Method and device search in the directory codes to encode the audio signal and the cellular communication system

 

(57) Abstract:

To encode the audio signal search a Handbook of codes. This reference code consists of many combinations of amplitudes and positions of pulses, each of which defines L different positions and includes pulses of zero amplitude, and the pulse nonzero amplitude assigned to the respective positions p = 1, 2,..., L the combination, with each pulse nonzero amplitude assumes the presence of at least one of q possible amplitudes. The technical result is to reduce the complexity of the search. To reduce the complexity of the search, pre-select a subset of combinations of amplitudes and positions of the pulses of the reference codes in connection with a sound signal and carry out the search only in the subset of combinations. Pre-selecting a subset of the combinations is that pre-set in connection with a sound signal, the functional dependence of the SP between the respective positions p = 1, 2,..., L and q possible amplitudes, with the search limited to those combinations of reference codes with non-zero pulses of amplitude, which correspond to the prior is by pre-assigning one of the q possible amplitudes of each position p, with pre-installed functional dependence enters into force when each pulse is non-zero amplitude combination has an amplitude equal to the amplitude of the Sp, the pre-assigned position p of the pulse. 3 C. and 24 C.p. f-crystals, 5 Il.

The background to the invention

The technical field

The present invention relates to an improved method of digitally encoding a sound signal, in particular, but not exclusively, of the speech signal, in connection with the transfer and synthesis of such a sound.

Prior art

Currently there is a growing need for effective methods of digital coding of speech, providing acceptable subjective compromise between quality and speed of data transmission in bits, for use in various systems, for example when sending voice messages via satellites, terrestrial mobile stations, digital radio or packet network, devices for storing voice messages, the device for transmitting the reply voice message and radiotelephony.

One of the best known ways in which to achieve an acceptable compromise between quality and Korolev). In accordance with this method, make a selection and processing of the speech signal in blocks of L samples (i.e., vectors), where L is a given number. The way LPCV based on the use of reference codes.

In the context of LPCV codes directory is indexed by the set of all sequences of length L samples, these sequences will be referred to as L-dimensional code vectors (combinations of pulses defining L different positions and containing pulses of zero amplitude, and the pulse nonzero amplitude assigned to the respective positions p = 1, 2,... L combination). The codes directory contains the index k in the range from 1 to M, where M represents the size of the reference codes, sometimes expressed by the number of bits b:

M = 2b.

Reference codes can be stored in physical memory (e.g., as a screening table), or you can use the mechanism that links the index with the corresponding code vector (for example, the appropriate formula).

To synthesize speech in accordance with the method LPCV, each block of samples of the speech synthesized by filtering the appropriate code vector from the reference codes using gestational synthesized output signal to all or some subset of the code vector candidates from the directory codes (search in directory codes). It fixed code vector gives the synthetic output signal, which is the closest to the original speech signal, respectively, somewhat perceptual weighted distortion.

The first type of reference codes are the so-called "stochastic" reference codes. The disadvantage of these directories is that they often occupy a significant amount of physical memory. They are stochastic, i.e., random in the sense that the path from the index to the associated code vector uses lookup tables, which are the result of combining the randomly generated numbers, or statistical methods used to great speech training data sets. The volume of the stochastic reference codes, as a rule, limited memory and/or the complexity of the search.

The second type of reference codes are algebraic reference codes. Unlike stochastic reference codes, algebraic reference codes are not random and do not require memory. Algebraic codes directory - ordered this set of code vectors, where the amplitude and position of the pulses of the k-th code vector can receive the parameter memory. Therefore, the volume of algebraic codes directory is not limited by memory requirements. Algebraic reference codes can also be used for effective search.

The present invention is to provide a method and device that provides a significant reduction in the complexity of the search codes directory after encoding the audio signal, applicable to a large class of reference codes.

Another objective of the present invention is to provide a method and device that enables a priori selection of a subset of combinations of pulses of the reference codes and restrictions subject to search combinations of this subset to reduce the complexity of the search code.

An additional object of the present invention is to increase the size of the reference codes, allowing individual pulses nonzero amplitude code vector, presumably with at least one of q possible amplitudes, without increasing the complexity of the search.

The invention

More specifically, in accordance with the present invention, developed and manner of performing the search in the background is ITUD and positions of the pulses, each combination of amplitudes and positions of the pulses determines the L different positions and includes pulses of zero amplitude, and a pulse of zero amplitude, assigned to the respective positions p = 1, 2,... L combination, and each pulse has a non-zero amplitude assumes the presence of one of the q possible amplitudes, namely, that

pre-selected from reference codes a subset of combinations of amplitudes and positions of the pulses in connection with a sound signal, and

search only in this subset combinations of the amplitudes and positions of the pulses to encode the audio signal, thereby reducing the complexity of the search, because the search carried out in only one subset combinations of the amplitudes and positions of the pulses of the reference codes.

Pre-selection includes preliminary determination, in connection with a sound signal, the functional dependence of Sppre-set positions p = 1, 2, ... L the actual amplitude of the q possible amplitudes, and the step of searching includes searching only those combinations of amplitudes and positions of the pulses of the reference codes, pulses having nonzero amplitudes, which correspond to the pre-mouth is but a device for implementing directory search codes to encode the audio signal, moreover, the reference code consists of many combinations of amplitudes and positions of the pulses, each combination of amplitudes and positions of the pulses determines the L different positions and includes pulses of zero amplitude and the non-zero pulses in amplitude, assigned to the respective positions p = 1, 2, . .. L combinations, and each pulse has a non-zero amplitude assumes the presence of one of the q possible amplitudes containing

tool pre-selection of reference codes a subset of combinations of amplitudes and positions of the pulses in connection with a sound signal and

search only in this subset combinations of the amplitudes and positions of the pulses due to the encoding of the audio signal, which decreases the complexity of the search, because the search is performed only in one subset combinations of the amplitudes and positions of the pulses of the reference codes.

