# Method and apparatus for encoding and decoding audio signals (versions)

FIELD: information technology.

SUBSTANCE: input signal is converted to spectral coefficients; the spectral coefficients are grouped into frequency bands and standards are estimated for each band as the average energy in the band; the spectrum is normalised based on the estimated standards; the standards are weighted based on psycho-acoustic properties of sound; bit distribution is calculated based on the weighted standards; the spectrum is quantised and encoded by the obtained number of bits; the method is characterised by that bit distribution is calculated based on a psycho-acoustic model built on quantised standards. Also disclosed is a device for implementing this method.

EFFECT: low level of distortions and easier encoding.

26 cl, 15 dwg

The invention relates to a method for compression of digital signals such as audio signals; and more particularly to a bit allocation algorithm, the substitution of noise and effective adaptive compression coefficient quantization.

Currently, the most AudioCodes based on the encoding in the frequency domain. These encoders consist of a transform module signal from the temporary shape in the frequency, quantizer and module lossless compression. The quantization error is controlled using information from the psychoacoustic model, and the quantized spectral coefficients, together with some additional information are encoded losslessly.

The G.719 audio encoder is a standard by ITU-T [1] by the encoder based on frequency conversion with adaptive bit allocation and vector quantization. The input time signal is converted into spectral coefficients using a transform MDCP (Modified Discrete Cosine Transform). The spectral coefficients are grouped into frequency bands and the energy bands (standards) is estimated for each band. These rules are used to normalize the spectrum, and the bit allocation algorithm. Normalized spectral coefficients vector quanthouse and the encoded number of bits that has been defined for each band. Before the algorithm races is determining the bit norms additionally weighted with the help of some psychoacoustic criteria, such as the masking effect. The pattern of bits uses the quantized weighted norms for the distribution of the total available number of bits between the bands within a single frame. The algorithm allocates one bit to the frequency for each band up until all selected bits will not be spent. At each iteration, selects the largest norm, bits allocated to the respective bands, and the norm is reduced by 6 dB. The bit allocation algorithm also uses the weighting of the spectrum on the basis of psychoacoustic criteria, however, the weighting factors largely limited, therefore, improves the quality slightly. In addition, the accuracy of the distribution of bits from an integer calculations may be insufficient for effective compression ratio at low speed.

In the decoder for G.719 spectral coefficients that were not transmitted due to insufficient number of available bits, the algorithm is used to fill the spectrum. In the bands, where he was transferred to the at least one coefficient, the filling is not performed. At low speed, where the number of bands with multiple encoded frequencies is large enough, there are many gaps in the spectrum, resulting in audible distortion.

The standard speech coding ITU-T G.718 [2] based on a factorial pulse coding (FPC) to encode the coefficients M is CP. This encoding method, as is known, effectively encodes the amplitude of the individual pulses and uses the calculation of combinatorial functions. These calculations are computationally complex, requiring many operations of multiplication and division, especially if the length of the signal is high, or the signal has a large amplitude. Given allocated to strip bits, FPC estimates the number of pulses in these bands. As a rule, to determine the relationship between beats and pulses uses a table and a well-known binary search algorithm. This algorithm is simple enough, however, the operation to calculate the logarithm required for each comparison, is quite complicated. These approaches to varying degrees implemented in known from the prior art solutions [3], [4] and [5].

The challenge which seeks the invention is to overcome the above disadvantages, namely to reduce the level of distortion and the high complexity of the encoding FPC.

The technical result is achieved due to the development of an improved method for more efficient allocation of bits among the frequency bands for encoding audio signals on the basis of transformations, as well as through the development of an improved device for encoding/decoding an audio signal. In the framework of the inventive CSP is both described the two-stage algorithm to determine the number of pulses with low complexity - prediction of the number of pulses in the band, and then a binary search within a small sub-band using the table of constants. In addition, the claimed solution contains a padding scheme spectrum for the substitution of noise instead of zero coefficients, and effective FPC coding with adaptive transmission ratio norms. Additionally, reduced complexity estimates of the number of pulses.

Specifically, the main variant of the inventive method provides for the encoding of a temporary audio signal, which is that the input signal is converted into spectral coefficients are grouped spectral coefficients in the frequency band, and evaluate standards for each band as the average energy in the band, normalize spectrum on the basis of the evaluated standards, weighed norms based on psychoacoustic properties of sound, calculated bit allocation based on weighted norms, quantum and encode the spectrum of the received bits, and the bit distribution calculated on the basis of psychoacoustical models based on the quantized codes.

When encoding at low speed bits must be allocated efficiently from the point of view of the quality of perception of sound. In the invention there are two methods of allocating bits based on the psychoacoustic model: forehand is her additional information, and without it. Both methods provide the possibility of implementing low complexity.

The quantized spectrum MDCP can contain many zero coefficients when the coding rate is low, resulting in audible distortion in the form of a metallic overtones. Filling noise is a good way to disguise gaps in the spectrum, however, this noise can ruin tones - they become noisy and dull. Proposed in the present invention, the lookup algorithm provides adaptive noise filling the spectrum taking into account the tone encoded spectral coefficients.

As to reduce the complexity of the encoding FPC, in the framework of the invention it is also proposed two-stage algorithm to determine the number of pulses with low complexity - prediction of the number of pulses in the band, and then a binary search within a small sub-band using the table of constants.

According to one of the proposed variants of the invention, determining the bit allocation is performed based on the criterion of relationship energy signal-to-masking threshold.

According to one of the proposed variants of the invention the calculation of the number of pulses is performed based on the criterion of relationship energy signal-to-masking threshold.

According to one variant of the invention the number of bits op is edalat by the formula factorial pulse coding (FPC) from a known number of pulses.

According to one variant of the invention in the proposed encoding method calculates the fill parameters noise for quantized to zero spectral coefficients, with the purpose of masking of failures of the spectrum, then the arguments are passed in the data stream.

According to one variant of the invention in the proposed encoding method, the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity.

According to one variant of the invention in the present application proposes a method of decoding an encoded audio signal, comprising: decoding and recovery standards, the calculation of the distribution of bits based on the reconstructed standards, spectrum decoding and inverse transform of the spectral coefficients in the time domain signal, wherein the bit distribution estimate based on psychoacoustic models based on recovered rules.

According to one variant of the invention in the proposed method of decoding an encoded audio signal, the noise parameters decode the data stream, and quantized to zero spectral coefficients filled with noise with the purpose of masking of failures of the spectrum.

