Vector joint encoding/decoding method and vector joint encoder/decoder

Abstract

A vector joint encoding/decoding method and a vector joint encoder/decoder are provided, more than two vectors are jointly encoded, and an encoding index of at least one vector is split and then combined between different vectors, so that encoding idle spaces of different vectors can be recombined, thereby facilitating saving of encoding bits, and because an encoding index of a vector is split and then shorter split indexes are recombined, thereby facilitating reduction of requirements for the bit width of operating parts in encoding/decoding calculation.

Description

FIELD OF THE INVENTION

The present invention relates to a vector joint encoding/decoding method and a vector joint encoder/decoder.

BACKGROUND OF THE INVENTION

When a signal is processed, generally the signal may be divided into multiple vectors. For example, referring to FIG. 1, the signal sampling is classified, several sampling values are classified as a group, which may be referred to as a vector, and the number of this group of sampling values is referred to as the dimension of the vector. In a voice encoder based on a code excited linear prediction (CELP) model, generally several pulses are used to denote quasi-white noise excitation. In order to reduce the complexity of the encoder, generally an input signal is divided into several sub-frames to be processed, and during processing a fixed code book, generally a signal of each sub-frame is further divided into several tracks. Based on the foregoing understanding about the vector, it may be regarded that sampling points on a track may form a multi-dimensional vector. For example, for a track having sixteen positions, it is assumed that a signal sampling case on a track is described using a pulse, and a value may be represented as [−1, −5, 0, −5, 12, 0, . . . , 0]. From the perspective of the pulse, it is denoted that this track has four non-zero pulses, that is, a first position has a pulse with amplitude being 1, and a symbol being “−”; a second position has a pulse with amplitude being 5, and a symbol being “−”; a third position has a pulse with amplitude being zero, and the rest may be deduced by analogy. From the perspective of the vector, a 16-dimensional vector has four non-zero components.

It is well-known that, in the voice encoding field, a voice encoder based on the CELP model is applied very widely, such as G.729, Global System for Mobile Communications (GSM) and G.723.1, which are already widely applied in various digital signal processors (DSP), embedded devices, mobile phones and personal computers (PC). Compared with other types of voice encoders, the voice encoder/decoder based on the CELP model can achieve good voice quality in a case of an extremely low code rate, and still have excellent performance in a case of a high code rate.

In the vector encoding technology, usually an algebraic code book is used to perform quantization encoding on a residual signal after adaptive filtering. After information of a position and a symbol of an optimal algebraic code book pulse on a track is obtained through searching, a corresponding index value is obtained through encoding calculation, so that a decoding end can reestablish a pulse sequence according to the index value. Under the premise of ensuring lossless reestablishment, to reduce bits, which are needed for an encoding index value, as much as possible is one of main objectives of the research and development of an algebraic code book pulse encoding method.

The following takes an example of a preferable encoding method in voice encoding an adaptive multi-rate wideband (AMR_WB+) encoding method, to introduce a specific encoding method adopted for an existing algebraic code book pulse. According to the difference between encoding code rates, 1 to N pulses may be encoded on each track, and it is assumed that each track has M=2^m positions, and procedures for encoding one to six pulses on each track in AMR_WB+ are respectively described as follows.

(1) One Pulse is Encoded on Each Track.

Because each track has 2^m positions, a position index of a pulse on each track needs to be encoded with m bits, and a symbol index of the pulse needs to be encoded with 1 bit. An index value for encoding one pulse with a symbol is I_1p(m)=p+s×2^m, where p∈[0, 2^m−1] is a position index of the pulse; s is a symbol index of the pulse, and when a symbol of the pulse is positive, s is set to 0, and when the symbol of the pulse is negative, s is set to 1; I_1p∈[0, 2^m+1−1].

The number of bits needed for encoding one pulse on each track is m+1.

(2) Two Pulses are Encoded on Each Track

According to a result of (1), m+1 bits are needed for encoding one pulse on each track, and m bits are needed for encoding a position index of another pulse; because the pulse order is not particularly needed, a symbol of another pulse may be denoted through a size relationship obtained by arranging position indexes of the pulses. An index value for encoding two pulses is I_2p(m)=p1+I_1p0×2^m=p1+p0×2^m+s×2^2m, where p0 and p1∈[0, 2^m−1] are position indexes of two pulses, respectively; s is a symbol index of pulse p0; a specific denotation rule for a symbol of pulse p1 is p0<p1 denotes that symbols of the two pulses are the same, and p0>p1 denotes that the symbols of the two pulses are opposite; I_2p∈[0, 2^2m+1−1].

The number of bits needed for encoding two pulses on each track is 2m+1.

(3) Three Pulses are Encoded on Each Track.

Each track is divided into two sections, Section A and Section B, and each section includes 2^m−1 positions. A certain section at least includes two pulses, and according to a result of (2), 2×(m−1)+1=2m−1 bits are needed for encoding the section; another pulse searches on the entire track, and according to a result of (1), m+1 bits are needed; furthermore, one bit is further needed to indicate a section including two pulses. An index value for encoding three pulses is I_3p(m)=I_2p(m−1)+k×2^2m−1+I_1p(m)×2^2m, where k is an index of Section; I_3p∈[0, 2^3m+1−1].

The number of bits needed for encoding three pulses on each track is 3m+1.

(4) Four Pulses are Encoded on Each Track.

Each track is divided into two sections, Section A and Section B, and each section includes 2^m−1 positions. Cases of combinations of the numbers of pulses included in each section are shown in the following table:

	Number of pulses in	Number of pulses in	Number of
Type	Section A	Section B	bits needed
0	0	4	4 m−3
1	1	3	4 m−2
2	2	2	4 m−2
3	3	1	4 m−2
4	4	0	4 m−3

In the above table, the basis of the number of bits needed, which correspond to each type is that, for type 0 and type 4, a method similar to (4) is adopted in a section having four pulses, but the number of pulses for performing whole searching is two, it is equivalent to I_2p(m−2)+k×2^2m-3+I_2p(m−1)×2^2m−2; for type 1, it is equivalent to I_1p(m−1)+I_3p(m−1)×2^m; for type 2, it is equivalent to I_2p(m−1)+I_2p(m−1)×2^2m−1; and for type 3, it is equivalent to I_3p(m−1)+I_1p(m−1)×2^3m−2.

Type 0 and type 4 are regarded as a kind of possible case, type 1 to type 3 each are used as a case, and totally there are four cases; therefore, two bits are needed to denote a corresponding case, thus type 1 to type 3 all need 4m−2+2=4m bits; moreover, for the case including type 0 and type 4, one bit is further needed for distinguishing, and therefore, type 0 and type 4 need 4m−3+2+1=4m bits.

The number of bits needed for encoding four pulses on each track is 4m.

(5) Five Pulses are Encoded on Each Track.

Each track is divided into two sections, Section A and Section B, and each section includes 2^m−1 positions. A certain section at least includes three pulses, and according to a result of (3), 3×(m−1)+1=3m−2 bits are needed for encoding the section; the rest two pulses search on the entire track, and according to a result of (2), 2m+1 bits are needed; furthermore, one bit is further needed to indicate a section including three pulses. An index value for encoding five pulses is I_5p(m)=I_3p(m−1)+k×2^3m-2+I_1p(m)×2^3m-1.

The number of bits needed for encoding five pulses on each track is 5m.

(6) Six Pulses are Encoded on Each Track.

Each track is divided into two sections, Section A and Section B, and each section includes 2^m−1 positions. Cases of combinations of the numbers of pulses included in each section are shown in the following table:

	Number of pulses in	Number of pulses in	Number of
Type	Section A	Section B	bits needed
0	0	6	6 m−5
1	1	5	6 m−5
2	2	4	6 m−5
3	3	3	6 m−4
4	4	2	6 m−5
5	5	1	6 m−5
6	6	0	6 m−5

In the above table, the basis of the number of bits needed, which correspond to each type, may be deduced following (4), and is not repeated again.

Type 0 and type 6, type 1 and type 5, type 2 and type 4 each are regarded as a possible case, type 3 is used as a case independently, and totally there are four cases; therefore, two bits are needed to denote a corresponding case, and type 3 needs 6m−4+2=6m−2 bits; for those cases including a combined type, one bit is further needed for distinguishing, and therefore, other types except for type 3 require 6m−5+2+1=6m−2 bits.

The number of bits needed for encoding six pulses on each track is 6m−2.

In the process of proposing the present invention, it is found that, In the algebraic pulse encoding method provided by AMR_WB+, an encoding logic similar to recursion is adopted to split a case of encoding a large number of pulses into several cases of encoding a small number of pulses for processing, in which the calculation complexity is large, and meanwhile, with the increase of the number of pulses encoded on a track, the redundancy of an encoding index is gradually accumulated, thereby easily causing waste of encoding bits.

SUMMARY OF THE INVENTION

An embodiment of the present invention provides a vector joint encoding method capable of helping to save encoding bits.

A vector joint encoding method includes calculating an encoding index (Ind_t) of each vector, where a subscript t denotes a t^th vector, t∈[0, T−1], and T is the number of vectors and is an integer greater than or equal to 2; splitting at least one Ind_t at least once into at least two sections, where the splitting at least once is equivalent to splitting the Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor (SLF_t), the SLF_t is a positive integer, the Ind_t0 denotes a serial number of an interval to which the Ind_t belongs, the Ind_t1 denotes a serial number of the Ind_t in the interval to which the Ind_t belongs, the length of the interval is not greater than the SLF_t, and Ind_t≤Ind_t0×SLF_t+Ind_t; combine a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index Ind_SLF; and encode according to the combined index and other uncombined split indexes.