Means of pre-selecting includes means first establishing, in connection with a sound signal, the functional dependence of Sppre-set positions p = 1, 2,... L the actual amplitude of the q possible amplitudes, and the search tool provides a means of limiting the search to those combinations amout predetermined functional dependence.

Also in accordance with the present invention, designed a cellular communication system for servicing a large area, divided into a number of cells containing

mobile transceiver units,

cellular base stations are respectively located in said cells,

controls communication between the cellular base stations,

subsystem two-way radio communications between each mobile unit located in one of the cells, and the cellular base station of the cell containing the mobile unit and the cellular base station transmitter includes a means of encoding the speech signal and means for transmitting the encoded speech signal, and the receiver includes means receiving the transmitted encoded speech signal and means for decoding the received encoded speech signal.

A means of encoding the speech signal contains a device for searching the reference codes for coding the speech signal, and the reference code consists of many combinations of amplitudes and positions of the pulses, each combination of amplitudes and positions of the pulses determines the L different positions and contains zero pulses is but every impulse of nonzero amplitude assumes the presence of one of the q possible amplitudes, the device to search for contains

tool pre-selection of reference codes a subset of combinations of amplitudes and positions of the pulses in relation to the speech signal and

search only in this subset combinations of the amplitudes and positions of the pulses for encoding the speech signal, which decreases the complexity of the search, because the search is performed only in one subset combinations of the amplitudes and positions of the reference codes,

the tool pre-selection includes a tool pre-setting, in connection with a sound signal, the functional dependence of Sppre-set positions p = 1, 2,... L the actual amplitude of these q possible amplitudes, and the search tool provides a means of limiting the search to those combinations of amplitudes and positions of the pulses of the reference codes, pulses having nonzero amplitudes, which correspond to the pre-installed functional dependency.

In accordance with a preferred embodiment of the invention, by means of functional dependencies Spone of the q possible amplitudes of the pre-field groups (authoriz the I dependence is effective, when each pulse of the non-zero amplitude combinations of amplitudes and positions of the pulses has an amplitude equal to the amplitude of the Sppre-assigned position p of the pulse is non-zero amplitude.

Preferably, pre-assigning one of the q possible amplitudes of each position p is that

process the audio signal with getting back the filtered target signal D and the residual signal R' to the remote main tone,

compute the vector B evaluation of the amplitude is inversely filtered target signal D and the residual signal R' to the remote main tone, and

for each position p quantuum assessment Bpthe amplitude of the vector B to obtain the amplitude selected for the position p.

The calculation of the vector B of amplitude evaluation primarily involves adding back the filtered target signal D in normalized form

< / BR>
residual signal R' to the remote main tone in normalized form

< / BR>
to obtain the vector B evaluation of the amplitude in the form

< / BR>
where is a fixed constant, preferably having a value between 0 and 1.

According to another preferred variant implementation Dodi vector B using the following expression:

< / BR>
where the denominator is

< / BR>
is a normalization factor, which represents the maximum amplitude of the pulses nonzero amplitude.

Each combination of pulses can contain a number N of non-zero pulses of the amplitude and the position p of non-zero pulses of the amplitude is preferably limited in accordance with at least one code permutation N of intermittent single pulses.

Search in the reference code preferably includes the maximization of the given fraction with a denominatork2computed by means of N nested loops in accordance with the following relationship:

ak2= U'(p1p1) + U'(p2p2) + 2 U'(p1p2) + U'(p3p3) + 2 U'(p1p3) + 2 U'(p2p3). . . . . . ... + U'(pNpN) + 2 U'(p1pN) + 2 U'(p2pN) + ... 2 U'(pN-1pN)

where the calculation for each cycle recorded in a separate line from the outermost loop to the extreme inner loop of the N nested loops, where pn- the position of the n-th pulse is non-zero amplitude combinations and where U'(pxpy) function that depends on the amplitude , pre-assigned positions pxnom calculation at least at the inner loop of the N nested loops can be skipped whenever we have the following inequality:

< / BR>
where Spnthe amplitude of the pre-assigned positions pnDpn- pn-th component of the target vector D, and TD- threshold value associated with back filtered target vector d

The objectives, advantages and other features of the present invention are explained in the following description of a preferred variant of the invention, given only as an example and with reference to the accompanying drawings.

Brief description of drawings

On the accompanying drawings show the following:

Fig.1 is a block diagram of the encoder of the audio signal containing the amplitude selector and optimizing the control device in accordance with the present invention,

Fig. 2 is a block diagram of a decoding device associated with the encoding device shown in Fig.1,

Fig. 3a is a sequence of basic operations for quick search in the directory codes in accordance with the present invention, based on the amplitudes of the pulses selected in relation to the signal,

Fig. 3c is a sequence of operations performed when searching using the N nested loops, in which the extreme inner loop is skipped whenever the contribution of the rst N-1 pulses in the numerator DAKTis considered to be insufficient,

Fig.4 is a schematic representation of the N nested loops used to search the directory codes, and

Fig. 5 is a block diagram illustrating the infrastructure of a typical cellular communication system.

A detailed description of the preferred option exercise

In Fig.5 shows a typical infrastructure of the cellular communication system 1.

Although the use of the method and device for searching of the present invention, in the cellular system disclosed in this description as an example, do not enforce any restrictions, it should be borne in mind that the method and apparatus can be used with the same benefits in many other types of communication systems that require the encoding of the audio signal.