According to one variant of the invention in the proposed method, zakodirovana the encoded audio signal, the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity.

According to another variant of the invention in the present application proposes a method of coding of a temporary audio signal, which consists in the fact that the input signal is converted into spectral coefficients are grouped spectral coefficients in the frequency band, and evaluate standards for each band as the average energy in the band, normalize spectrum on the basis of the evaluated standards, weighed norms based on psychoacoustic properties of sound, calculated bit allocation based on weighted norms, quantum and encode the spectrum obtained by the number of bits, wherein the bit allocation is calculated on the basis of psychoacoustic models based on spectral coefficients.

Under the proposed option in the encoding method of the temporary audio signal of the psychoacoustic properties of the signal estimate based on the coefficients of the modified discrete cosine transform (MDCP).

Under the proposed option in the encoding method of the temporary audio signal of the bit distribution quantuum and passed as additional information.

Under the proposed option in the encoding method of the temporary audio signal of the determination of the allocation of bits based on the criteria of the signal energy to the masking threshold.

According to the about the proposed variant in the coding method of the temporary audio signal of the calculation of the number of pulses based on the criteria of the signal energy to the masking threshold.

Under the proposed option in the encoding method of the temporary audio signal number of bits is determined by the formula FPC from a known number of pulses.

Under the proposed option in the encoding method of the temporary audio signal to calculate population parameters noise for quantized to zero spectral coefficients, with the purpose of masking of failures of the spectrum, then the arguments are passed in the data stream.

Under the proposed option in the encoding method of the temporary audio signal the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity.

According to another variant of the invention in the present application proposes a method of decoding an encoded audio signal, comprising: decoding and recovery standards; calculation of bit allocation based on the reconstructed codes; decoding the spectrum and the inverse transform of the spectral coefficients in the time domain signal, characterized in that the distribution of bit decode the data stream.

Under the proposed alternative in the method of decoding an encoded audio signal, the noise parameters decode the data stream and quantized to zero spectral coefficients filled with noise, with the aim of masking failures spectrum

Under the proposed alternative in the method of decoding an encoded audio signal, the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity.

In this application also proposes a device for encoding/decoding an audio signal that contains the encoder and its associated decoder

this encoder includes the following blocks:

block MDCP conversion to convert the input signal into spectral coefficients;

- the evaluation unit and the quantization rules made with the possibility of grouping the spectral coefficients in the frequency band and evaluation standards for each band, as the average energy in the band;

block coding standards;

- the building block psychoacoustic model for quantized standards, designed to determine the importance of the strips;

- the first block of the bit allocation calculation performed by the calculation of the distribution of bits based on the data about the importance of psychoacoustic models based on the quantized standards;

- block quantization and coding of the spectrum made with the possibility of coding the spectrum obtained by the number of bits.

a multiplexer for transmission of the coded data in the bitstream;

and the decoder includes the following after avatele linked blocks:

- demultiplexer designed to break and decrypt the data stream;

block decoding standards;

unit dekvantovanie standards;

- the building block psychoacoustic model restored regulations;

- the second block of the bit allocation calculation performed by the calculation of the distribution of bits based on the data psychoacoustic models based on recovered regulations;

block decoding and dekvantovanie spectrum with spectrum decoding based on the distribution of bits.

block scaling the decoded spectral coefficients in accordance with the recovered regulations;

- block inverse transform of the spectral coefficients in the signal in the time domain.

According to one variant of the claimed invention, the encoder of the proposed device for encoding/decoding an audio signal further comprises a computing unit noise parameters for quantized to zero spectral coefficients, transmitting the calculated parameters in the data stream.

According to one variant of the claimed invention, the decoder of the proposed device for encoding/decoding an audio signal further comprises the substitute block noise made with the possibility of the recovery of substitution of noise, decode is skilled at zero spectral coefficients.

According to another variant of the claimed invention in the present application is also a device for encoding/decoding an audio signal that contains the encoder and its associated decoder, the encoder includes the following blocks:

block MDCP conversion to convert the input signal into spectral coefficients;

- the evaluation unit and the quantization rules made with the possibility of grouping the spectral coefficients in the frequency band and evaluation standards for each band as the average energy in the band;

block coding standards;

- the building block psychoacoustic model spectral coefficients, designed to determine the importance of spectral coefficients;

- block bit allocation calculation performed by the calculation of the distribution of bits based on the data about the importance of psychoacoustic model;

- block quantization and coding of the spectrum made with the possibility of coding the spectrum obtained by the number of bits.

block coding bit allocation;

a multiplexer for transmission of the coded data in the bitstream;

and the decoder includes the following blocks:

- demultiplexer designed to break and decrypt the data stream;

block decoding standards;

unit dekvantovanie standards;

- BC is to decode the bit allocation, on which input data is received from the stream;

block decoding and dekvantovanie spectrum, the input of which receives data on the distribution of bit data stream;

- the unit normalization of the decoded spectral coefficients in accordance with the recovered regulations;

- block inverse transform of the spectral coefficients in the signal in the time domain.

According to the variant of the claimed invention, the encoder of the proposed device for encoding/decoding an audio signal further comprises a computing unit noise parameters for quantized to zero spectral coefficients, transmitting the calculated parameters in the data stream.

According to the variant of the claimed invention, the decoder of the proposed device for encoding/decoding an audio signal further comprises the substitute block noise made with the possibility of the recovery of substitution decoded noise at zero spectral coefficients.

For a better understanding of the claimed invention, the following is a detailed description with the appropriate drawings.

Figure 1 shows a diagram of an encoder in accordance with the invention, where the spectral coefficients are calculated using MDK conversion, and the average energy is calculated for the spectrum bands. COI is lsua average energy in the band or energy spectral coefficients, psychoacoustic model (US) computes the parameters of importance. The output of the module WE used for the adaptive bit allocation. Normalized spectral coefficients quanthouse and encoded a certain number of bits in each band.

Figure 2 shows a diagram of a decoder in accordance with the invention, where the average energy bands and data FPC quantization decoded from the stream. Next, it calculates the bit allocation using the PAM module, based on energy bands. Using this information are decoded spectral coefficients.

Figure 3 shows another example of the arrangement of the encoder in accordance with the invention. Information on the distribution of bits is encoded and transmitted via the bitstream.

Figure 4 shows another example of the arrangement of a decoder in accordance with the invention. Information on the distribution of bits is decoded from the bit stream and is used for decoding spectral coefficients.