An embodiment of the present invention further provides a multi-stage cascade encoding method based on the foregoing vector joint encoding method.

Embodiments of the present invention also provide a decoding method and an encoder/decoder corresponding to the foregoing encoding method.

One aspect of present disclosure discloses a vector joint encoding method, comprising calculating an encoding index Ind_t of each vector, wherein a subscript t denotes a t^th vector, t∈[0, T−1], and T is the number of vectors and is an integer greater than or equal to 2; splitting at least one Ind_t at least once into at least two sections, wherein the splitting at least once is equivalent to splitting the Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t, the SLF_t is a positive integer, the Ind_t0 denotes a serial number of an interval to which the Ind_t belongs, the Ind_t denotes a serial number of the Ind_t in the interval to which the Ind_t belongs, the length of the interval is not greater than the SLF_t, and Ind_t<Ind_t0×SLF_t+Ind_t1; combining a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index Ind_SLF; and performing encoding according to the combined index and other uncombined split indexes.

Where the vector is represented as pulse distribution on a track, and the encoding index is an index used for indicating the pulse distribution on the track.

Wherein the splitting the Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t is Ind_t0=Int(Ind_t/SLF_t), wherein Int( ) denotes rounding down to an integer, and Ind_t1=Ind_t% SLF_t, wherein % denotes taking a remainder.

Wherein SLF_t=2^(K_t), wherein K_t is a positive integer, and the combining a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index Ind_SLF is, for a vector t providing a split index to participate in combination, selecting the Ind_t0 to participate in combination; or, SLF_t=Int(Ind_t,max/2^(K_t)), wherein Ind_t,max denotes a maximal value of the Ind_t, and the combining a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index (Ind_SLF) is, for the vector t providing a split index to participate in combination, selecting the Ind_t1 to participate in combination.

Wherein K_t=K_t,max−2, or K_t=K_t,max−3, or K_t=K_t,max−4, wherein the K_t,max is a length of a bit sequence of the Ind_t,max.

Wherein the combining a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index (Ind_SLF) is, for a vector t providing a split index to participate in combination, selecting a split index capable of embodying space occupancy characteristics of the Ind_t to participate in combination, wherein being capable of embodying space occupancy characteristics of the Ind_t refers to that, compared with other split indexes, an occupancy rate of a value range of a selected split index for an encoding space is the closest to an occupancy rate of a value range of the Ind_t for the encoding space.

Wherein the splitting the Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t is selecting a value of bits of the Ind_t as the Ind_t0, the bits being located at a set position, and selecting a value of bits at a remaining position as the Ind_t1, wherein SLF_t=2^(K_t), the number of the bits at the set position is K_t0, the number of the bits at the remaining position is K_t, K_t0+K_t=K_t,max, K_t,max is a length of a bit sequence of Ind_t,max, and the Ind_t,max denotes a maximal value of the Ind_t.

Wherein combining split indexes from at least two vectors to generate a combined index (Ind_SLF) is, for a vector t providing a split index to participate in combination, selecting, in a bit sequence with a length being K_t,max, a split index of the Ind_t to participate in combination, wherein the split index at least comprises a value of the highest two bits.

Wherein the performing encoding according to the combined index and other uncombined split indexes is splitting the combined index into T1 recombined indexes Ind_t0′ according to a set value range, wherein T1 is less than or equal to the number of vectors generating the combined index, a value range of at least one Ind_t0′ is greater than a value range of the split index of a corresponding vector t, wherein the split index participates in combination, and a value range of at least one Ind_t0′ is less than the value range of the split index of the corresponding vector t, wherein the split index participates in combination; and respectively combining each recombined index and an uncombined split index of a corresponding vector and then performing encoding, and if a vector without being allocated a recombined index exists, encoding an uncombined split index of the vector.

Wherein the splitting the combined index into T1 recombined indexes Ind_t0′ according to a set value range is splitting a total length K_SLF of the bit sequence of the combined index into T1 sections according to a set length, wherein a value of each section corresponds to one Ind_t0′, the K_SLF is the length of the bit sequence of an Ind_SLF,max, and the Ind_SLF,max denotes a maximal value of the Ind_SLF.

Wherein the splitting the Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t is, in the bit sequence with the length being K_t,max, selecting a value of K_t0 bits, starting from the highest bit, of the Ind_t, as the Ind_t0, and selecting a value of remaining bits as the Ind_t1, wherein SLF_t=2^(K_t), K_t0+K_t=K_t,max, the K_t,max is the length of the bit sequence of the Ind_t,max, and the Ind_t,max denotes a maximal value of the Ind_t; and the combining a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index Ind_SLF is, for a vector t needs to provide a split index, selecting the Ind_t0 to participate in combination; and the splitting the total length K_SLF of the bit sequence of the combined index into T1 sections according to a set length is splitting the K_SLF according to a K_t0 value used by the vector t generating the combined index, wherein the number of bits split by each Ind_t0′ correspondingly is less than or equal to the K_t0 value used by the corresponding vector t.

Wherein the performing encoding according to the combined index and other uncombined split indexes is comparing the combined index Ind_SLF and adjusting a threshold value THR, wherein THR≤2^(K_SLF)−Ind_SLF,max, the K_SLF is the length of the bit sequence of the Ind_SLF,max, and the Ind_SLF,max denotes a maximal value of the Ind_SLF; if the Ind_SLF is less than the THR, encoding the Ind_SLF using a first number of encoding bits; otherwise, encoding the Ind_SLF added with an offset value THR₀ using a second number of encoding bits, wherein THR≤THR₀≤2^(K_SLF)−Ind_SLF,max, the first number is less than the second number, the second number is less than or equal to the K_SLF, and the first number and the second number are both positive integers; and encoding other uncombined split indexes.

One aspect of the present disclosure discloses a vector joint pulse encoding method, comprising grouping vectors participating in joint encoding, wherein each group at least comprises two vectors; in each group, calculating an encoding index of each vector; splitting at least two encoding indexes; splitting each encoding index at least once into at least two sections, wherein splitting at least once is equivalent to splitting an encoding index into two one-stage split indexes according to a set one-stage factor, one one-stage split index denotes serial numbers of several intervals with a length not greater than a set value, the other one-stage split index denotes a serial number of the encoding index in an interval to which the encoding index belongs; and combining one-stage split indexes from at least two vectors to generate a one-stage combined index; starting from m=2, repeating the following operation of generating an m-stage combined index until m=M; splitting at least two (m−1)-stage combined indexes, wherein m∈[2, M], M is an integer greater than or equal to 2; splitting each (m−1)-stage combined index at least once into at least two sections, wherein splitting at least once is equivalent to splitting an (m−1)-stage combined index into two m-stage split indexes according to a set m-stage factor; and combining m-stage split indexes from at least two (m−1)-stage combined indexes to generate an m-stage combined index; and encoding according to the M-stage combined index and other uncombined one-stage to (M−1)-stage split indexes.

Wherein the one-stage split index used for generating the one-stage combined index is a value intercepted from a set bit of a corresponding encoding index starting from the highest bit; and the m-stage split index used for generating the m-stage combined index is a value intercepted from a set bit of a corresponding (m−1)-stage combined index starting from the highest bit.

One aspect of the present disclosure discloses a vector joint pulse decoding method, comprising acquiring a joint code, and acquiring, from the joint code, a combined index and an uncombined split index corresponding to a vector; splitting the combined index into split indexes corresponding to the vector, or splitting the combined index into a split index and an encoding index that correspond to the vector; for each vector participating in splitting the encoding index, splicing, according to a split manner of an encoding end, a split index not participating in combination and a split index participating in combination of the vector to generate the encoding index of the vector; and reestablishing the vector according to the encoding index of the vector for each vector.

Wherein the acquiring, from the joint code, a combined index and an uncombined split index corresponding to each vector is extracting, from the joint code, a code corresponding to each vector; splitting a recombined index and an uncombined split index from the code of each vector; and if a code comprising no recombined index exists, directly obtaining an uncombined split index corresponding to a corresponding vector; and splicing all recombined indexes into a combined index according to the split manner of the encoding end.

One aspect of the present disclosure discloses a vector joint pulse decoding method, comprising acquiring a joint code, and acquiring, from the joint code, an M-stage combined index, an uncombined m-stage split index corresponding to each (m−1)-stage combined index, and an uncombined one-stage split index corresponding to each vector, wherein M is an integer greater than or equal to 2, and m∈[2, M]; starting from m=M, repeating the following operation of generating an (m−1)-stage combined index until m=2; splitting each m-stage combined index into an m-stage split index corresponding to each (m−1)-stage combined index for generating the m-stage combined index; and for each (m−1)-stage combined index, splicing, according to a split manner of an encoding end, an m-stage split index not participating in combination and an m-stage split index participating in combination of the (m−1)-stage combined index to generate the (m−1)-stage combined index according to a split manner of an encoding end; splitting each one-stage combined index into a one-stage split index corresponding to each vector in a vector group for generating the one-stage combined index; and for each vector in each vector group, splicing, according to the split manner of the encoding end, a one-stage split index not participating in combination and a one-stage split index participating in combination of the vector to generate an encoding index; and reestablishing the vector according to the encoding index for each vector in each vector group.

One aspect of the present disclosure discloses a vector joint pulse encoder, comprising an encoding index calculation unit configured to calculate an encoding index (Ind_t) of each vector, wherein a subscript t denotes a t^th vector, t∈[0, T−1], and T is an integer greater than or equal to 2; a vector index splitting unit configured to split at least one Ind_t at least once into at least two sections, wherein splitting at least once is equivalent to splitting the Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t, the SLF_t is a positive integer, the Ind_t0 denotes a serial number of an interval to which the Ind_t belongs, the Ind_t1 denotes a serial number of the Ind_t in the interval to which the Ind_t belongs, a length of the interval is not greater than the SLF_t, and Ind_t≤Ind_t0×SLF_t+Ind_t1; an index combination unit configured to combine a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index (Ind_SLF); and an encoding unit configured to perform encoding according to the combined index and other uncombined split indexes.