In the cellular system, for example, in system 1, the telecommunication service is provided within a larger territorial areas by splitting this big area for some quantity which I radio audio and data channels.

The signaling channels for Radiocommunication used for the search of cellular phones (mobile transceiver blocks), for example 3, within the service area (cell) of a base station of a cell and to establish communication calls with other phones as within the cell of the base station, and outside or through another network, such as the public switched telephone network of General use (PSTN) 4.

As soon as the phone 3 communicating on call or received a call, set or audio information channel with the cellular base station 2 corresponding to the cell in which the mobile phone 3, and the connection between the base station 2 and the mobile phone 3 to this audio or data channel. Apparatus 3 can also receive information management or synchronization channel signalling during call.

If the phone 3 leaves the cell during a call and enters another cell, it transfers the call to an existing audio or information channel in the new cell. Similarly, if the current call is not, by channel alarm is sent to the control message, so other communication in a wide territorial area.

The cellular communication system 1 also contains the terminal 5 to manage communication between the cellular base stations 2 and the PSTN 4, for example, during a communication session between the mobile phone 3 and the PSTN 4, or between the mobile phone 3 in the first cell and the mobile phone 3 in the second cell.

Of course, to establish a link between each phone 3, placed in one cell, and the cellular base station 2 in this cell, the required subsystem two-way radio. This two-way radio system typically contains in the apparatus 3 and the cellular base station 2 transmitter for encoding the speech signal and for transmitting the encoded speech signal through the antenna, for example 6 or 7, and a receiver for receiving the transmitted encoded speech signal through the same antenna 6 or 7 and to decode the encoded speech signal. As is well known to specialists in this field of technology to reduce the bandwidth required for voice transmission system two-way radio, for example between a mobile phone 3 and the base station 2 requires the encoding of voice messages.

The present invention is the creation of an effective method of digital coding of speech with brieanna transmission of voice signals between the cellular base station 2 and the mobile phone 3 through audio or data channel. In Fig.1 depicts a block diagram of the digital coding of speech suitable for the implementation of this effective method.

Device for encoding speech, is shown in Fig.1, is identical to the device of the speech code, is shown in Fig.1 primary application for U.S. patent N 07/927528, which introduced the selector 112 amplitude in accordance with the present invention. The primary application for U.S. patent N 07/927528 filed September 10, 1992, "Dynamic reference codes for efficient speech coding based on algebraic codes".

The analog speech signal is sampled and subjected to the processing unit. It should be borne in mind that the present invention is not limited to the application to the speech signal. You can also perform encoding other types of sound.

In the above example, the block discrete input speech signal S (Fig. 1) contains L consecutive samples. In accordance with the method LPCV, L is called the length of the "subsets of the data, and it usually ranges from 20 to 80. In addition, blocks of L samples is called the L-dimensional vectors. B the encoding process get different L-dimensional vector. A list of these vectors, which is shown in Fig.1 and 2, as well as the vector of the input speech signal,

R' is the residual vector with a remote the main tone,

X is the target vector,

D - back filtered target vector,

Akcode vector index k of algebraic codes directory, and

Ck- vector update (filtered code vectors).

The list of parameters to pass

k - the index of the code vector (input parameter algebraic codes directory),

g - gain,

CEC (STP) parameters short-term forecast (define A(z)), and

Chipboard (LTP) parameters long-term forecast (which determine the amplification factor b of the main tone and the delay T of the fundamental tone).

The principle of decoding

It seems preferable to describe the first device encoding speech, is shown in Fig.2, illustrating different steps performed when switching from digital input signal (input demultiplexer 205) to the discretized output speech signal (the output of the synthesizing filter 204).

The demultiplexer 205 distinguishes four different parameter of the binary information received from the digital input channel, namely the index k, the gain g, the parameters of the short-term forecast of the PSC and settings long is aetsa below.

The decoding device of the speech shown in Fig. 2, includes a dynamic reference codes 208, consisting of generator 201 algebraic codes and adaptive pre-filter 202, an amplifier 206, an adder 207, device 203 long-term forecasting and synthesizing filter 204.

At the first stage of algebraic codes generator 201 generates a code vector of Akin response to the index k.

In the second stage code vector Akprocessed by the adaptive pre-filter 202, which serves the parameters of short-term forecast of the PSC and/or the parameters of the long term forecast chipboard, to obtain the output vector pack Ck. The purpose of the adaptive pre-filter 202 is in the dynamic control of the frequency content of the output vector pack Ckto improve the quality of speech, i.e., reducing audible distortion caused by frequency irritating to the human ear. A typical transfer function F(z) for the adaptive pre-filter 202 below

< / BR>
Fa(z) is the formant pre-filter, in which 0 < Y1< Y2< 1 are constants. This pre-filter improves the shape of the B>(z) is the pre-filter main tone in which T is the time-varying delay of the fundamental tone, and b0either a constant or a value equal to the quantized parameter long-term forecast of the basic tone from the current or previous groups of data. Fb(z) acts very effectively to increase the frequencies of the harmonics of the fundamental tone at all baud rates. Therefore F(z), as a rule, includes pre-filter main tone, sometimes in combination with formant pre-filter, namely:

F(z) = Fa(z)Fb(z).