Figure 5 illustrates a block diagram of a psychoacoustic model that uses the average energy in the bands to calculate the sound pressure level and the masking threshold of hearing. To interpolate points between the centers of the bands are different features of sound propagation in the case of the masking threshold of hearing and sound pressure level.

6 illustrates a block diagram of psychoac the static model, using the spectral coefficients to calculate the sound pressure level and the masking threshold of hearing. Function of sound propagation is used to simulate the masking effect.

7 depicts a block diagram of the bit allocation algorithm, in which the number of pulses in the band is determined by the difference between sound pressure level and a minimum value of the masking threshold. According to the obtained number of pulses is determined by the required number of bits. Then apply restrictions on the number of bits imposed by the FPC algorithm.

Fig illustrates the algorithm for determining the level of a tone signal and applying the diffusion of sound with low computational complexity by eliminating the convolution operation.

Figure 9 shows the scheme of redistribution bit on the conceptual level, where restrictions on the number of bits in the band, resulting in the FPC algorithm.

Figure 10 shows the block diagram of the scaling algorithm bits, to maintain a given speed encoding.

11 illustrates the lookup process noise using a guard interval. The spectral coefficients quantized to zero, reversed the generation of random noise. Guard interval reduces the excess noise of the decoded sound in the case of tonal fragments with the drove.

Fig illustrates the encoder lookup noise. Noise information is calculated based on the energy of the spectral coefficients that quanthouse to zero. Information about the noise is averaged to calculate the overall noise level on the frame.

Fig illustrates a decoding device lookup noise. Noise information is decoded from the data stream and restored by the operation of the inverse quantization. For all bands, with zero quantized coefficients, the random noise is inserted in the signal quality. Guard interval is used to reduce the noise level in the band with the tone fragments of sound.

Fig illustrates the coding and decoding device of the adaptive transmission coefficient FPC. The coefficient FPC is only for bands with increased pitch, in the case of other bands information about the FPC factor is not sent.

Fig illustrates a fast algorithm for the estimation of the number of pulses for a given number of bits. The algorithm consists of two levels. On the first level are computed lower and upper bounds on the number of pulses. At the second level uses a binary search with a small dynamic range.

The block diagram of audiocamera used as the example shown in Figure 1. The input time signal hits the block 101 MDCP. In matter re odny or stationary type of signal was determined,
applied adaptive frequency transformation. For non-stationary signals has a higher time resolution. The spectral coefficients are grouped in bands of unequal lengths and the average energy in the band (norm) is calculated for each band. Standards quanthouse and encoded (blocks 102, 104). In block 106 PAM analyzed the importance of each band or frequency, from the point of view of human perception, and calculates the masking threshold. For transient frames received spectral coefficients of podkatov interspersed before grouping, in order to use the masking effect for neighboring frequencies in block 106 PAM. For example, if each frame consists of four podkatov and minimum strip length is equal to eight, then two factor from each podagra will premiani and divided into strips. In the next step the bits for coding the spectral coefficients are divided between the bands, on the basis of the results of the operation unit 107 to US. From the obtained number of bits of the calculated number of pulses within the bands and the spectral coefficients are encoded using a factorial pulse coding (block 108). The FPC algorithm uses combinatorial scheme of spectral coefficients y={y_{1}, y_{2}, y_{3}the...y_{k-1}}, keeping the minimum RMS error and the restriction on the total number of the number of pulses of

Figure 2 illustrates a block diagram of a decoding device, which corresponds codereuse the device of figure 1. Data stream broken and decrypted in block 211. Data standards are decoded in block 212 and restored in block 216 using the inverse quantization. Information on the distribution of bits obtained in the encoding device is also required for decoding in a decoding device. In order to recover the distribution of bits in the decoding unit uses the same psychoacoustic model (block 206) and on the side of the coding unit by using the decoded rules (block 216). The total number of bits is allocated among the bands in block 207 on the basis of the data psychoacoustic model. The bit allocation algorithm described in more detail in the next section. The spectral coefficients are decoded and restored in block 214, using the operation fo the strong quantization. The coefficients of the spectrum, the decoded zero, restored by substitution of noise in block 215. The recovered spectrum coefficients are scaled in accordance with the rules in block 217. Then apply the inverse transform in block 218 for recovery time signal. Psychoacoustic model is shown in Figure 5 and is based on the approximation of the masking threshold of hearing and sound pressure level, using the average energy in the bands. Initially, the level of a tone signal is estimated for each lane in block 512, in order to obtain an approximation of the sound pressure level:

where

where

The level of a tone signal strips to build the approximation masking threshold of hearing is calculated in block 510:

where

where Bark(i) - frequency scale Barca for the spectral index of the coefficient of i, i and j are the indices of the spectral coefficients, maskHihg and maskLow - factors that determine the gradient of the function of sound propagation. The coefficients are functions of sound propagation (3) and (4) may be different.

The computational complexity psychoacoustic model based on the strips several times less complexity psychoacoustic model for each spectral coefficient. In the implemented invention proposes a fast algorithm for determining the level of a tone signal and applying the diffusion of sound, shown in Fig. In order to reduce the computational complexity of the proposed test is masked if the current level of a tone signal (block 800) previous level tone stripes. If the current level of a tone signal is masked, then any distribution of the audio data level of the tone signal is masked, therefore, there is no need to perform any calculations to the level of a tone signal of the current band. Only for all direct level tone (block 801) evaluate the function of sound propagation. For each level of a tone signal left and right tilt functions of sound propagation is analyzed and determined the point of intersection (block 803). In the end, it calculates the equation of a straight line between the intersection point and level tone (blocks 802 and 804) and calculated once the value of the function of sound propagation. Thus, there is no need to apply convolution in (2) and (4), which significantly reduces the computational complexity. Also, in case of using a psychoacoustic model based on the strips do not need to pass the bit distribution, since it is enough to repeat the same calculations in the decoding device. The output of the psychoacoustic model used for the distribution of bits between the bands of the spectrum.