Wherein the encoding unit comprises a recombining sub-unit configured to split the combined index into T1 recombined indexes Ind_t0′ according to a set value range, wherein T1 is less than or equal to the number of vectors generating the combined index, a value range of at least one Ind_t0′ is greater than a value range of the split index, participating in combination, of a corresponding vector t, and a value range of at least one Ind_t0′ is less than the value range of the split index, participating in combination, of the corresponding vector t; and an encoding sub-unit configured to combine each recombined index and an uncombined split index of a corresponding vector respectively and then perform encoding, and if a vector without being allocated a recombined index exists, encode an uncombined split index of the vector.

One aspect of the present disclosure discloses a vector joint pulse encoder, comprising a vector index calculation unit configured to group vectors participating in joint encoding, wherein each group at least comprises two vectors; in each group, calculate an encoding index of each vector; split at least two encoding indexes; split each encoding index at least once into at least two sections, wherein splitting at least once is equivalent to splitting an encoding index into two one-stage split indexes according to a set one-stage factor, one one-stage split index denotes serial numbers of several intervals with a length not greater than a set value, the other one-stage split index denotes a serial number of the encoding index in an interval to which the encoding index belongs; and combine one-stage split indexes from at least two vectors to generate a one-stage combined index; a cascade calculation unit configured to, starting from m=2, repeat the following operation of generating an m-stage combined index until m=M; split at least two (m−1)-stage combined indexes, wherein m∈[2, M], M is an integer greater than or equal to 2; split each (m−1)-stage combined index at least once into at least two sections, wherein splitting at least once is equivalent to splitting an (m−1)-stage combined index into two m-stage split indexes according to a set m-stage factor; and combine m-stage split indexes from at least two (m−1)-stage combined indexes to generate an m-stage combined index; and an encoding unit configured to perform encode according to the M-stage combined index and other uncombined one-stage to (M−1)-stage split indexes.

One aspect of the present disclosure discloses a vector joint pulse decoder, comprising a decoding unit configured to acquire a joint code, and acquire, from the joint code, a combined index and an uncombined split index corresponding to a vector; an index splitting unit configured to split the combined index into split indexes corresponding to the vector, or split the combined index into a split index and an encoding index that correspond to the vector; a vector index reestablishing unit configured to, for each vector participating in splitting the encoding index, splice, according to a split manner of an encoding end, a split index not participating in combination and a split index participating in combination of the vector to generate the encoding index of the vector; and a vector reestablishing unit configured to reestablish the vector according to the encoding index of the vector for each vector.

Wherein the decoding unit comprises a decoding sub-unit configured to extract, from the joint code, a code corresponding to each vector; split a recombined index and an uncombined split index from the code of each vector; and if a code comprising no recombined index exists, directly obtain an uncombined split index corresponding to a corresponding vector; and a splicing sub-unit configured to splice all recombined indexes into a combined index according to the split manner of the encoding end.

One aspect of the present disclosure discloses a vector joint pulse decoder, comprising a decoding unit configured to acquire a joint code, and acquire, from the joint code, an M-stage combined index, an uncombined m-stage split index corresponding to each (m−1)-stage combined index, and an uncombined one-stage split index corresponding to each vector, wherein M is an integer greater than or equal to 2, and m∈[2, M]; a cascade recovering unit configured to, starting from m=M, repeat the following operation of generating an (m−1)-stage combined index until m=2; split each m-stage combined index into an m-stage split index corresponding to each (m−1)-stage combined index for generating the m-stage combined index; and for each (m−1)-stage combined index, splice, according to a split manner of an encoding end, an m-stage split index not participating in combination and an m-stage split index participating in combination of the (m−1)-stage combined index to generate the (m−1)-stage combined index; a vector index reestablishing unit configured to split each one-stage combined index into a one-stage split index corresponding to each vector in a vector group for generating the one-stage combined index; and for each vector in each vector group, splice, according to the split manner of the encoding end, a one-stage split index not participating in combination and a one-stage split index participating in combination of the vector to generate an encoding index; and a vector reestablishing unit configured to reestablish the vector according to the encoding index for each vector in each vector group.

In the embodiments of the present invention, more than two vectors are jointly encoded, and an encoding index of at least one vector is split and then combined between different vectors, so that encoding idle spaces of different vectors can be recombined, thereby helping to save encoding bits, and because an encoding index of a vector is split and then shorter (compared with an index before splitting) split indexes are recombined, thereby helping to reduce requirements for the bit width of operating parts in encoding/decoding calculation.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of vector division of a signal;

FIG. 2 is a schematic flowchart of an encoding method according to Embodiment 1 of the present invention;

FIG. 3 is a schematic flowchart of an encoding method according to Embodiment 2 of the present invention;

FIG. 4 is a schematic flowchart of an encoding method according to Embodiment 3 of the present invention;

FIG. 5 is a schematic flowchart of an encoding method according to Embodiment 4 of the present invention;

FIG. 6 is a schematic flowchart of an encoding method according to Embodiment 5 of the present invention;

FIG. 7 is a schematic flowchart of a decoding method according to Embodiment 6 of the present invention;

FIG. 8 is a schematic flowchart of a decoding method according to Embodiment 7 of the present invention;

FIG. 9 is a schematic logic structural diagram of an encoder according to Embodiment 8 of the present invention;

FIG. 10 is a schematic logic structural diagram of an encoder according to Embodiment 9 of the present invention;

FIG. 11 is a schematic logic structural diagram of a decoder according to Embodiment 10 of the present invention;

FIG. 12 is a schematic logic structural diagram of a decoder according to Embodiment 11 of the present invention;

FIG. 13 is a schematic diagram of a 4-track highest-4-bits joint encoding procedure based on an embodiment of the present invention; and

FIG. 14 is a schematic diagram of a 4-track 2-stage-cascade highest-8-bits joint encoding procedure based on an embodiment of the present invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

An embodiment of the present invention provides a vector joint encoding method, more than two vectors are jointly encoded, and an encoding index of at least one vector is split and then combined between different vectors, thereby saving encoding bits and reducing the length of data participating in calculation. Embodiments of the present invention further provide a vector joint decoding method and a vector joint encoder/decoder correspondingly.

In order to make the description concise and understandable, pulse distribution on a track is used as an example for a vector below, and joint encoding for multiple tracks is described to embody joint encoding for multiple vectors. It is easily understood that, the present invention is not limited to encoding/decoding for track pulse distribution, and as long as it is needed to perform encoding/decoding processing on multiple vectors formed of a group of data, the solution provided in the present invention may be adopted, and “track” and “pulse” merely endow data in a vector with a concrete meaning, and do not form a substantial limitation.

In a voice encoder, information about positions and symbols (if related) of all pulses on each track is obtained through code book searching, and the information needs to be completely transferred to a decoding end, so that the information about the positions and the symbols (if related) of all the pulses can be uniquely recovered at the decoding end, and meanwhile, in order to reduce a bit rate as much as possible, it is expected to use bits as fewer as possible to transfer the information.

It can be known through theoretical analysis that, the permutation and combination number of positions and symbols (if related) of all pulses on a same track is a minimal value of a codebook space, and the number of corresponding encoding bits is a theoretical lower limit value. The total number of positions and the total number of pulses on a track are fixed. As far as a case that the total number of positions and the total number of pulses on a track are different values is concerned, the permutation and combination number of positions and symbols of all pulses is not always an integer power of 2, and therefore the theoretical lower limit value of the number of encoding bits is not always an integer, and in this case, the actual number of encoding bits for single-track encoding is at least the integer section of the theoretical lower limit value plus 1, so that occurrence of partial idle codebook space is unavoidable. For example, Table 1 provides a theoretical lower limit value and an actual lower limit value of the number of encoding bits, and a situation of an idle space, which are on a track with the total number of positions being 16 in a case that the total number N of pulses needed to be encoded is 1 to 7 (that a pulse has a symbol is taken into account).

	TABLE 1
	Number of bits (bit) needed
	Total		Actual
	permutation		lower limit
	and	Theoretical	value for	Number of
	combination	lower limit	single-track	idle	Idle
N	number	value	encoding	combinations	proportion
1	32	5	5	0	0
2	512	9	9	0	0
3	5472	12.4179	13	2720	33.2%
4	44032	15.4263	16	21504	32.8%
5	285088	18.1210	19	239200	45.6%
6	1549824	20.5637	21	547328	26.1%
7	7288544	22.7972	23	1100064	15.1%

It may be seen from Table 1 that, in a majority of cases, the actual lower limit value still brings about large waste of the codebook space, and therefore, the present invention proposes that, more than two tracks are jointly encoded, an encoding index of at least one track is split into at least two split indexes, and then a split index of a track and split indexes or encoding indexes of other tracks are combined and then used for encoding.

This method is based on an idea as follows: joint encoding for more than two tracks may enable idle codebook spaces in single-track encoding to be combined, and once combined idle spaces are sufficient, one actual encoding bit may be reduced. However, if encoding indexes on more than two tracks are directly combined, a final encoding length may be large, or even may exceed the bit width (such as 64 bits) generally used for operating, and at this time, it is needed to design dedicated calculation processing procedure codes for operations of encoding or decoding such as addition, subtraction, multiplication, and division, thereby resulting in increase of the processing complexity.