In accordance with the method LPCV, the discretized output speech signal is obtained by first scaling vector pack Ckfrom the directory codes 208 using the gain g provided by amplifier 206. Then, the adder 207 adds the scaled signal gCkwith the output signal E (component of long-term prediction excitation signal synthesizing filter 204) device 203 long-term forecasting, which serves the parameters of the DSP included in the feedback circuit and having a transfer function B(z) defined as follows:

B(z) = bz-T,krepresents the excitation signal synthesizing filter 204, which has a transfer function 1/A(z) (A(z)) is determined in the following description). Pre-filter 204 provides the correct formation of the spectrum in accordance with the latest accepted parameters of the PSC. More specifically, the filter 204 simulates the resonance frequencies (formants) of speech. The output block is synthesized sampled speech signal, which can be converted into an analog signal by proper filtering to eliminate the effects of aliasing, which is well known in the technique.

There are many ways to build a generator 201 algebraic codes. The preferred method disclosed in the aforementioned application for U.S. patent N 07/927528, is to use at least one code permutation N of intermittent single pulses.

This idea can be illustrated using a simple generator 201 algebraicas is plitude Sp1, Sp2, Sp3, Sp4, Sp5. Here piindicates the location of the i-th pulse in the subset of the data (i.e. piis in the range from 0 to L-1). Assume that the pulse Sp1limited to eight possible positions p1as follows:

p1= 0, 5, 10, 15, 20, 25, 30, 35 = 0+8m1, M1=0,1 ... 7.

Within these eight items, which can be called "track" N 1, it is possible to freely conduct mutual permutation Sp1and seven pulses of zero amplitude. This code permutation of single pulses. The alternation of five such "codes permutations of single pulses can be obtained by limiting exposures of the other pulses in the same way (i.e., track # 2 track # 3 track # 4 track # 5).

p1- 0, 5, 10, 15, 20, 25, 30, 35 - 0+8m1,

p2= 1, 6, 11, 16, 21, 26, 31, 36 = 1+8m2,

p3= 2, 7, 12, 17, 22, 27, 32, 37 - 2+8m3,

p4= 3, 8, 13, 18, 23, 28, 33, 38 = 3+8m4,

p5= 4, 9, 14, 19, 24, 29, 34, 39 = 4+8m5.

Note that the integers mi= 0, 1,..., 7 fully define the position of pieach pulse Spi.

Thus, a simple index position kpcan be obtained through direct SUB> + m5.

It should be emphasized that the preceding tracks pulses can be obtained and other references code.

For example, you can only use 4 of the pulse, where the first 3 pulse position in the first three tracks, respectively, while the fourth pulse is either the fourth or fifth track, one bit indicates which track is occupied. This configuration allows you to create reference codes with the positions of 13 bits.

In the prior art, it was assumed that the non-zero pulses have a fixed amplitude amplitude for all practical purposes, for reasons of complexity of the search code vector. In fact, if we can assume that the momentum of Spihas one of q possible amplitudes, the search process will have to consider qNpossible combinations of pulses and amplitudes. For example, assuming that 5 pulses from the first example can be one of the q = 4 possible values of the amplitude, for example, Spi= +1, -1, +2, -2, instead of a fixed amplitude, the size of the algebraic code reference increases from 15 to 15+(5x2) bit = 25 bits, which is a thousand times more complicated.

In the present invention billsof q, amplitudes at low cost. The solution is to search for a certain limited subset of the code vectors. The method of selecting the code vectors associated with the input speech signal, as explained in the following description.

Practical benefits of the present invention is to provide increased dynamic algebraic codes directory 208, allowing a variety of possible amplitudes of the individual pulses without increasing the complexity of the search code vector.

The principle of coding

The selected speech signal S encode the block using a coding system depicted in Fig. 1, which consists of 11 modules, indicated by the positions 102 to 112. The function and operation of these modules is left unchanged relative to the description of the primary applications for U.S. patent N 07/927528. Therefore, although in the following description, and will be, at least briefly explained the function and operation of each module, it will be focused on the subject that is relatively new description primary application for U.S. patent N 07/927528.

For each block of L samples of the speech signal many encoding parameters by the method of linear dig through analyzer 102 spectrum CMLP. More specifically, the analyzer 102 simulates the spectral characteristics of each block S of L samples.

The input block S of L samples is processed whitening filter 103 having the following transfer function based on the current values of the parameters of the PSC:

< / BR>
where a0= 1, a(z) is the variable of the so-called z - transformation. As shown in Fig. 1, the whitening filter 103 generates the residual vector R.

For calculation and quantization parameters chipboard, namely, the delay T of the fundamental tone and the gain g of the fundamental tone, is used to allocate 104 main tone. The initial state of the selector 104 is also set equal to the value of the signal SF from the selector 110 initial state. Detailed calculation procedure and the quantization parameters of the DSP described in the initial application for U.S. patent N 07/927528 and well-known specialists in the field of technology and advanced will not be considered.

To calculate the characteristic response of the filter HOF intended for use in the subsequent steps, characterizaton 105 response of the filter (Fig. 1) serves the parameters of the PSC and particleboard. Information HOF consists of the following three components, where n = 1, 2, ... L.

f(n): h(n): response 1/A(zy-1) on f(n)

where y is the factor of perception. More generally, h(n) is the impulse response F(z)W(z)/A(z), which is the cascade of the pre-filter F(z), the filter W(z) perceptual weighting and synthesis filter 1/A(z). Note that F(z) and 1/A(z) are the same filters that are used in the decoding device shown in Fig.2.

U(i,j): autocorrelation of h(n) according to the following expression,

< / BR>
The device 106 long-term forecasting is fed past excitation signal (i.e., E+gCkprevious subsets of the data) to form a new component of E, using the appropriate delay T and gain b of the main tone.