3 the ill is Trichet another example of the coding device in accordance with the claimed invention. MDCP (block 301) is used to calculate spectral coefficients from the temporary signal. The spectral coefficients are grouped by irregularly distributed stripes. For each band is computed rate representing the average energy in the band. Standards quanthouse in block 302 and encoded in block 304. Psychoacoustic model (block 306) analyzes the subjective importance of each spectral coefficient and calculates a masking threshold of hearing. The total number of bits is distributed between the bands on the basis of psychoacoustic information (block 307). The spectral coefficients quanthouse the FPC algorithm in block 308 in accordance with the bit allocation of the bands. Information on the distribution of bits between the bands is encoded in block 310 and transmitted in the data stream (block 311). For all spectral coefficients quantized to zero, the necessary parameters are calculated noise using a low-speed algorithm in block 309. For example, the coding bit allocation can be organized using vector quantization. In this case, there is no need on the side of the decoding device to perform calculations related to the psychoacoustic model, since the distribution of the bits can be recovered from information transmitted in the data stream. 6 illustrates a block diagram of psychoacoustic the Russian models are built for each spectral coefficient. The sound pressure level for each spectral coefficient is calculated in block 600. The level of tonal and noise signal is calculated in block 601 and 604. Function of sound propagation is used (blocks 602 and 605) for tonal and noise level of the signal to calculate the masking threshold of hearing in the block (603). Received the masking threshold is compared to the absolute hearing threshold in block 606, and selects the largest. Figure 4 illustrates a block diagram of a decoding device, which corresponds codereuse the device of figure 3. The data stream is split and decrypted in block 411. Norms are decoded in block 412 and restored in block 416. The distribution of bits between the bands is decoded from the stream in block 413. The spectral coefficients are decoded and restored in block 414. The coefficients of the spectrum, the decoded zero, restored by substitution of noise in block 215. The decoded spectral coefficients are normalized in accordance with the recovered norms in block 417. In conclusion, the inverse transform MDCP (block 418) is used for recovery time signal.

The purpose of the bit allocation algorithm is to distribute the available bits between the frequency bands. The proposed algorithm uses the masking threshold and the sound pressure levels calculated in modulating the psychoacoustic analysis.

Regardless of the module PAM-based energy spectral bands (106) or spectral coefficients (306), the bit allocation algorithm works similarly. First it calculates the maximum signal-to-masking threshold (CMP) inside the band, instead of using individual SMR for each frequency. This is done due to the fact that the band are encoded together, and the total allowable when the quantization distortion should not exceed specified in the PAM for each frequency. To calculate the maximum CMP, is the minimum value of the masking threshold in the band by the formula 5:

where MTH_{i}is the masking threshold calculated in the MEMORY for the i-th frequency, MTH_{min -}this is the minimum masking threshold among not exceeding the value of energy corresponding spectral coefficients in the band.

The number of selected pulses m one spectral band can be estimated by the formula 6:

where E_{i}mean energy of the i-th frequency in logarithmic scale (dB).

Using the above calculated number of pulses, it is possible to estimate the number of bits per band, using equation 7:

where n is the length of the strip, m is the number of pulses. Formula 7 requires high vychislitelnykhsistem, in the claimed invention, a method of reducing the complexity of the algorithm for estimating the number of bits by the number of pulses. A detailed description of this method will be given below.

As another example of allocation of the bit, the sound pressure level is calculated by the average energy of the band and the minimum masking threshold in the band, which is then used to estimate the number of pulses in the band:

Once the optimal number of bits per band was determined for all bands, it is necessary to bring the total number of bits on all bands to the value defined in the control unit of bit rate. Figure 10 shows the block diagram of this method as proposed in the invention. First, it calculates the number of pulses at one frequency for each band. By adding or animania a controlling factor With this number it is possible to change the total number of bits per frame. The number of pulses that is changed after the control algorithm of bit rate, is calculated as shown in formula 9:

After this limits the bit rate imposed due to the properties of the algorithm FPC. For bands where the number of allocated bits was more possible in the algorithm FPC, over b is t will be stored in the bit reservoir. Similarly, the bits will be extracted from strips, where the number of allocated bits turned out to be less than the minimum possible for FPC algorithm.

Then stored in the tank bits will be evenly distributed between the bands with non-zero number of bits, but less possible for FPC. If all non-null bandwidth is already equal to the maximum threshold, and the bits are still there, then the remainder will be distributed environment most of the low-frequency bands of the spectrum.

Module substitution generates noise random noise as spectral coefficients, which have been encoded and restored as zero. The main idea of the substitution of noise is masking the spectral holes random noise. 11 illustrates the principle of substitution of noise. Fig illustrates a block diagram of a coding device for the substitution of noise. The original and reconstructed spectra are used to determine the position of the spectrum in which you want to apply a substitution of noise. For areas of application of substitution calculates the average noise amplitude noise on the reference block 1200:

where E_{b}the average amplitude of the noise on the reference in the band room b, S_{b}- number of samples quantized to zero for band number b, y_{i}spectral coefficients of the original spectrum in the b band. The amount of the free period of quantized to zero is calculated in block 1201:

where

where E_{b}the average amplitude of the noise in the band b,

The amplitude of the noise (12) is calculated for each band in which there is at least one spectral coefficient, the quantized to zero. Further, the amplitude of the noise is averaged on the frame due to the reduction of bit cost. The average magnitude of the noise on the frame quantized logarithmic scale on 8 levels in block 1203 and the limitation on the maximum value of the quantum in block 1205:

where E_{b}the average amplitude of the noise in the band b, q is the quantum number, F the number of bands where applicable, the substitution of noise. In conclusion, the average amplitude of the noise frame is encoded with three bits in block 1204.

Fig illustrates a block diagram of a decoding device for the substitution of noise. The amplitude of the noise is decoded in block 1301 and is restored in block 1302:

E=2^{-q},

where E is the average amplitude of the mind on the frame,
q is the number of the quantum. The number of samples restored to zero determines in block 1306, as well as in the encoding device according to formula (11). For each band with S_{b≤}2 calculates the sound pressure level in block 1308 on the basis of normal values and the reconstructed spectrum:

where L is the length of the spectrum,

where spl_{i}- sound pressure level for a non-zero reconstructed spectrum, g_{i}i - guard interval, Thr_{j}the threshold to determine whether the use of guard interval with length j.