Therefore, it is taken into account that, an encoding index of at least one track is split, inter-track combination is performed with at least one split index, and in this way, inter-track idle spaces can be combined to a certain extent, and meanwhile, the length of a value participating in operating is also reduced.

The theoretical analysis for the principle of the multi-track joint encoding of the present invention is provided above, and below, various preferable solutions are illustrated in detail with specific embodiments, respectively.

Embodiment 1

A track joint pulse encoding method, as shown in FIG. 2, includes the following steps.

A1: Calculate an encoding index Ind_t of each track, where a subscript t denotes a t^th track, t∈[0, T−1], and T is the number of tracks and is an integer greater than or equal to 2.

Various existing methods may be adopted for calculating Ind_t of each track. For example, for the single-track encoding index calculation method provided in Chinese Patent Application Publication CN 101295506 published Oct. 29, 2008, reference may be made to row 18, page 13 to row 9, page 15 (Embodiment 2, FIG. 14 and FIG. 15) in the specification of the application document, and for a corresponding decoding calculation method, reference may be made to row 23, page 16 to row 12, page 17 (Embodiment 4) in the specification of the application document. Also for example, for the single-track encoding index calculation method provided in Chinese Patent Application Publication CN101388210 published Mar. 18, 2009, reference may be made to row 23, page 8 to row 7, page 10 (Embodiment 2, FIG. 7) in the specification of the application document, and for a corresponding decoding calculation method, reference may be made to row 10, page 21 to row 27, page 21 (Embodiment 6) in the specification of the application document.

In order to better save encoding bits, during selection of an Ind_t calculation method, a calculation method capable of reaching the theoretical lower limit value of the number of single-track encoding bits may be selected as much as possible. Moreover, a calculation method enabling the value range of Ind_t to be continuous or be continuous as much as possible is further preferably used, so as to utilize the idle space. It should be noted that, as long as both the encoding end and the decoding end may determine a calculation method adopted by a track, different tracks may adopt different Ind_t calculation methods.

A2: Split at least one Ind_t at least once into at least two sections, where splitting at least once is equivalent to splitting Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor (SLF_t).

The splitting of Ind_t may be understood as follows: that index information is borne by a parameter is converted into that index information is borne by more than two parameters. For example, that originally a parameter with the value range being [0, 99] indicates one hundred cases may be split into that two parameters with the value range being [0, 9] commonly indicate the one hundred cases.

Ind_t may be split into multiple split indexes, such as Ind_t0, Ind_t1, Ind_t2, Ind_t3, . . . , and the like. In actual calculation, a split index needed by the present invention may be obtained by directly splitting Ind_t, or may be obtained by splitting again a section split by Ind_t; no matter which case it is, the splitting for obtaining the split index (that is, a split index used for subsequent combination) needed by the present invention may be equivalently understood as a process of splitting Ind_t into two split indexes Ind_t0 and Ind_t1, one of which is a split index used for subsequent combination, and the other of which may be understood as a section including the rest information of Ind_t.

Therefore, the factor SLF_t may be understood as follows: the value range of the original Ind_t is divided into several intervals, the length of each interval is not greater than SLF_t, SLF_t is a positive integer, Ind_t0 denotes a serial number of an interval to which Ind_t belongs, and Ind_t denotes a serial number of Ind_t in the interval to which Ind_t belongs (Ind_t1≤SLF_t), and Ind_t≤Ind_t0×SLF_t+Ind_t1;

When the less-than sign is taken in the above formula, it means that, a space jointly provided by Ind_t0 and Ind_t1 may be greater than a space originally occupied by Ind_t; because generally the idle encoding space of Ind_t has some remainders, if the joint space of Ind_t0 and Ind_t1 is slightly larger, generally final encoding bit saving may not be influenced unfavorably.

The most economical case is that splitting is able to be performed as much as possible according to the space size of Ind_t, that is Ind_t0=Int(Ind_t/SLF_t), where Int( ) denotes rounding down to an integer, and Ind_t1=Ind_t% SLF_t, where % denotes taking a remainder.

In this case, if the value range of Ind_t is continuous, compared with the value range of Ind_t, what is idle of the space jointly provided by Ind_t0 and Ind_t1 is just a difference between a maximal value of Ind_t1 and SLF_t when Ind_t0 takes a maximal value.

Different tracks may adopt different SLF_t, and if multiple split indexes are split on a same track, each splitting may also use different SLF_t, as long as SLF_t used by a certain split index split by a certain track is determined.

A3: Combine a split index of at least one track and split indexes or encoding indexes of other tracks to generate a combined index (Ind_SLF).

Corresponding to “splitting”, the “combination” may be understood as follows: that index information is borne by more than two parameters is converted into that index information is borne by one parameter. The value range of a parameter after combination is greater than or at least equal to a product of value ranges of each parameter before combination. The combination of multiple parameters may be denoted with a formula as follows: AI=((((a₁×A_I-1+a_I-1) . . . )×A₂+a₂)×A₁+a₁)×A₀+a₀, where AI denotes a parameter after combination, a_i denotes I+1 parameters before combination, i∈[0, I], and A_i denotes the number of all values of a_i. This combination manner is the most compact, the value range of a parameter after combination is equal to a product of value ranges of each parameter before combination, and all parameter combination mentioned here may adopt this manner. Other combination manners may be further adopted. For example, the value range of a parameter after combination may be greater than a product of value ranges of each parameter before combination as long as a parameter after combination can completely retain information of each parameter before combination; if the value space of a parameter after combination does not exceed much compared with a product of value spaces of each parameter before combination, generally final encoding bit saving may not be influenced unfavorably.

In the present invention, a generated combined index at least includes a split index of a track, and furthermore, split indexes or encoding indexes of other tracks may participate in combination. It should be noted that, the present invention does not limit the number of combined indexes, and if a track provides multiple split indexes for participating in combination, the split indexes may be respectively combined into different combined indexes.

A4: Perform encoding according to a combined index and other uncombined split indexes.

(1) The combined index and each uncombined split index may be directly encoded respectively. Because each combination section in the combined index carries a part of an idle space from a track, it is possible to save encoding bits fixedly.

(2) Furthermore, a variable-length encoding manner may be further adopted for a combined index, so as to save more encoding bits of the combined index, that is comparing the combined index Ind_SLF and adjusting a threshold value (THR), where, THR≤2^(K_SLF)−Ind_SLF,max, 2^(K_SLF) denotes a K_SLF power of 2, the K_SLF is the length of the bit sequence of the Ind_SLF,max, and the Ind_SLF,max denotes a maximal value of the Ind_SLF; if the Ind_SLF is less than the THR, encoding the Ind_SLF using a first number of encoding bits; otherwise, encoding the Ind_SLF added with an offset value THR₀ using a second number of encoding bits, in which THR≤THR₀≤2^(K_SLF)−Ind_SLF,max, the first number is less than the second number, the second number is less than or equal to the K_SLF, and the first number and the second number are both positive integers.

For the principle, specific deduction and description of the foregoing method for saving bits, reference is made to Chinese Patent Application CN200910150637.8 filed Jun. 19, 2009.

(3) The combined index may be split, then combined with other uncombined split indexes and then encoded.

That is, the combined index is split into T1 recombined indexes Ind_t0′ according to a set value range, in which T1 is less than or equal to the number of tracks generating the combined index (the number of values of t here may be less than the number T of values of t originally denoting a t^th track, but because a recombined index is used for recombination with an uncombined split index of a corresponding track t, the subscript t is still used continuously to denote being corresponding to a track), a value range of at least one Ind_t0′ is greater than a value range of the split index, participating in combination, of the corresponding track t, and a value range of at least one Ind_t0′ is less than a value range of the split index, participating in combination, of the corresponding track t. Here, the splitting of a combined index may be performed with reference to the splitting of an encoding index in step A2, but it is needed to split T1 recombined indexes, each splitting may be based on the same or different value ranges, and the foregoing limitation to the value range of the split recombined index enables the idle space converged by the combined index to be concentrated on certain recombined indexes to a certain extent during splitting, thereby facilitating saving of encoding bits.

Each recombined index and an uncombined split index of a corresponding track are respectively combined and then encoded, and if a track without being allocated a recombined index exists, an uncombined split index of the track is encoded. For the combination of a recombined index and a split index of a corresponding track, reference may be made to the aforementioned combined index combination process. Because the combined index aggregates the idle space, in a case that the encoding bits of an original track for an allocated recombined index are not increased, the splitting of the combined index is completed before certain tracks may be allocated a recombined index, and in this case, only the uncombined split index of the track may be encoded.

If there is another track which does not provides a split index and whose encoding index does not participate in combination, the track may be regarded as an independent encoding track, and is not discussed in the present invention.

Embodiment 2

A track joint pulse encoding method. This embodiment provides a preferable solution for selecting split indexes to be combined on the basis of Embodiment 1, as shown in FIG. 3, which includes the following steps.

B1: Calculate an encoding index (Ind_t) of each track.

B2: Split at least one Ind_t at least once into at least two sections, where splitting at least once is equivalent to splitting Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t.

Step B1 and step B2 may be executed with reference to step A1 and step A2 of Embodiment 1.

B3: Combine a split index of at least one track and split indexes or encoding indexes of other tracks to generate a combined index (Ind_SLF), and, when a split index participating in combination is selected, select, for a track t, a split index capable of embodying space occupancy characteristics of the Ind_t to participate in combination.

Being capable of embodying space occupancy characteristics of the Ind_t refers to that, compared with other split indexes, an occupancy rate of a value range of a selected split index for an encoding space is the closest to an occupancy rate of a value range of the Ind_t for the encoding space.