The initial state of the filter of perception 107 installed in accordance with the magnitude of SF supplied from the selector 110 initial state. The residual vector R'=R-E with the remote main tone, calculated by the device subtraction 121 (Fig. 1), then served on a perceptual filter 107 for receiving the output of the last filter of the target vector X. As shown in Fig. 1, the IRT parameters are fed to the filter 107 to change its transfer function in connection with these options. Basically, X=R'-P, where P represents the contribution of the long-term p is th error (RMSE), applicable to , in the form of the following matrix entries:

< / BR>
where H is a matrix of smaller triangles Greenhouse m lxl formed from response h(n) as follows. The element h(0) is a diagonal matrix, a h(1), h(2),... h(L-1) are the corresponding lower diagonal.

Phase inverse filtering is performed by the filter 108 shown in Fig. 1. Setting to zero the derivative of the above equality by the gain g leads to the optimal gain as follows:

< / BR>
This value g minimization becomes:

< / BR>
The challenge is to find a specific index k, which is achieved by minimization. Note that since a fixed value, the same index can be found by minimizing the following values:

< / BR>
where D = (XH), and

In the reverse filter 108 is calculated inversely filtered target vector D= (XH). The term "inverse filtering" for this operation stems from the interpretation of (XH) as filtering turned in time X.

To the device shown in Fig.1 above, the primary application for U.S. patent N 07/927528, added only the amplitude selector 112. Function of the amplitude selector 112 is Ogre is the most promising code vectors Akin order to reduce the complexity of the search code vectors. As indicated in the above description, each code vector Akis the signal combination of the amplitudes and positions of the pulses determining L different positions p and containing pulses of zero amplitude, and the pulse nonzero amplitude assigned to the respective positions p = 1, 2,... L combination, and each pulse has a non-zero amplitude assumes the presence of at least one of q different possible amplitudes.

In accordance with Fig.3a, 3b and 3c, the purpose of the amplitude selector 112 is to pre-establish a functional dependence of Spbetween the positions p of the code vector signal and the q possible values of the amplitudes of the pulses. Pre-installed functional dependence of Spis determined in connection with a speech signal before searching in the reference codes. More specifically, prior to the establishment of this functional dependence is in the preliminary assignment, in connection with a speech signal, at least one of q possible amplitudes for each possible position p of the signal (step 301 in Fig. 3a).

To pre-assign one of the q amplitude of the residual vector R' with a remote the main tone. More specifically, the vector In the estimation of the amplitude is calculated by adding (step 301-1 in Fig.3b) is inversely filtered target vector B in normalized form

< / BR>
and the residual vector R' to the remote main tone in normalized form

< / BR>
so as to obtain the vector B evaluation of the amplitude in the form:

< / BR>
where is a fixed constant having a typical value of 1/2 (the value chosen between 0 and 1, depending on the percentage of non-zero pulses of the amplitude used in the algebraic code).

For each position q signal amplitude Sppreviously assigned to this position p is obtained by quantization of the corresponding evaluation of Bpthe amplitude of the vector B. More specifically, for each position p of the signal evaluation Bpthe normalized maximum value of the amplitude vector B quantuum (substep 301-2 in Fig.3b) using the following expression:

< / BR>
where Q(.) function quantization, and

< / BR>
the normalization factor representing the maximum amplitude of the pulses nonzero amplitude.

In the important special case when

q= 2, i.e. we can assume only two values of the amplitudes of the pulses (i.e., Spi= 1), and

the density N is Ki amplitude is simply reduced to the backward filtered target vector D and therefore, Sp= sign(Dp).

The purpose of optimizing the control unit 109 is to choose the best code vector Akof algebraic codes directory. The selection criteria set in the form of a ratio, calculated for each code vector Akand maximized all code vectors (step 303):

< / BR>
where D = (XH), a

Since Akis algebraic code vector having N pulses nonzero amplitudes corresponding to the amplitudes of the Spithe numerator is the square of the size

< / BR>
and the denominator is the energy component, which can be expressed as

< / BR>
where U(pipj- correlation associated with two pulses of unit amplitude, one is in the position of piand the other in the position of the pj. This matrix is calculated in accordance with the above equation in characterizatio 105 response of the filter and include many parameters, defined as the HOF on the block diagram shown in Fig.1.

A quick way to compute the denominator (step 304) provides N nested loops, is shown in Fig.4, in which is used a truncated lowercase entry s(i) and ss(i,j) instead of "Spiis neither. Calculations that contribute to thek2, which are carried out in each cycle, shown in Fig. 4, can be written in separate rows from the outermost loop to the extreme inner loop as follows:

< / BR>
where pi- the position of the i-pulse non-zero amplitude. Note that the presence of N nested loops in Fig.4 gives the possibility to limit the non-zero pulses of the amplitude code vector Akin accordance with codes of permutations of N alternating single pulses.

In the present invention, the complexity of the search is greatly reduced by limiting the subset we seek a code vector Akcode vectors, for which N pulses nonzero amplitudes correspond to functional dependencies pre-installed on the stage 301, shown in Fig.3a. Pre-installed functional dependence enters into force when each of the N pulses nonzero amplitude code vector Akhas an amplitude equal to the amplitude of the pre-assigned position p of the pulse is non-zero amplitude.

The restriction of a subset of the code vectors is carried out by first combining pre ustane N nested loops, it is shown in Fig.4, assuming that all pulses s(i) are fixed, positive and have a unit amplitude (step 303). Thus, even though the non-zero amplitude pulses may be any one of the q possible values in algebraic codes directory, search complexity is reduced to the case of fixed amplitude pulses. More specifically, the matrix U(i,j) element of which is formed by characterization 105 response of the filter, combined with pre-installed functional dependence in accordance with the following equation (step 302):

U'(i,j) =SiSjU(i,j)

where Siis determined by the operation of the amplitude selector 112, namely, Siis the amplitude selected for the line item i, taking into account the quantization of the corresponding estimate of the amplitude.