Random substitution of noise (block 1303) is based on the amplitude of the noise and limitations on the value of the average energy of the band and the maximum amplitude of noise:

where noise(X) is the generator of random noise, in which the average amplitude of the noise is equal to X,_{i}neighboring noise positions:

where_{i±j}- guard interval for reference i±j. The energy of each band is normalized in block 1304 in accordance with the original energy spectrum:

where

Fig illustrates a block diagram of the encoder and decoder devices for adaptive encoding of the coefficient of FPC. Quantization FPC (block 1403) is a uniform scalar quantization, in which the quantum number is the number of pulses:

where z_{i}the number of pulses in position i for strip b, m_{b}the total number of pulses in the band b, y_{i}spectral coefficient at position i in b band. The total number of pulses is determined in block 1400. The quantized spectrum is encoded as the number of combinations in block 1404. Equation (13) is the solution of the conditional extremal problems minimize quantization errors.
This solution uses the FPC factor (block 1401), which manages the energy bands:

where z_{i - b}the number of pulses at position i, y_{i}spectral coefficient at position i, G is the optimal ratio FPC for recovery of spectral coefficients. Thus, the recovery of spectral coefficients is the multiplication of the number of pulses z_{i}factor G:

where z_{i}the number of pulses at position i,

where N_{b -}the length of the strip b, z_{i}the number of pulses at position i. The relationship between the calculated coefficient G (14) and predicted G_{p}(15) is called the error. The transmission of the prediction error in the bit stream is sometimes redundant, since the exact value of the coefficient FPC is only important for tonal areas of the signal. Thus, the lines with tone signal is determined in block 1402 based on the total number of pulses or the number of bits for the band. For example, the criteria can be selected:

where bits is the number of bits in the band, BitsN_{b}- the number of bits for encoding N_{b}pulses, G - factor FPC, G_{p}- predicted ratio FPC. The prediction error coefficient FPC and adaptive quantized transmitted in the data stream in block 1406.

Fig illustrates a block diagram of a decoding device for FPC factor. The total number of pulses is determined in block 1409. The prediction error of the adaptive decoded using the same criteria? as the encoder. The number of combinations of multiple pulses is decoded in block 1410. Quantum spectral coefficients are restored in block 1412. The spectral coefficients are restored in block 1411.

Fig illustrates the proposed two-level algorithm for the estimation of the number of pulses for a given number of bits. This algorithm significantly outperforms the computational complexity of the method based on binary search on the table. On the ground level of the lower and upper bounds on the number of pulses are determined. At the second level performs a binary search in a small range of values. The lower bound on the number of pulses is estimated at block 1500, using the formula (21). The upper bound is estimated in block 1501, using the formula (20). A binary search between the lower and upper bound is used to determine the exact value of p is law in block 1502. This proposal reduces the number of comparisons several times.

In most cases, the number of pulses is calculated using binary search on a given table with a large enough range. The table typically contains more than 512 elements, which corresponds to at least 9 comparisons in the case of binary search. For each comparison is the analysis of the compliance of a given number of bits and is obtained assuming the number of pulses. The dependence of the number of pulses from the number of bits is quite complicated:

where b is the number of bits of the n - length vector with heartbeats, m is the number of pulses.

Expression (16) can be simplified:

where b is the number of bits of the n - length vector with heartbeats, m is the number of pulses z(m,n) be the polynomial defined as:

In fact, in the formula (17) is true equality z(n,m)=z(m,n), because the order of the arguments is not important. Consequently, it is possible to estimate the number of pulses for a given number of bits, using the formula:

where b is the number of bits of the n - length vector with heartbeats, m is the number of pulses, Z(n,m) is a polynomial.

Expressions (17) and (18) are accurate, but the computational complexity them enough height is CA. In this invention it is proposed to divide (18) by two levels. On the first level are estimated lower and upper bound, and the second is the exact value. The lower bound on the estimate of bit cost can be expressed as:

Expression (19) allows you to find lower and upper bounds on the number of pulses depending on the given quantities:

where b is the number of bits of the n - length vector with heartbeats, m is the number of pulses, F - shift, which allows to determine the lower bound on the number of pulses:

where n is the length of the vector with heartbeats. Expression (21) can easily be determined for any value of n using a modeling expressions (20) and (16). In most cases, the length is less than 17, which allows to make a conclusion about the necessity of only one or two comparisons for a binary search on the second level of the proposed algorithm.

The invention can find application in devices for handling digital signals, in particular, to devices for compressing digital signals such as audio signals.

Links

1. ITU-T Rec. G.719, "Low-complexity, full-band audio coding for high-quality, conversational applications," 2008.

2. ITU-T Rec. G.718, "Frame error robust both narrowband and wideband embedded variable bit-rate coding of speech and audio from 8-32 kbt/s," 2008.

3. EN 2427978. The encoding and decoding audio.

4. EN 2428748. The encoding of the audio signal.

5. EN 2432624. A method of reducing the amount of data in wideband coding of speech signals.

1. The encoding method of the temporary audio signal, which consists in the fact that the input signal is converted into spectral coefficients are grouped spectral coefficients in the frequency band, and evaluate standards for each band as the average energy in the band, normalize spectrum on the basis of the estimated norms, characterized in that the bit distribution calculated on the basis of psychoacoustic models based on the quantized standards by performing the following operations:

- the spectral coefficients are grouped by irregularly distributed stripes;

- for each band to calculate the rate representing the average energy in the band;

- norm quantuum and encode;

- analyze using a psychoacoustic model of the subjective importance of each spectral coefficient, and calculates a masking threshold of hearing;

- the total number of bits is distributed between the bands on the basis of psychoacoustic information;

spectral coefficients quantum algorithm factorial encoding pulses (FPC) in accordance with the received bit distribution on the strips;

- information on the distribution of bits between bands is mi encode and transmit the data stream;

for all spectral coefficients quantized to zero, calculate the necessary parameters of the noise using a low-speed algorithm.

2. The method according to claim 1, characterized in that the determination of the allocation of bits based on the criteria of the signal energy to the masking threshold.

3. The method according to claim 1, characterized in that the calculation of the number of pulses based on the criteria of the signal energy to the masking threshold.

4. The method according to claim 3, characterized in that the number of bits is determined by the formula factorial encoding pulses (FPC) from a known number of pulses.

5. The method according to claim 1, characterized in that the calculated population parameters noise for quantized to zero spectral coefficients, with the purpose of masking of failures of the spectrum and the calculated parameters are passed in the data stream.

6. The method according to claim 1, characterized in that the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity, providing the following operations:

- at the first stage determines the lower and upper bounds on the number of pulses;

- on the second level, perform a binary search in a small range of values;

- evaluate the lower bound on the number of pulses, using the formula

- estimate the upper limit on the number of pulses using the ormolu

determine the exact value of pulses based on a binary search between the lower and upper bounds.