In the present invention, because idle spaces of different tracks are combined to assist in saving encoding bits, in order to achieve better effect of utilizing the idle space, it is expected that the value range of the split index representing the track t to perform combination can maintain the idle proportion of the encoding space of the value range of Ind_t as much as possible, which means that the value range of the rest section of Ind_t is closer to full utilization of the encoding space, such as, close to an integer power of 2.

B4: Perform encoding according to the combined index and other uncombined split indexes.

Step B4 may be executed with reference to step A4 of Embodiment 1.

Adoption of the solution of this embodiment can further ensure the effect that a combined index saves encoding bits.

Embodiment 3

A track joint pulse encoding method. This embodiment provides a preferable solution for splitting an encoding index on the basis of Embodiment 1, as shown in FIG. 4, which includes the following steps.

C1: Calculate an encoding index Ind_t of each track.

Step C1 may be executed with reference to step A1 of Embodiment 1.

C2: Split at least one Ind_t at least once into at least two sections, where splitting at least once is equivalent to splitting Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t, Ind_t0=Int(Ind_t/SLF_t), and Ind_t1=Ind_t% SLF_t, where, SLF_t=2^(K_t), or SLF_t=Int(Ind_t,max/2^(K_t)), K_t is a positive integer, and Ind_t,max denotes a maximal value of Ind_t.

(1) When SLF_t=2^(K_t), Ind_t0=Int(Ind_t/2^(K_t)), and Ind_t1=Ind_t%2^(K_t).

In this case, the value range of Ind_t1 can fully utilize an encoding space, what is idle is just a difference between a maximal value of Ind_t1 and 2^(K_t) when Ind_t0 takes a maximal value, and the value range of Ind_t0 fully retains the space occupancy characteristics of Ind_t. The smaller the K_t is, the closer the space characteristics of Ind_t0 are to Ind_t, and when K_t is 0, Ind_t0 is degenerated into Ind_t.

(2) When SLF_t=Int(Ind_t,max/2^(K_t)), this case is just opposite to (1), and actually, it is equivalent to that Ind_t0 and Ind_t1 in (1) exchange positions, the value range of Ind_t0 can fully utilize the encoding space, and the value range of Ind_t1 fully retains the space occupancy characteristics of Ind_t. The smaller the K_t is, the closer the space characteristics of Ind_t1 are to Ind_t, and when K_t is 0, Ind_t1 is degenerated into Ind_t.

C3: Combine a split index of at least one track and split indexes or encoding indexes of other tracks to generate a combined index Ind_SLF.

It may be known according to the analysis in Embodiment 2 that, if a combined index needs to achieve better effect of saving encoding bits, it is needed to select a split index that retains the space characteristics of Ind_t as much as possible, and therefore, for the track t providing a split index to participate in combination, if SLF_t=2^(K_t), it is appropriate to select Ind_t0 to participate in combination, and if SLF_t=Int(Ind_t,max/2^(K_t)), it is appropriate to select Ind_t1 to participate in combination.

Moreover, it may be known according to the analysis of step C2 that, the smaller the K_t is, the better the space characteristics retained by the split index selected to participate in combination are. However, the length of a corresponding combined index is increased, and therefore, the length of the split index participating in combination may be determined according to the condition of an actual track (such as, the length of the original encoding index, and the idle extent of the space). During actual design, proportions of idle spaces, which can be retained by split indexes used for combination and corresponds to different values of K_t, may be calculated one by one, and a K_t value, which can maintain an idle space of a high proportion and does not cause a split index participating in combination to be excessively long, is selected. For example, it may be selected that K_t=K_t,max−2, or K_t=K_t,max−3, or K_t=K_t,max−4, where the K_t,max is the length of a bit sequence of the Ind_t,max.

C4: Perform encoding according to the combined index and other uncombined split indexes.

Step C4 may be executed with reference to step A4 of Embodiment 1.

The Ind_t splitting manner provided in this embodiment can ensure that one of split indexes retains the space occupancy characteristics of Ind_t as much as possible, so that the combined index may better save encoding bits.

Embodiment 4

A track joint pulse encoding method. This embodiment provides a simple, convenient and easy solution for splitting an encoding index on the basis of Embodiment 1, as shown in FIG. 5, which includes the following steps.

D1: Calculate an encoding index Ind_t of each track.

Step D1 may be executed with reference to step A1 of Embodiment 1.

D2: Split at least one Ind_t at least once into at least two sections, where splitting at least once is equivalent to splitting Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t, and the specific practice is, selecting a value of a bit located at a set position of Ind_t as Ind_t0, and a value of bits at remaining positions as Ind_t1.

In this embodiment, a split index is directly obtained from Ind_t according to a bit splitting manner. In this case, SLF_t=2^(K_t), where the number of bits at the set position is K_t0, the number of bits at the remaining position is K_t, K_t0+K_t=K_t,max, K_t,max is the length of the bit sequence of Ind_t,max, and Ind_t,max denotes a maximal value of Ind_t.

It is easily understood that, when the set bits forming a certain split index are not continuous, it means that the value range of Ind_t is discontinuously allocated into several intervals with a length being SLF_t, but occupancy characteristics of the split index for the encoding space is not influenced by whether an included value range is continuous, and is only relevant to whether the intervals are fully padded.

It should be noted that, when a parameter is split in a bitwise manner, a position of a bit is described according to the bit sequence length of the maximal value of the parameter. For example, if the bit sequence length of Ind_t,max is 10, a value of Ind_t may be merely 2 at a certain time (unless otherwise specially illustrated, the values used here are all decimal), and the effective bit sequence length of Ind_t are only 2 bits, but when the parameter is split in a bitwise manner, a position needed to split a bit is still calculated starting from the highest 10 bits.

Moreover, it may be seen in contrast with the situation (1) described in step C2 of Embodiment 3 that, the situation (1) is actually equivalent to that in the bit sequence with a length being K_t,max, a value of K_t0 bits starting from the highest bit of Ind_t is selected as Ind_t0, and a value of remaining bits is selected as Ind_t1, that is, Ind_t is divided into two segments according to a bit priority, and a value of each segment is corresponding to a split index.

D3: Combine a split index of at least one track and split indexes or encoding indexes of other tracks to generate a combined index Ind_SLF, and the specific practice is, selecting a split index of Ind_t at least including values of the highest two bits to participate in combination.

It may be known according to the analysis in Embodiment 2 that, if a combined index is needed to achieve better effect of saving encoding bits, it is needed to select a split index that retains the space characteristics of Ind_t as much as possible, and when Ind_t is split in a bitwise manner, several highest bits are the most capable of reflecting the space characteristics of Ind_t, and therefore, no matter whether bits forming a split index are continuous, a split index including values of the highest two bits is preferably selected to participate in combination. A more preferable situation is, when Ind_t is split according to the bit priority of the bit sequence, a split index corresponding to a split segment located at a high bit is selected to participate in combination, such as Ind_t0 in the situation (1) described in step C2 of Embodiment 3.

D4: Perform encoding according to the combined index and other uncombined split indexes.

Step D4 may be executed with reference to step A4 of Embodiment 1. Moreover, if a manner is adopted in which the combined index is split, then combined with other uncombined split indexes and then encoded, the splitting of the combined index may also be performed with reference to the foregoing step D2, that is, splitting the total length K_SLF of the bit sequence of the combined index into T1 sections according to a set length, in which T1 is less than or equal to the number of tracks generating the combined index, a value of each section is corresponding to one Ind_t0′, the K_SLF is the length of the bit sequence of the Ind_SLF,max, and the Ind_SLF,max denotes a maximal value of the Ind_SLF. For example, splitting the K_SLF according to a K_t0 value used by the track t generating the combined index, in which the number of bits split by each Ind_t0′ correspondingly is less than or equal to the K_t0 value used by a corresponding track t.

For example, if four tracks exist, and each track respectively extracts the highest four bits of an encoding index as a split index for participating in combination, after the combination, the combined index may be split likewise with 4 bits as a segment, and is used as a recombined index to replace the highest four bits of an original encoding index. The combined index combines an idle space, and the bit sequence length may be less than 4+4+4+4, so the recombined index obtained by a certain track may only have 3 bits or even less, which become encoding bits which are fixedly saved. It should be noted that, when split indexes are combined into a combined index, the operation of saving bits is completed, so when the combined index is split in a bitwise manner, no special rule is needed, and as long as an encoding end and an decoding end adopt a same rule, bits may be continuously segmented or may not be continuously extracted.

This embodiment provides a simple, convenient and easy Ind_t splitting manner, which can not only ensure that one of split indexes retains the space occupancy characteristics of Ind_t as much as possible, but also is implemented conveniently.

Embodiment 5

A track joint pulse encoding method. This embodiment provides a layered cascade track joint pulse encoding method on the basis of Embodiment 1, as shown in FIG. 6, which includes the following steps.

E1: Group tracks participating in joint encoding, in which each group at least includes two tracks.

For example, sixteen tracks participate in joint encoding, every four tracks form one group, and totally four groups are formed.

E2: Split and combine an encoding index in each group with reference to the manner of Embodiment 1.

That is, an encoding index of each track is calculated; at least two encoding indexes are split; each encoding index is split at least once into at least two sections, where splitting at least once is equivalent to splitting an encoding index into two one-stage split indexes according to a set one-stage factor, one one-stage split index denotes serial numbers of several intervals with a length not greater than a set value, the other one-stage split index denotes a serial number of the encoding index in the interval to which the encoding index belongs; and one-stage split indexes from at least two tracks are combined to generate a one-stage combined index.

E3: Starting from m=2, repeat the following operation of generating an m-stage combined index until m=M; split at least two (m−1)-stage combined indexes, in which m∈[2, M], M is an integer greater than or equal to 2; split each (m−1)-stage combined index at least once into at least two sections, where the splitting at least once is equivalent to splitting an (m−1)-stage combined index into two m-stage split indexes according to a set m-stage factor; and combine m-stage split indexes from at least two (m−1)-stage combined indexes to generate an m-stage combined index.

