to the present invention sets the priority of bit allocation according to the energy distribution of the input sound signal, and most closely approximates the human auditory characteristics by the psychoacoustic model. By quantizing by bit allocation, a large number of bits are adaptively allocated to subbands containing information important for encoding and decoding an acoustic signal, thereby minimizing quantization error due to encoding a fixed number of bits. It has the effect of reproducing the sound signal close to the auditory characteristics.

Claims (10)

An apparatus for encoding an acoustic signal, comprising: split band decomposition means for dividing an input acoustic signal into predetermined frequency bands; Energy analysis means for calculating energy for each of the acoustic signals divided by the split band decomposition means; Psychoacoustic model means for receiving the acoustic signal and modeling it according to the human auditory characteristics, and allocating coded bits by receiving energy information of each divided acoustic signal calculated from the energy analyzing means; And quantization means for allocating and quantizing the sound signal applied through the energy analysis means according to the coded bit allocation information applied from the psychoacoustic model means. 2. The apparatus of claim 1, wherein the energy analyzing means includes means for calculating energy at each sampling point and means for calculating total energy in each subband. The apparatus of claim 2, wherein the energy of the sampling point is calculated by the following equation. Here, X (k) is energy at the kth sampling point in one subband, N is the number of sampling points in one subband, and x (n) is a sound signal given in the time domain. The apparatus of claim 2 or 3, wherein the total energy in the subbands is calculated by the following equation. Here, S _ij [X (k)] is total energy in one subband, and i to j are ranges of subbands for obtaining energy. The apparatus of claim 1, wherein the psychoacoustic model means allocates coded bits according to an energy distribution of an acoustic signal applied from the energy analyzing means. The method of claim 1, wherein the psychoacoustic model means assigns a coded bit to a low frequency band acoustic signal sensitive to a human ear according to information modeling human auditory characteristics, and the energy analysis means for the remaining band acoustic signal. Adaptive encoding apparatus for an acoustic signal, characterized in that for allocating the coded bits in accordance with the energy distribution of the sound signal applied from. An apparatus for decoding sound signals for reproducing encoded sound data, comprising: inverse quantization means for inversely quantizing the sound data; Energy analysis means for calculating energy for each subband of the sound data; And split-band synthesizing means for band-synthesizing the dequantized sound signal according to the energy distribution calculated by the energy analysis means. 8. The apparatus of claim 7, wherein the energy analyzing means comprises means for calculating energy at each sampling point and means for calculating total energy in each subband. 10. The apparatus of claim 8, wherein the energy of the sampling point is calculated by the following equation. Here, X (k) is energy at the kth sampling point in one subband, N is the number of sampling points in one subband, and x (n) is a sound signal given in the time domain. 10. The apparatus of claim 8 or 9, wherein the total energy in the subbands is calculated by the following equation. Here, S _ij [X (k)] is total energy in one subband, and i to j are ranges of subbands for implementing energy.