7. The method of decoding an encoded audio signal, comprising: decoding and recovery standards, the calculation of the distribution of bits based on the reconstructed standards, spectrum decoding and inverse transform of the spectral coefficients in the time domain signal, wherein the bit distribution calculated by performing the following operations:

- decode and restore data standards using the inverse quantization;

- used for decoding the information on the distribution of bits obtained in the encoding device;

- restore the distribution of bits in the decoding device using the same psychoacoustic model, and on the side of the coding unit by using the decoded standards;

- distribute the total number of bits among the bands on the basis of the data psychoacoustic model;

- decode and recover the spectral coefficients using the inverse quantization;

- restore the spectrum coefficients decoded to zero, the substitution of noise.

8. The method according to claim 7, characterized in that the parameters of the noise decode the data stream and quantized to zero spectral coefficients fill out the noise of the m order masking of failures of the spectrum.

9. The method according to claim 7, characterized in that the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity, providing the following operations:

- at the first stage determines the lower and upper bounds on the number of pulses;

- on the second level, perform a binary search in a small range of values;

- evaluate the lower bound on the number of pulses, using the formula

- estimate the upper limit on the number of pulses, using the formula

determine the exact value of pulses based on a binary search between the lower and upper bounds.

10. The encoding method of the temporary audio signal, which consists in the fact that the input signal is converted into spectral coefficients are grouped spectral coefficients in the frequency band, and evaluate standards for each band as the average energy in the band, normalize spectrum on the basis of the evaluated standards, weighed norms based on psychoacoustic properties of sound, calculated bit allocation based on weighted norms, quantum and encode the spectrum obtained by the number of bits, wherein the bit distribution calculated on the basis of psychoacoustic models based on spectral coefficients performed by the I following operations:

- the spectral coefficients are grouped by irregularly distributed stripes;

- for each band to calculate the rate representing the average energy in the band;

- norm quantuum and encode;

- analyze using a psychoacoustic model of the subjective importance of each spectral coefficient, and calculates a masking threshold of hearing;

- the total number of bits is distributed between the bands on the basis of psychoacoustic information;

spectral coefficients quantum algorithm factorial encoding pulses (FPC) in accordance with the received bit distribution on the strips;

- information on the distribution of bits between the bars encode and transmit the data stream;

for all spectral coefficients quantized to zero, calculate the necessary parameters of the noise using a low-speed algorithm.

11. The method according to claim 10, wherein the psychoacoustic properties of the signal estimate based on the coefficients of the modified discrete cosine transform (MDCP).

12. The method according to claim 10, characterized in that the bit distribution quantuum and passed as additional information.

13. The method according to claim 10, characterized in that the determination of the allocation of bits based on the criteria of the signal energy to the masking threshold.

14. The method according to item 13, characterized in that Thu is the calculation of the number of pulses based on the criteria of the signal energy to the masking threshold.

15. The method according to 14, characterized in that the number of bits is determined by the formula factorial pulse coding (FPC) from a known number of pulses.

16. The method according to claim 10, characterized in that the calculated population parameters noise for quantized to zero spectral coefficients, with the purpose of masking of failures of the spectrum, the parameters passed in the data stream.

17. The method according to claim 10, characterized in that the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity, providing the following operations:

- at the first stage determines the lower and upper bounds on the number of pulses;

- on the second level, perform a binary search in a small range of values;

- evaluate the lower bound on the number of pulses, using the formula

- estimate the upper limit on the number of pulses, using the formula

determine the exact value of pulses based on a binary search between the lower and upper bounds.

18. The method of decoding an encoded audio signal, comprising: decoding and recovery standards, the calculation of the distribution of bits based on the reconstructed standards, spectrum decoding and inverse transform of the spectral coefficients in the signal in the time domain, the ex is different, however,
the distribution of bit decode of the data stream by performing the following operations:

data flow split and decode;

data standards decode and restore using the inverse quantization;

- information on the distribution of bits obtained in the encoding device, used for decoding;

- restore the distribution of bits in the decoding device using the same psychoacoustic model, and on the side of the coding unit by using the decoded standards;

- distribute the total number of bits among the bands on the basis of the data psychoacoustic model;

- decode and recover the spectral coefficients using the inverse quantization;

- restore the spectrum coefficients decoded to zero, the substitution of noise.

19. The method according to p, wherein the noise parameters decode the data stream and quantized to zero spectral coefficients filled with noise with the purpose of masking of failures of the spectrum.

20. The method according to p, characterized in that the number of pulses specified by bits in the band is determined using a two-stage algorithm with low computational complexity, providing the following operations:

- at the first stage determines the lower and upper bounds on the number of pulses;

on the second level, the implementation of AWT binary search in a small range of values;

- evaluate the lower bound on the number of pulses, using the formula

- estimate the upper limit on the number of pulses, using the formula

determine the exact value of pulses based on a binary search between the lower and upper bounds.

21. Device for encoding/decoding an audio signal that contains the encoder and its associated decoder, characterized in that

the encoder includes the following blocks:

block modified discrete cosine transform (MDCP), configured to convert the input signal into spectral coefficients;

the evaluation unit and the quantization rules made with the possibility of grouping the spectral coefficients in the frequency band and evaluation standards for each band as the average energy in the band;

block coding standards;

the building block psychoacoustic model quantized by regulations made with the possibility of analysis of subjective importance of each spectral coefficient;

the first block of the bit allocation calculation made with the possibility of calculating the bit allocation based on the subjective importance of each spectral coefficient;

block quantization and coding of the spectrum made with the possibility of coding the spectrum obtained by the number of bits.

the mule is diplexer for transmission of the coded data in the bitstream;

the decoder includes the following sequentially linked blocks:

demultiplexer, configured to break and decrypt the data stream;

the block decoding standards;

the unit dekvantovanie Nord;

the building block psychoacoustic model restored regulations;

the second set of bit allocation calculation performed by the calculation of the distribution of bits based on the data psychoacoustic models based on recovered norms

the block decoding and dekvantovanie spectrum with spectrum decoding based on the bit distribution,

block scaling the decoded spectral coefficients in accordance with the recovered norms

block inverse transform of the spectral coefficients in the signal in the time domain.

22. The device according to item 21, wherein the encoder further comprises a computing unit noise parameters for quantized to zero spectral coefficients, transmitting the calculated parameters in the data stream.

23. The device according to item 21, wherein the decoder further comprises a substitute block noise made with the possibility of the recovery of substitution decoded noise at zero spectral coefficients.