By regarding the (m−1)-stage combined index needed to be split as the encoding index in each aforementioned embodiment, the foregoing splitting and combination procedure for each stage may be performed with reference to corresponding description, and is not repeated again. For example, with reference to Embodiment 4, the one-stage split index used for generating the one-stage combined index may be a value intercepted from a set bit of a corresponding encoding index starting from the highest bit; and the m-stage split index used for generating the m-stage combined index may be a value intercepted from a set bit of a corresponding (m−1)-stage combined index starting from the highest bit.

E4: Perform encoding according to the M-stage combined index and other uncombined one-stage to (M−1)-stage split indexes.

Similar to step A4 in Embodiment 1, the M-stage combined index and other uncombined one-stage to (M−1)-stage split indexes may be directly encoded respectively. Or further, the M-stage combined index may be split, then combined with other uncombined (M−1)-stage split indexes and then encoded, which is not repeated again here.

It should be noted that, although in the foregoing encoding procedure, in order to make description clear and be provided with cyclical operability, splitting and combination of each stage are consistently described, actually, it may also be the same as Embodiment 1, a split index is permitted to participate in combination, and also an encoding index (for first-stage combination) or combined index (for combination above second-stage) is permitted to directly participate in combination, or even a split index may be permitted to participate in combination (such as, an uncombined one-stage split index participates in generation of a three-stage combined index) by skipping a stage, but in these cases, it is needed to set a splitting and combination rule for each stage, so that the encoding end and the decoding end maintain consistent.

This embodiment is applicable to joint encoding for excessive tracks, such as 16 or 32 or even more tracks, in a multi-track case, even if each track only extracts several bits of split indexes to be combined, the combined index is also made excessively long, and in this case, the foregoing layered cascade manner is adopted, through multi-layered splitting and combination, it can be ensured that idle spaces are fully combined, and the combined index is not made excessively long.

Embodiment 6

A track joint pulse decoding method. The decoding method provided in this embodiment decodes the joint code obtained according to the encoding method of Embodiment 1 to Embodiment 4, and a decoding procedure is an inverse procedure for an encoding procedure, as shown in FIG. 7, which includes the following steps.

F1: Acquire a joint code, and acquiring, from the joint code, a combined index and an uncombined split index corresponding to a track.

The process of extracting each index from the joint code may be performed according to the inverse process of executing an operation on each index during encoding.

For example, if an encoding end directly encodes a combined index and each uncombined split index respectively, each index is directly decoded.

Also for example, if the encoding end adopts a variable-length encoding manner for the combined index, the encoding length of the combined index is determined and then is correspondingly decoded (with reference to Chinese Patent Application CN200910150637.8).

Further for example, if the encoding end splits the combined index, which is then combined with other uncombined split indexes and then encoded, a code corresponding to each track is extracted from the joint code, and a recombined index and an uncombined split index are split from a code of each track; if a code including no recombined index exists, an uncombined split index corresponding to a corresponding track is directly obtained, and then all recombined indexes are spliced into a combined index according to the splitting manner of the encoding end.

F2: Split the combined index into split indexes corresponding to the track, or split the combined index into a split index and an encoding index that correspond to the track.

This step is performed inversely with reference to the manner in which the encoding end generates the combined index.

For example, if the encoding end generates a combined index in an AI=((((a_i×A_I-1+a_I-1) . . . )×A₂+a₂)×A₁+a₁)×A₀+a₀ manner, during decoding, a value of a₀ may be obtained by calculating AI % A₀, then a value of a₁ may be obtained by calculating Int(AI/A₀) % A₁, and the rest may be deduced by analogy, until all a_i is obtained.

F3: For each track participating in splitting the encoding index, splice, according to a split manner of an encoding end, a split index not participating in combination and a split index participating in combination of the track to generate the encoding index of the track.

For those tracks that are not split but directly use an encoding index to participate in combination, step F2 may obtain the encoding index of those tracks, and this step may be omitted.

F4: Reestablish a pulse sequence on the track according to the encoding index of the track for each track.

The encoding end uses the manner in which the split indexes are combined, so likewise, a decoding end may obtain the benefit of reducing the bit width requirement for value processing.

Embodiment 7

A track joint pulse decoding method. The decoding method provided in this embodiment decodes the joint code obtained according to the layered cascade encoding method of Embodiment 5, and a decoding procedure is an inverse procedure of an encoding procedure, as shown in FIG. 8, which includes the following steps.

G1: Acquire a joint code, and acquiring, from the joint code, an M-stage combined index, an uncombined m-stage split index corresponding to each (m−1)-stage combined index, and an uncombined one-stage split index corresponding to each track, in which M is an integer greater than or equal to 2, and m∈[2, M].

With reference to step F1 of Embodiment 6, the index extracting process is performed according to the inverse process of executing an operation on each index during encoding likewise, and is not repeated again.

G2: Starting from m=M, repeat the following operation of generating an (m−1)-stage combined index until m=2; split each m-stage combined index into an m-stage split index corresponding to each (m−1)-stage combined index for generating the m-stage combined index; and for each (m−1)-stage combined index, splice, according to a split manner of an encoding end, an m-stage split index not participating in combination and an m-stage split index participating in combination of the (m−1)-stage combined index to generate the (m−1)-stage combined index.

G3: Split each one-stage combined index into a one-stage split index corresponding to each track in a track group for generating the one-stage combined index; and for each track in each track group, splice, according to the split manner of the encoding end, a one-stage split index not participating in combination and a one-stage split index participating in combination of the track to generate an encoding index.

G4: Reestablish a pulse sequence on the track according to the encoding index for each track in each track group.

Embodiment 8

A vector joint pulse encoder 10. The encoder provided in this embodiment may be used for executing the encoding method provided in Embodiment 1 to Embodiment 4, as shown in FIG. 9, which includes an encoding index calculation unit 101 configured to calculate an encoding index Ind_t of each vector, where a subscript t denotes a t^th vector, t∈[0, T−1], and T is an integer greater than or equal to 2; a vector index splitting unit 102 configured to split at least one Ind_t at least once into at least two sections, where splitting at least once is equivalent to splitting the Ind_t into two split indexes Ind_t0 and Ind_t1 according to a set factor SLF_t, the SLF_t is a positive integer, the Ind_t0 denotes a serial number of an interval to which the Ind_t belongs, the Ind_t1 denotes a serial number of the Ind_t in the interval to which the Ind_t belongs, a length of the interval is not greater than the SLF_t, and Ind_t≤Ind_t0×SLF_t+Ind_t1; an index combination unit 103 configured to combine a split index of at least one vector and split indexes or encoding indexes of other vectors to generate a combined index Ind_SLF; and an encoding unit 104 configured to perform encoding according to the combined index generated by the index combination unit 103 and other uncombined split indexes split by the vector index splitting unit 102.

Further, the encoding unit 104 may include a recombining sub-unit 1041 configured to split the combined index into T1 recombined indexes Ind_t0′ according to a set value range, in which T1 is less than or equal to the number of vectors generating the combined index, a value range of at least one Ind_t0′ is greater than a value range of the split index, participating in combination, of the corresponding vector t, and a value range of at least one Ind_t0′ is less than the value range of the split index, participating in combination, of the corresponding vector t; and an encoding sub-unit 1042 configured to combine each recombined index and an uncombined split index of a corresponding vector respectively and then encode, and if a vector without being allocated a recombined index exists, encode an uncombined split index of the vector.

Embodiment 9

A vector joint pulse encoder 20. The encoder provided in this embodiment may be used for executing the encoding method provided in Embodiment 5, as shown in FIG. 10, which includes a vector index calculation unit 201 configured to group vectors participating in joint encoding, in which each group at least includes two vectors; in each group, calculate an encoding index of each vector; split at least two encoding indexes; split each encoding index at least once into at least two sections, where the splitting at least once is equivalent to splitting an encoding index into two one-stage split indexes according to a set one-stage factor, one one-stage split index denotes serial numbers of several intervals with a length not greater than a set value, the other one-stage split index denotes a serial number of the encoding index in an interval to which the encoding index belongs; and combine one-stage split indexes from at least two vectors to generate a one-stage combined index; a cascade calculation unit 202 configured to, starting from m=2, repeat the following operation of generating an m-stage combined index until m=M; split at least two (m−1)-stage combined indexes, in which m∈[2, M], M is an integer greater than or equal to 2; split each (m−1)-stage combined index at least once into at least two sections, where splitting at least once is equivalent to splitting an (m−1)-stage combined index into two m-stage split indexes according to a set m-stage factor; and combine m-stage split indexes from at least two (m−1)-stage combined indexes to generate an m-stage combined index; and an encoding unit 203 configured to perform encoding according to the M-stage combined index and other uncombined one-stage to (M−1)-stage split indexes.

During practical implementation, the cascade calculation unit 202 may multiplex a section in the vector index calculation unit 201, in which the section is used for splitting and combining an encoding index.

Embodiment 10

A vector joint pulse decoder 30. The decoder provided in this embodiment may be used for executing the decoding method of Embodiment 6, as shown in FIG. 11, which includes a decoding unit 301 configured to acquire a joint code, and acquire, from the joint code, a combined index and an uncombined split index corresponding to a vector; an index splitting unit 302 configured to split the combined index acquired by the decoding unit 301 into split indexes corresponding to the vector, or split the combined index into a split index and an encoding index that correspond to the vector; a vector index reestablishing unit 303 configured to, for each vector participating in splitting the encoding index, splice, according to a split manner of an encoding end, a split index not participating in combination and a split index participating in combination of the vector to generate the encoding index of the vector; and a vector reestablishing unit 304 configured to reestablish the vector according to the encoding index of the vector for each vector.