For this new matrix is computed for each cycle of the fast algorithm can be written in a separate line, from outermost to outermost internal Nikl, as follows:

< / BR>
where px- the position of the x-th pulse is non-zero amplitude signal and where U'(pxpyfunction that depends on the amplitude of the pre-assigned positions pxamong pozicka to reduce the complexity of the search, can be omitted (see Fig. 3c), in particular, but not exclusively, at the inner loop whenever a true inequality:

< / BR>
where Spnthe amplitude of the pre-assigned positions pnDpn- pn- th component of the target vector D, and TD- threshold value associated with back filtered target vector d

The signal E+gCkthe total excitation signal is calculated by the adder 120 (Fig. 1) the signal gCkfrom the control unit 109 and the output signal E from a forecasting device 106. The module selector 110 initial state consisting of a perceptual filter with transfer function 1/A(zy-1), which changes in relation with the parameters of the PSC, is subtracted from the residual signal R, the signal E + gCkthe excitation signal is simply to obtain the final state of the filter SF to use it as the initial condition in the filter 107 and the selector 104 to the main tone.

Many of the four parameters k, g, particle Board and the PSC is converted to the proper format digital channel multiplexer 111, the final procedure coding block S of samples of the speech signal.

Although the present invention has been described above with syncategorematic within the scope of the accompanying claims, without changing the essence of the invention.

1. The way search reference code consisting of a set of combinations of amplitudes and positions of the pulses to encode the audio signal, and each combination of amplitudes and positions of the pulses determines the L different positions and includes pulses of zero amplitude, and the pulse nonzero amplitude assigned to the respective positions p = 1, 2, ..., L the combination, and each pulse has a non-zero amplitude assumes the presence of one of the q possible amplitudes, characterized in that what is pre-selected from reference codes a subset of combinations of amplitudes and positions of the pulses in connection with a sound signal and carry out the search only in the subset of combinations of amplitudes and positions of the pulses to encode the audio signal, thereby reducing the complexity of the search, because the search carried out in only one subset combinations of the amplitudes and positions of the pulses of the reference codes, and the step of pre-selecting includes preliminary determination, in connection with a sound signal, the functional dependence of Sppre-set positions p = 1, 2, ... L the actual amplitude of the above q possible AB, having pulses of a non-zero amplitude, which correspond to the pre-installed functional dependency.

2. The method according to p. 1, wherein the step of establishing a functional dependency includes pre-assignment by functional dependency Spone of the possible amplitudes as the actual amplitude of each position p, and pre-installed functional dependence is effective, boiler each pulse nonzero amplitude combinations of amplitudes and positions of the pulses has an amplitude equal to the amplitude of the pre-assigned functional dependence of Spthe position p of the pulse is non-zero amplitude.

3. The method according to p. 2, wherein the step of assigning one of the q possible amplitudes of each position p lies in the fact that processes the audio signal from the receiving back the filtered target vector D and the residual signal R' to the remote main tone, compute the vector B evaluation of the amplitude is inversely filtered target vector D and the residual signal R' to the remote main tone and for each of the mentioned products p quantuum assessment Bpample p. 3, characterized in that the step of calculating the vector B of amplitude evaluation includes adding back the filtered vector D in normalized form

< / BR>
residual signal R' to the remote main tone in normalized form

< / BR>
to obtain the vector B evaluation of the amplitude in the form

< / BR>
where is a fixed constant.

5. The method according to p. 4, characterized in that it is a fixed constant having a value between 0 and 1.

6. The method according to p. 3, characterized in that for each of these positions p phase quantization includes quantization assessment Bpthe normalized maximum value of the amplitude of the vector B using the following expression:

< / BR>
where the denominator is

< / BR>
is a normalization factor representing the maximum amplitude of the pulses nonzero amplitude.

7. The method according to p. 1, characterized in that each of the mentioned combinations of pulses contains the number N of pulses of the non-zero amplitude, and the above-mentioned method further includes limiting positions p of non-zero pulses of amplitude in accordance at least with the same code permutation N of intermittent single pulses.

8. The method according to p. 3, Otley is oulevay amplitude, and the step of searching includes the maximization of the given fraction, the denominator of the k2which is computed by means of N nested loops in accordance with the following relationship:

< / BR>
where the calculation for each cycle recorded in a separate line from the outermost loop to the extreme inner loop of the N nested loops;

Pn- the position of the n-th pulse is non-zero amplitude combination;

U'(PxPy) function that depends on the amplitude of the Spxpre-assigned positions Pxfrom the number of positions p, and the amplitude of Spypre-assigned positions Pyfrom the number of positions p.

9. The method according to p. 8, characterized in that the step of maximizing the above given fraction comprises skipping at least at the inner loop of the N nested loops whenever we have the following inequality:

< / BR>
where Spnthe amplitude of the pre-assigned positions Pn;

Dpn- Pn-I component of the target vector D;

TD- threshold value corresponding to the back of filtered target vector d

10. A device for implementing directory search code consisting of a set of combinations of amplitudes and positions delaet L different positions and includes pulses of zero amplitude, and the non-zero pulses of amplitude, assigned to the respective positions p = 1, 2, ..., L the combination, and each pulse has a non-zero amplitude assumes the existence of one possible amplitudes containing the tool pre-selection of these reference codes of subset combinations of the amplitudes and positions of the pulses in connection with a sound signal and a search tool only in the above-mentioned subset combinations of the amplitudes and positions of the pulses to encode the audio signal, to reduce the complexity of the search by searching only in one subset combinations of the amplitudes and positions of the pulses of the reference codes, the tool pre-selection includes a tool pre-setting, in connection with a sound signal, the functional dependence of Spfor pre-assigning positions p = 1, 2, ..., L the actual amplitudes of these q possible amplitudes, and the search tool provides a means of limiting the search to those combinations of amplitudes and positions of the pulses of the above-mentioned reference codes with non-zero pulses of amplitude, which correspond to the pre-installed functional dependency.