24. Device for encoding/decoding an audio signal, containing the its encoder and an associated decoder,
characterized in that

the encoder includes the following blocks:

block modified discrete cosine transform (MDCP), configured to convert the input signal into spectral coefficients;

the evaluation unit and the quantization rules made with the possibility of grouping the spectral coefficients in the frequency band and evaluation standards for each band as the average energy in the band;

block coding standards;

the building block psychoacoustic model spectral coefficients, configured to determine the subjective importance of each spectral coefficient;

the block bit allocation calculation made with the possibility of calculating the bit allocation based on the subjective importance of each spectral coefficient;

block quantization and coding of the spectrum made with the possibility of coding the spectrum obtained by the number of bits.

block coding bit allocation;

the multiplexer for transmission of the coded data in the bitstream;

the decoder includes the following blocks:

demultiplexer, configured to break and decrypt the data stream;

the block decoding standards;

the unit dekvantovanie standards;

the block decoding of bit allocation, the input of which receives the data from the stream;

the block decoding, dequan the Finance spectrum,
on which input data concerning the distribution of the bit data stream;

the unit normalization of the decoded spectral coefficients in accordance with the recovered standards;

block inverse transform of the spectral coefficients in the signal in the time domain.

25. The device according to paragraph 24, wherein the encoder further comprises a computing unit noise parameters for quantized to zero spectral coefficients, transmitting the calculated parameters in the data stream.

26. The device according to paragraph 24, wherein the decoder further comprises a substitute block noise made with the possibility of the recovery of substitution decoded noise at zero spectral coefficients.

**Same patents:**

FIELD: radio engineering, communication.

SUBSTANCE: method of transmitting information bits includes a step of dividing the information bits to be transmitted into at least two groups. Further, according to the method, the information bits in each group to be transmitted are encoded to obtain at least two groups of encoded bits. Said at least two groups of encoded bits are combined to obtain a full sequence of encoded bits. The full sequence of encoded bits is obtained by dividing the encoded bits in each group into N subgroups and reordering said subgroups in each group of encoded bits. Subgroups in at least one group of the encoded bits are discontinuously distributed in the full sequence of encoded bits after reordering.

EFFECT: improved reception quality.

16 cl, 9 dwg, 2 tbl

FIELD: radio engineering, communication.

SUBSTANCE: apparatus for decoding block turbo codes has a first random-access memory unit 1, a second random-access memory unit 2, a third random-access memory unit 3, a SISO decoder 4, a decision unit 5, a first limiter 6, a read-only memory unit 7, a multiplier unit 8, a second limiter 9. The SISO decoder has a random-access memory unit 10, a clock generator 11, a switch 12, a counter 13, a read-only memory unit 14, a Walsh function coefficient signal former 15, an analysed sequence former 16, a first adder 17, a first subtractor unit 18, a doubling unit 19, a multiplier unit 20, a first divider unit 21, a second adder 22, a third adder 23, a second subtractor unit 24, a second divider unit 25, a third divider unit 26, a limiter 27.

EFFECT: high noise immunity of block turbo codes.

3 cl, 6 dwg

FIELD: information technology.

SUBSTANCE: transmitting device comprises: means of generating frames, which is configured to arrange signal and pilot signal data in each of at least two signal code combinations in a frame, each signal code combination having the same length, and arrange data in said at least one code combination in a frame, a conversion means which is configured to convert said signal code combinations and said data code combinations from a frequency domain into a time domain to generate a time-domain transmission signal, and a transmitting means which is configured to transmit said time-domain transmission signal. Method is intended to be implemented by the given device.

EFFECT: enabling flexible tuning to the required portion of the transmission band and reduced content of service data.

20 cl, 15 dwg

FIELD: information technology.

SUBSTANCE: intra prediction modes are coded in a bit stream. Brightness and chroma components can potentially have different prediction modes. For chroma components, there are 5 different modes defined in AVC: vertical, horizontal, DC, diagonal down right, and "same as brightness". Statistics show that the "same as brightness" mode is frequently used, but in AVC, this mode is encoded using more bits than other modes during entropy coding, therefore the coding efficiency is decreased. Accordingly, a modified binarisation/codeword assignment for chroma intra mode signalling can be used for high efficiency video coding (HEVC), the next generation video coding standard.

EFFECT: high coding efficiency.

18 cl, 4 dwg

FIELD: radio engineering, communication.

SUBSTANCE: method of generating codes for generating signal ensembles involves generating a source code of N≥4 elements, a number K≥1 of codes of N elements to be generated, as well as a target function for a set of L states of the code elements, and corresponding values of given signal parameters, characterised by an array of states of L×N×K peaks on N×K levels, connected by edges, wherein each of the L states is the initial state; generating codes; selecting a path with the extremum value of the target function, after which each generated code is assigned a symbol which corresponds to the edge of the path with the extremum value of the target function, and selecting 2≤M≤K codes with the maximum value of the ratio of the amplitude of the central peak of the autocorrelation function to the magnitude of the amplitude of the maximum lateral peak of the autocorrelation function and the minimum duration of the section of the autocorrelation function between the point of the maximum of the central peak and the point where the autocorrelation function becomes zero for the first time.

EFFECT: high jamming resistance of signals generated based on corresponding codes.

5 cl, 7 dwg

FIELD: radio engineering, communication.

SUBSTANCE: receiving apparatus, which corresponds to the digital television standard T.2, known as DVB-T2, is configured to perform low-density parity-check (LDPC) decoding for physical layer channels (PLC), which denote data streams, and layer 1 (L1), which represents physical layer transmission parameters. The receiving apparatus includes a LDPC decoding apparatus which is configured such that, when a LDPC encoded data signal and a LDPC encoded transmission control signal are transmitted multiplexed, said LDPC decoding apparatus decodes both the data signal and the transmission control signal. The receiving apparatus also includes a storage device configured to be placed in front of the LDPC decoding device and to store the transmission control signal when receiving the data signal and the transmission control signal.

EFFECT: enabling simultaneous reception of data and control signals using the same apparatus.

4 cl, 12 dwg

FIELD: radio engineering, communication.

SUBSTANCE: receiving apparatus, which corresponds to the digital television standard T.2, known as DVB-T2, is configured to perform low-density parity-check (LDPC) decoding for physical layer channels (PLC), which denote data streams, and layer 1 (L1), which represents physical layer transmission parameters. The receiving apparatus includes a LDPC decoding apparatus which is configured such that, when a LDPC encoded data signal and a LDPC encoded transmission control signal are transmitted multiplexed, said LDPC decoding apparatus decodes both the data signal and the transmission control signal. The receiving apparatus also includes a storage device configured to be placed in front of the LDPC decoding device and to store the transmission control signal when receiving the data signal and the transmission control signal.