Further, the decoding unit 301 may include a decoding sub-unit 3011 configured to extract, from the joint code, a code corresponding to each vector; split a recombined index and an uncombined split index from the code of each vector; and if a code including no recombined index exists, directly obtain an uncombined split index corresponding to a corresponding vector; and a splicing sub-unit 3012 configured to splice all recombined indexes into a combined index according to the split manner of the encoding end.

Embodiment 11

A vector joint pulse decoder 40. The decoder provided in this embodiment may be used for executing the decoding method of Embodiment 7, as shown in FIG. 12, which includes a decoding unit 401 configured to acquire a joint code, and acquire, from the joint code, an M-stage combined index, an uncombined m-stage split index corresponding to each (m−1)-stage combined index, and an uncombined one-stage split index corresponding to each vector, in which M is an integer greater than or equal to 2, and m∈[2, M]; a cascade recovering unit 402 configured to, starting from m=M, repeat the following operation of generating an (m−1)-stage combined index until m=2; split each m-stage combined index into an m-stage split index corresponding to each (m−1)-stage combined index for generating the m-stage combined index; and for each (m−1)-stage combined index, splice, according to a split manner of an encoding end, an m-stage split index not participating in combination and an m-stage split index participating in combination of the (m−1)-stage combined index to generate the (m−1)-stage combined index; a vector index reestablishing unit 403 configured to split each one-stage combined index into a one-stage split index corresponding to each vector in a vector group for generating the one-stage combined index; and for each vector in each vector group, splice, according to the split manner of the encoding end, a one-stage split index not participating in combination and a one-stage split index participating in combination of the vector to generate an encoding index; and a vector reestablishing unit 404 configured to reestablish the vector according to the encoding index for each vector in each vector group.

During practical implementation, the cascade recovering unit 402 may multiplex a section in the vector index reestablishing unit 403, in which the section is used for splitting and splicing a one-stage combined index.

In order to better understand the present invention, two specific examples based on the track joint encoding of the present invention are provided below.

Example 1

it is assumed that four tracks participate in joint encoding, and five pulses with symbols are encoded on each track (the number of pulses encoded on each track may also vary). Each track splits a split index participating in combination in a manner of intercepting the highest four bits of an encoding index (the number of bits intercepted by each track may also vary). As shown in FIG. 13, the encoding procedure is as follows.

(1) Encoding indexes Ind₀, Ind₁, Ind₂, and Ind₃ of four tracks are calculated respectively, and if the total number of track positions is 16, the value range of a 5-pulse encoding index is 0 to 285088, and the number of bits needed is 19.

(2) The highest four bits of each encoding index are intercepted to be used as the split index Ind_t0 participating in combination, and the remaining 15 bits are used as the split index Ind_t not participating in combination. That is, it is equivalent to that SLF_t=2^15.

(3) Ind₀₀, Ind₁₀, Ind₂₀, and Ind₃₀ are combined to generate the combined index Ind_SLF, and because the number of values of the 5-pulse encoding index is 285088, which is divided by SLF_t=2^15=32768 to obtain a quotient being 8.7001953125, the value range of Ind_t0 is 0 to 8, that is, the number of values of Ind_t0 is 9 (the number of values of Ind_t0 with another number of pulses may be deduced by analogy, which is not repeated again), and therefore,Ind_SLF=((Ind₃₀×9+Ind₂₀)×9+Ind₁₀)×9+Ind₀₀

It may be known that the value range of Ind_SLF is 0 to 6560, and a bit sequence length is 13.

(4) In this case, Ind_SLF, Ind₀₁, Ind₁₁, Ind₂₁, and Ind₃₁ may be directly encoded, a total encoding length is 13+15+15+15+15, and three bits are saved compared with single-track encoding.

Also, Ind_SLF may be split, then combined with Ind_t1 and then encoded, that is, Ind_SLF is split with 4 bits as a group (the splitting of the combined index according to the length of the split index participating in combination may intuitively embody the effect of saving bits, and definitely the splitting may also be performed according to other lengths; bits of each recombined index may vary, and bits forming a same recombined index may be unnecessarily continuous, which does not influence the effect of saving bits), Ind₃₀′, Ind₂₀′, Ind₁₀′, and Ind₀₀′ are obtained, and the length of the last recombined index Ind₃₀′ is only 1 bit; and then each recombined index is combined with an uncombined split index of a corresponding track, that is, the highest four bits of the encoding index of the corresponding track are replaced with each recombined index to obtain new encoding indexes Ind₃′, Ind₂′, Ind₁′, and Ind₀′, and the length of the new encoding index Ind₀′ is only 16 bits, and therefore three bits are fixedly saved.

Example 2

It is assumed that four tracks are divided into two groups (two tracks form a group, and the number of tracks included in each group may also vary) to participate in two-stage cascade joint encoding, and five pulses with symbols are encoded on each track (the number of pulses encoded on each track may also vary). Each track splits a split index participating in combination in a manner of intercepting the highest eight bits of an encoding index (the number of bits intercepted by each track may also vary). As shown in FIG. 14, the encoding procedure is as follows.

(1) Encoding indexes Ind₀, Ind₁, Ind₂, and Ind₃ of respective two tracks in two groups are calculated respectively, where, Ind₀, and Ind₁ are encoding indexes of two tracks of the first group, and Ind₂, and Ind₃ are encoding indexes of two tracks of the second group; if the total number of track positions is 16, the value range of the 5-pulse encoding index is 0 to 285088, and the number of bits required is 19.

(2) The highest 8 bits of an encoding index in the first group and the second group are intercepted to be used as a one-stage split index Ind_1,t0 participating in combination (the first subscript 1 denotes the number of stages, and it is the same below), and remaining 11 bits are used as a one-stage split index Ind_1,t1 not participating in combination. That is, it is equivalent to that the one-stage factor SLF_1,t=2^11.

(3) Ind_1,00 and Ind_1,10 are combined to generate a one-stage combined index Ind_1,SLF,0, and Ind_1,20 and Ind_1,30 are combined to generate a one-stage combined index Ind_1,SLF,1; the number of values of the 5-pulse encoding index is 285088, which is divided by SLF_t=21 1=2048 to obtain a quotient being 139.203125, so the value range of Ind_1,t0 is 0 to 139, that is, the number of values of Ind_1,t0 is 140, and therefore, Ind_1,SLF,0=Ind_1,10×140+Ind_1,00, Ind_1,SLF,1=Ind_1,30×140+Ind_1,20.

It may be known that the number of values of the one-stage combined indexes Ind_1,SLF,0 and Ind_1,SLF,1 is 19600, and a bit sequence length is 15, and therefore, each one-stage combined index saves one bit, and the one-stage joint encoding saves two bits.

(4) The two-stage joint encoding operation continues to be performed based on Ind_1,SLF,0 and Ind_1,SLF,1 (when multi-stage joint is performed, a previous stage combined index may be regarded as an encoding index in one-stage joint to be similarly split and combined, and therefore a subscript t is subsequently used continuously to denote a t^th previous stage combined index); the highest 8 bits of Ind_1,SLF,t are intercepted to be used as a two-stage split index Ind_2,SLF,t0 participating in combination, and the remaining 7 bits are used as a two-stage split index Ind_2,SLF,t1 not participating in combination. That is, it is equivalent to that the two-stage factor SLF₂,t=2^7.

(5) Ind_2,SLF,00 and Ind_2,SLF,10 are combined to generate a two-stage combined index Ind_2,SLF; because the number of values of Ind_1,SLF,t is 19600, which is divided by SLF_2,t=2^7=128 to obtain a quotient being 153.125, the value range of Ind_2,SLF,t0 is 0 to 153, that is, the number of values of Ind_2,SLF,t0 is 154, and therefore, Ind_2,SLF=Ind_2,SLF,10×154+Ind_2,SLF,00

It may be known that, the number of values of the two-stage combined index Ind_2,SLF is 23716, and the bit sequence length is 15, so the two-stage joint encoding saves one bit again.

(6) In this case, Ind_2,SLF, Ind_2,SLF,01, Ind_2,SLF,11, Ind_1,01, Ind_1,11, Ind_1,21, and Ind_1,31 may be directly encoded, a total encoding length is 15+7+7+11+11+11+11=73, and three bits are saved totally compared with single-track encoding.

Also, Ind_2,SLF may be split, then combined with Ind_1,SLF,t1 and then encoded, that is, Ind_2,SLF is split with 8 bits as a group (the splitting of the m-stage combined index according to the length of the m-stage split index participating in combination may intuitively embody the effect of saving bits, and definitely the splitting may also be performed according to other lengths; bits of each recombined index may vary, and bits forming a same recombined index may be unnecessarily continuous, which does not influence the effect of saving bits), Ind_2,SLF,10′ and Ind_2,SLF,00′ are obtained, and the length of the last recombined index Ind_2,SLF,00′ is only 7 bits; and then each recombined index is combined with an uncombined two-stage split index of a corresponding one-stage combined index, that is, the highest four bits of the corresponding one-stage combined index are replaced with each recombined index to obtain new one-stage combined indexes Ind_1,SLF,1′ and Ind_1,SLF,0′, and the length of the new one-stage combined index Ind_1,SLF,0′ is only 14 bits.

In this case, Ind_1,SLF,1′, Ind_1,SLF,0′, Ind₀₁, Ind₁₁, Ind₂₁, and Ind₃₁ may be directly encoded, the total encoding length is 15+14+11+11+11+11=73, and three bits are saved totally compared with single-track encoding.