11. The device according to p. 10, great for alternova assignments through the functional dependence of Spone of the q possible amplitudes as the actual amplitude of each position p, and pre-installed functional dependence enters into force when each pulse is non-zero amplitude combinations of amplitudes and positions of the pulses has an amplitude equal to the amplitude of the pre-assigned functional dependence of Spthe position p of the pulse is non-zero amplitude.

12. The device according to p. 11, characterized in that the means of pre-assigning one of the q possible amplitudes of each position contains a means of processing the audio signal with getting back the filtered target vector D and the residual signal R' to the remote main tone, the tool will calculate the vector B evaluation of the amplitude is inversely filtered target vector D and the residual signal R' to the remote main tone and a means of quantization for each of these positions p, evaluation of Bpthe amplitude of the above-mentioned vector B with obtaining the amplitude selected for the mentioned position p.

13. The device according to p. 12, characterized in that the said means of calculation of the vector B of amplitude evaluation provides a means of adding back tfilter in normalized form

< / BR>
to obtain the vector B evaluation of the amplitude in the form

< / BR>
where is a fixed constant.

14. The device according to p. 13, characterized in that it is a fixed constant having a value between 0 and 1.

15. The device according to p. 12, characterized in that the said quantization means includes quantization means, for each of these positions p, evaluation of Bpnormalized maximum amplitude values of the vector B using the following expression:

< / BR>
where the denominator is

< / BR>
is a normalization factor representing the maximum amplitude of the pulses nonzero amplitude.

16. The device according to p. 10, characterized in that each of the mentioned combinations of amplitudes and positions of the pulses contains the number N of pulses of the non-zero amplitude, while the above device further comprises a means of limiting the positions p of non-zero pulses of amplitude in accordance at least with the same code permutation N of intermittent single pulses.

17. The device according to p. 12, characterized in that each of the mentioned combinations of amplitudes and positions of the pulses contains the number N of pulses of the non-zero amplitude, and the medium is I mentioned the denominator of the k2by N nested loops in accordance with the following relationship:

< / BR>
where the calculation in each cycle recorded in a separate line from the outermost loop to the extreme inner loop of the N nested loops, where Pp- the position of the n-th pulse is non-zero amplitude combinations and where U'(PxPy) function that depends on the amplitude of the SPxpre-assigned positions Pxfrom the number of positions p, and the amplitude of SPypre-assigned positions Pyfrom the number of positions p.

18. The device under item 17, characterized in that the said means of calculation of the denominator of the k2contains the tool passes, at least at the inner loop of the N nested loops whenever we have the following inequality:

< / BR>
where SPnthe amplitude of the pre-assigned position of the Pn;

Dpn- Pn-th component of the target vector D;

TD- threshold value related to a backward filtered target vector d

19. The cellular communication system for servicing a large area, divided into a number of cells containing mobile transceiver units, cellular base stations, setpagestep two-way radio communications between each mobile unit, located in one of the cells, and the cellular base station is referred to cells that contain and in the mobile unit and the cellular base station transmitter includes means encoded speech signal and means for transmitting the encoded speech signal, and the receiver includes means receiving the transmitted encoded speech signal and means for decoding the received encoded speech signal, characterized in that the means of encoding the speech signal contains a device for searching the reference code consists of many combinations of amplitudes and positions of the pulses for encoding the speech signal, moreover, each combination of amplitudes and positions of the pulses determines the L different positions and includes pulses of zero amplitude, and the pulse nonzero amplitude assigned to the respective positions p = 1, 2, ..., L the combination, and each pulse has a non-zero amplitude assumes the presence of one of the q possible amplitudes, the device for searching includes means pre-selection of reference codes a subset of combinations of amplitudes and positions of the pulses in connection with a speech signal and means search only mentioned the falsity of the search by searching only in one subset of combinations of amplitudes and positions in the reference code, the tool pre-selection includes a tool pre-setting, in connection with a sound signal, the functional dependence of Sppre-set positions p = 1, 2, ..., L the actual amplitude of the above q possible amplitudes, and thus the search tool provides a means of limiting the search to those combinations of amplitudes and positions of the pulses of the reference codes, pulses having nonzero amplitudes, which correspond to the preset function.

20. The system under item 19, characterized in that the tool prior establishment of a functional dependency holds the tool prior appropriation, by the functional dependence of Spone of the q possible amplitudes as the correct amplitude of each position p, and pre-installed functional dependence enters into force when each pulse is non-zero amplitude combinations of amplitudes and positions of the pulses has an amplitude equal to the amplitude of the pre-assigned functional dependence of Spthe position p of the pulse is non-zero amplitude.

21. System on p. 20, characterized in that the means of pre-assigning one is th filtered target vector D and the residual signal R' to the remote main tone, the means of calculating the vector B evaluation of the amplitude is inversely filtered target vector D and the residual signal R' to the remote main tone and a means of quantization for each of these positions P, evaluation of Bpthe amplitude of the above-mentioned vector B with obtaining the amplitude selected for the mentioned position p.