EFFECT: enabling simultaneous reception of data and control signals using the same apparatus.

4 cl, 12 dwg

FIELD: radio engineering, communication.

SUBSTANCE: method of generating codes for generating signal ensembles involves generating a source code of N≥4 elements, a number K≥1 of codes of N elements to be generated, as well as a target function for a set of L states of the code elements, and corresponding values of given signal parameters, characterised by an array of states of L×N×K peaks on N×K levels, connected by edges, wherein each of the L states is the initial state; generating codes; selecting a path with the extremum value of the target function, after which each generated code is assigned a symbol which corresponds to the edge of the path with the extremum value of the target function, and selecting 2≤M≤K codes with the maximum value of the ratio of the amplitude of the central peak of the autocorrelation function to the magnitude of the amplitude of the maximum lateral peak of the autocorrelation function and the minimum duration of the section of the autocorrelation function between the point of the maximum of the central peak and the point where the autocorrelation function becomes zero for the first time.

EFFECT: high jamming resistance of signals generated based on corresponding codes.

5 cl, 7 dwg

FIELD: information technology.

SUBSTANCE: intra prediction modes are coded in a bit stream. Brightness and chroma components can potentially have different prediction modes. For chroma components, there are 5 different modes defined in AVC: vertical, horizontal, DC, diagonal down right, and "same as brightness". Statistics show that the "same as brightness" mode is frequently used, but in AVC, this mode is encoded using more bits than other modes during entropy coding, therefore the coding efficiency is decreased. Accordingly, a modified binarisation/codeword assignment for chroma intra mode signalling can be used for high efficiency video coding (HEVC), the next generation video coding standard.

EFFECT: high coding efficiency.

18 cl, 4 dwg

FIELD: information technology.

SUBSTANCE: transmitting device comprises: means of generating frames, which is configured to arrange signal and pilot signal data in each of at least two signal code combinations in a frame, each signal code combination having the same length, and arrange data in said at least one code combination in a frame, a conversion means which is configured to convert said signal code combinations and said data code combinations from a frequency domain into a time domain to generate a time-domain transmission signal, and a transmitting means which is configured to transmit said time-domain transmission signal. Method is intended to be implemented by the given device.

EFFECT: enabling flexible tuning to the required portion of the transmission band and reduced content of service data.

20 cl, 15 dwg

FIELD: radio engineering, communication.

SUBSTANCE: apparatus for decoding block turbo codes has a first random-access memory unit 1, a second random-access memory unit 2, a third random-access memory unit 3, a SISO decoder 4, a decision unit 5, a first limiter 6, a read-only memory unit 7, a multiplier unit 8, a second limiter 9. The SISO decoder has a random-access memory unit 10, a clock generator 11, a switch 12, a counter 13, a read-only memory unit 14, a Walsh function coefficient signal former 15, an analysed sequence former 16, a first adder 17, a first subtractor unit 18, a doubling unit 19, a multiplier unit 20, a first divider unit 21, a second adder 22, a third adder 23, a second subtractor unit 24, a second divider unit 25, a third divider unit 26, a limiter 27.

EFFECT: high noise immunity of block turbo codes.

3 cl, 6 dwg

FIELD: radio engineering, communication.

SUBSTANCE: method of transmitting information bits includes a step of dividing the information bits to be transmitted into at least two groups. Further, according to the method, the information bits in each group to be transmitted are encoded to obtain at least two groups of encoded bits. Said at least two groups of encoded bits are combined to obtain a full sequence of encoded bits. The full sequence of encoded bits is obtained by dividing the encoded bits in each group into N subgroups and reordering said subgroups in each group of encoded bits. Subgroups in at least one group of the encoded bits are discontinuously distributed in the full sequence of encoded bits after reordering.

EFFECT: improved reception quality.

16 cl, 9 dwg, 2 tbl

FIELD: information technology.

SUBSTANCE: input signal is converted to spectral coefficients; the spectral coefficients are grouped into frequency bands and standards are estimated for each band as the average energy in the band; the spectrum is normalised based on the estimated standards; the standards are weighted based on psycho-acoustic properties of sound; bit distribution is calculated based on the weighted standards; the spectrum is quantised and encoded by the obtained number of bits; the method is characterised by that bit distribution is calculated based on a psycho-acoustic model built on quantised standards. Also disclosed is a device for implementing this method.

EFFECT: low level of distortions and easier encoding.

26 cl, 15 dwg

FIELD: radio engineering, communication.

SUBSTANCE: information 1 consisting of five pulses is encoded in form of a series of one positive pulse, two positive pulses, each magnified N times, one negative pulse magnified N times and one positive pulse, and an information 0 consisting of five pulses is encoded in form of a series of one negative pulse, two negative pulses, each magnified N times, one positive pulse magnified N times and one negative pulse, wherein N is a positive number greater than 1; the obtained sequences are transmitted to a data transmitting medium, and the received signal is compared with a reference signal by cross-correlation at the receiving side.

EFFECT: obtaining a clear signal with high level of interference and longer range of signal transmission.

2 cl, 5 dwg

FIELD: radio engineering, communication.

SUBSTANCE: invention relates to a coding method in a wireless mobile communication system. More specifically, the present invention relates to a convolutional turbo coding (CTC) method and a device for implementing the method. The method for CTC includes steps of encoding information bits A and B using a constituent encoder, and outputting parity sequences Y_{1} and W_{1}, interleaving the information bits A and B using a CTC interleaver to obtain information bits C and D, and encoding the interleaved information bits C and D using the constituent encoder to obtain parity sequences Y_{2} and W_{2}, interleaving the information bits A and B, the parity sequences Y_{1} and W_{1} and the parity sequences Y_{2} and W_{2}, respectively, wherein the bits in at least one of a bit group consisting of the information bits A and B, a bit group consisting of the sequences Y_{1 }and W_{1}, and a bit group consisting of the sequences Y_{2} and W_{2} are alternately mapped to bits of constellation points with high reliability and low reliability and puncturing the interleaving result to obtain the encoded bit sequences.

EFFECT: high reliability of encoding with bit mapping of high order modulation.

12 cl, 7 dwg

FIELD: radio engineering, communication.

SUBSTANCE: method of decoding convolutional codes involves receiving radio signals, automatic gain control, demodulation, first deinterleaving, Viterbi algorithm decoding, amplitude detection, averaging, second deinterleaving, nonlinear conversion and multichannel multiplication-summation.

EFFECT: low error probability when decoding and high noise-immunity of transmitted information.

7 dwg