22. System on p. 21, characterized in that the said means of calculation of the vector B of amplitude evaluation provides a means of adding back the filtered target vector D in normalized form

< / BR>
residual signal R' to the remote main tone in normalized form

< / BR>
to obtain the vector B evaluation of the amplitude in the form

< / BR>
where is a fixed constant.

23. The system according to p. 22, characterized in that it is a fixed constant having a value between 0 and 1.

24. System p. 23, characterized in that the said quantization means includes quantization means, for each of these positions p, evaluation of Bpnormalized maximum value of the amplitude of the vector B using the following expression

< / BR>
where the denominator is

< / BR>
is a normalization factor. represents the maximum ampli is combinaci amplitudes and positions of the pulses contains the number N of pulses of the non-zero amplitude, as mentioned, the device further comprises a means of limiting the positions p of non-zero pulses of the amplitude in accordance with at least one of the codes permutation N of intermittent single pulses.

26. The system according to p. 22, characterized in that each of the mentioned combinations of amplitudes and positions of the pulses contains the number N of pulses of the non-zero amplitude, while the search tool provides a means of maximising the given fraction with a denominator of k2and the means of calculating the above denominator k2by N nested loops in accordance with the following relationship:

< / BR>
where the calculation in each cycle recorded in a separate line from the outermost loop to the extreme inner loop of the N nested loops, where Pn- the position of the n-th pulse is non-zero amplitude combinations, and where U'(PxPy) function that depends on the amplitude of the Spxpre-assigned positions Pxfrom the number of positions p, and the amplitude of Spypre-assigned positions Pyfrom the number of positions p.

27.The system under item 26, characterized in that the said means of calculation of the denominator of the k2contains a means passes at least extreme in the UB>pnthe amplitude of the pre-assigned position of the Pn;

DPn- Pn-th component of the target vector D;

TD- threshold value related to a backward filtered target vector d

Priority points:

06.02.95 - p. 1, 10;

28.07.95 - p. 19.

 

Same patents:

The invention relates to the field of modern mobile radio and can be used when creating user terminals (UTS) of such systems

The invention relates to the field of technology encrypted communication

The invention relates to a mobile communication system, and more particularly to a method and apparatus for managing a service channel of a base station to improve the efficiency and stability of the system characteristics, multiple access, code-division multiplexing

The invention relates to systems in the mobile-cellular telephone, and more particularly to methods and devices designed to ensure the use of various designs of subscriber telephone sets with regard to compromise the demands for efficiency and effectiveness

The invention relates to communication technology

The invention relates to communication systems and can be used to provide voice and data services for the population of terminals

The invention relates to the field of processing of speech signals, more specifically to the creation of new and improved methods and devices for compressing speech signals, in which the compression ratio is dynamically changed to minimize the impact of compression on the quality of the reproduced speech

FIELD: radio communications.

SUBSTANCE: proposed method intended for single-ended radio communications between mobile objects whose routes have common initial center involves radio communications with aid of low-power intermediate transceiving stations equipped with non-directional antennas and dropped from mobile object, these intermediate transceiving drop stations being produced in advance on mentioned mobile objects and destroyed upon completion of radio communications. Proposed radio communication system is characterized in reduced space requirement which enhances its effectiveness in joint functioning of several radio communication systems.

EFFECT: reduced mass and size of transceiver stations, enhanced noise immunity and electromagnetic safety of personnel.

1 cl, 7 dwg, 1 tbl

FIELD: radio communications.

SUBSTANCE: proposed method intended for data transfer from mobile object to stationary one residing at initial center of common mobile-object route using electronic means disposed on stationary and mobile objects involves radio communications with aid of low-power intermediate transceiving stations equipped with non-directional antennas and dropped from mobile object, these intermediate transceiving drop stations being produced in advance on mobile object. Proposed radio communication system is characterized in reduced space requirement which enhanced its effectiveness in joint functioning with several other radio communication systems.

EFFECT: reduced mass and size of transceiver stations, enhanced noise immunity and electromagnetic safety of personnel.

2 cl, 6 dwg

FIELD: radio communications.

SUBSTANCE: proposed method intended for data transfer to mobile object from stationary one residing at initial center of mobile-object route using electronic means disposed on stationary and mobile objects involves radio communications with aid of low-power intermediate transceiving stations equipped with non-directional antennas and dropped from mobile object, these intermediate transceiving drop stations being produced in advance on mobile object. Proposed radio communication system is characterized in reduced space requirement which enhances its effectiveness in joint functioning with several other radio communication systems.

EFFECT: reduced mass and size of transceiver stations, enhanced noise immunity and electromagnetic safety of personnel.

2 cl, 6 dwg, 1 tbl

FIELD: radio communications.

SUBSTANCE: proposed method for single-ended radio communications between mobile objects whose routes have common initial center involves use of low-power intermediate transceiving stations equipped with non-directional antennas and dropped from mobile objects. Proposed radio communication system is characterized in reduced space requirement and, consequently, in enhanced effectiveness when operating simultaneously with several other radio communication systems.

EFFECT: reduced mass and size, enhanced noise immunity and electromagnetic safety for attending personnel.

2 cl, 7 dwg, 1 tbl

FIELD: radio communications.

SUBSTANCE: proposed method intended for data transfer to mobile objects from stationary one residing at initial center of common mobile-objects route using electronic means disposed on stationary and mobile objects involves radio communications with aid of low-power intermediate transceiving stations equipped with non-directional antennas and dropped from first mobile object. Proposed radio communication system is characterized in reduced space requirement which enhances its effectiveness in simultaneous functioning of several radio communication systems.

EFFECT: reduced mass and size of transceiver stations, enhanced noise immunity and electromagnetic safety of personnel.

2 cl, 7 dwg, 1 tbl

Up!