Also, Ind_1,SLF,1′ and Ind_1,SLF,0′ may be split, then combined with Ind_1,t1 and then encoded, that is, Ind_1,SLF,1′ is split with 8 bits as a group to obtain Ind_1,30′ and Ind_1,20′, and the length of the recombined index Ind_1,20′ is only 7 bits; Ind_1,SLF,0′ is split with 8 bits as a group to obtain Ind_1,10′ and Ind_1,00′, and the length of the recombined index Ind_1,00′ is only 6 bits, and then, each recombined index is combined with an uncombined one-stage split index of a corresponding encoding index, that is, the highest four bits of the encoding index of the corresponding encoding index are replaced with each recombined index to obtain new encoding indexes Ind₃′, Ind₂′, Ind₁′, and Ind₀′, and the length of Ind₂′ is only 18 bits, the length of Ind₀′ is only 17 bits, and therefore three bits are fixedly saved.

Table 2 provides a case that during 4-track joint encoding, the highest-four-bit joint encoding saves bits in a case of different number of pulses (pulse with a symbol) on a track, and other cases that a multi-track joint or split index intercepts different bits may be deduced by analogy.

TABLE 2
		Highest-four-bit joint	Total
Number	Lower limit	encoding range and	number
of pulses	value for	the number of bits	of bits	Number
on 4	single-track	Value	Number of	of joint	of bits
tracks	encoding bit	range	encoding bits	encoding	saved
3, 3, 3, 3	52	0~14640	14	50	2
3, 3, 3, 4	55	0~14640	14	53	2
3, 3, 4, 4	58	0~14640	14	56	2
3, 4, 4, 4	61	0~14640	14	59	2
4, 4, 4, 4	64	0~14640	14	62	2
4, 4, 4, 5	67	0~11978	14	65	2
4, 4, 5, 5	70	0~9800	14	68	2
4, 5, 5, 5	73	0~8019	13	70	3
5, 5, 5, 5	76	0~6560	13	73	3
5, 5, 5, 6	78	0~8747	14	76	2
5, 5, 6, 6	80	0~11663	14	78	2
5, 6, 6, 6	82	0~15551	14	80	2
6, 6, 6, 6	84	0~20735	15	83	1
6, 6, 6, 7	86	0~24191	15	85	1
6, 6, 7, 7	88	0~28223	15	87	1
6, 7, 7, 7	90	0~32927	16	90	0
7, 7, 7, 7	92	0~38415	16	92	0

It may be seen that, encoding bits can be effectively saved using the joint encoding method of the present invention. The bits saved using joint encoding may be used for reducing a transmission code rate, and may also be used for the ISF coefficient quantization, the pitch period precision, and the gain of other modules, so as to be used for improving the encoding quality.

For example, the number of pulses on a track may be increased (the number of ACELP excitation code pulses is increased) in a case of an invariable code rate, so as to enhance the encoding quality. For example, several ACELP high-code-rate fixed codebooks for AMR-WB+, after using the highest-four-bit joint encoding shown in Table 2, may be added with several pulses at an original code rate, and details are as follows:

(1) 3 pulses are added at 18.25 kbps

				(highest-four-bit joint
	(AMR-WB+)			encoding)
	Number of pulses on			Number of pulses on 4
	4 tracks	bits	−>	tracks	bits
18.25	4, 4, 4, 4	64		4, 5, 5, 5	70
kbps	4, 4, 4, 4	64		4, 4, 4, 4	62
	4, 4, 4, 4	64		4, 4, 4, 4	62
	4, 4, 4, 4	64		4, 4, 4, 4	62

(2) 6 pulses are added at 19.85 kbps

				(highest-four-bit joint
	(AMR-WB+)			encoding)
	Number of pulses on			Number of pulses on 4
	4 tracks	bits	−>	tracks	bits
19.85	4, 4, 5, 5	72		5, 5, 5, 5	73
kbps	4, 4, 5, 5	72		5, 5, 5, 5	73
	4, 4, 5, 5	72		5, 5, 5, 5	73
	4, 4, 5, 5	72		4, 4, 5, 5	68

(3) 6 pulses are added at 19.85 kbps

				(highest-four-bit joint
	(AMR-WB+)			encoding)
	Number of pulses on			Number of pulses on 4
	4 tracks	bits	−>	tracks	bits
19.85	4, 4, 5, 5	72		5, 5, 5, 5	73
kbps	4, 4, 5, 5	72		5, 5, 5, 5	73
	4, 4, 5, 5	72		4, 5, 5, 5	70
	4, 4, 5, 5	72		4, 5, 5, 5	70

(4) 7 pulses are added at 19.85 kbps

				(highest-four-bit joint
	(AMR-WB+)			encoding)
	Number of pulses on			Number of pulses on 4
	4 tracks	bits	−>	tracks	bits
19.85	4, 4, 5, 5	72		5, 5, 6, 6	78
kbps	4, 4, 5, 5	72		4, 5, 5, 5	70
	4, 4, 5, 5	72		4, 5, 5, 5	70
	4, 4, 5, 5	72		4, 5, 5, 5	70

(5) 8 pulses are added at 23.05 kbps

				(highest-four-bit joint
	(AMR-WB+)			encoding)
	Number of pulses on			Number of pulses on 4
	4 tracks	bits	−>	tracks	bits
23.05	6, 6, 6, 6	88		7, 7, 7, 7	92
kbps	6, 6, 6, 6	88		7, 7, 7, 7	92
	6, 6, 6, 6	88		6, 6, 6, 6	83
	6, 6, 6, 6	88		6, 6, 6, 6	83

Persons skilled in the art should understand that all or a part of the steps of the methods according to the embodiments may be implemented by a program instructing relevant hardware. The program may be stored in a computer readable storage medium, and the storage medium may include a read only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

A vector joint encoding/decoding method and a vector joint encoder/decoder provided in the embodiments of the present invention are introduced in detail above. Specific embodiments are used for illustrating principles and implementation manners of the present invention. The above descriptions of the embodiments are merely for understanding the method and core ideas of the present invention. Meanwhile, persons skilled in the art may make modifications to specific implementation manners and application ranges according to the idea of the present invention. In conclusion, the content of the specification shall not be regarded as a limitation to the present invention.

Claims

1. A method for vector joint encoding for a voice signal on a CELP (code excited linear prediction) model, comprising: calculating, by a processor of an encoder, an encoding index Indt of each vector from a plurality of vectors, wherein the each vector is determined based on the voice signal, wherein a subscript t denotes a tth vector and t∈[0, T−1], and wherein T represents number of vectors and is an integer greater than or equal to 2;splitting the Indt at least once into two split indexes Indt0 and Indt1 according to a set factor SLFt, wherein the set factor SLFt is a set split factor SLF for the tth vector that varies according to the tth vector, wherein the set factor SLFt is a positive integer, wherein the Indt0 denotes a serial number of an interval to which the Indt belongs, wherein the Indt1 denotes a serial number of the Indt in the interval to which the Indt belongs, wherein a length of the interval is not greater than the SLFt, and wherein Indt≤Indt0×SLFt+Indt1;obtaining a combined index IndSLF based on a split index of at least one vector and a split index of other vector for a further splitting if any of the Indt0 and Indt1 is uncombined; andencoding the final output of the previous steps if all of the Indt0 and Indt1 is combined.

2. The method according to claim 1, wherein each of the vectors is represented as pulse distribution on a track, and wherein the encoding index is an index used for indicating the pulse distribution on the track.

3. The method according to claim 1, wherein the Indt0 split from the Indt is equal to Indt0=Int(Indt/SLFt), wherein Int( ) denotes rounding down to an integer, and the Indt1 split from the Indt is equal to Indt1=Indt% SLFt, and wherein % denotes taking a remainder.

4. The method according to claim 3, wherein SLFt=2^(Kt), wherein Kt is a positive integer, and wherein the obtaining the combined index IndSLF comprises: selecting the Indt0 to participate in a combination for a vector t providing a split index to participate in the combination; orsetting SLFt=Int(Indt,max/2(Kt)), wherein Indt,max denotes a maximal value of the Indt, and wherein obtaining a combined index IndSLF comprises selecting the Indt1 to participate in a combination for the vector t providing a split index to participate the combination.

5. The method according to claim 4, wherein Kt=Kt,max−2, Kt=Kt,max−3, or Kt=Kt,max−4, and wherein the Kt,max is a length of a bit sequence of the Indt,max.

6. The method according to claim 1, wherein the obtaining the combined index IndSLF comprises selecting a split index for embodying space occupancy characteristics of the Indt to participate in a combination for a vector t providing a split index to participate in the combination, wherein the embodying space occupancy characteristics of the Indt refers to that, compared with other split indexes, an occupancy rate of a value range of a selected split index for an encoding space is closest to an occupancy rate of a value range of the Indt for the encoding space.

7. The method according to claim 1, wherein the splitting the Indt at least once into the two split indexes Indt0 and Indt1 according to the set factor SLFt comprises: selecting a value of bits of the Indt as the Indt0, the bits being located at a set position; andselecting a value of bits at a remaining position as the Indt1, wherein SLFt=2^(Kt), wherein number of the bits at the set position is Kt0, wherein number of the bits at the remaining position is Kt, wherein Kt0+Kt=Kt,max, wherein Kt,max is a length of a bit sequence of Indt,max and wherein the Indt,max denotes a maximal value of the Indt.

8. The method according to claim 7, wherein the obtaining the combined index IndSLF comprises selecting, in a bit sequence with a length being Kt,max, a split index of the Indt to participate in a combination for a vector t providing a split index to participate in the combination, wherein the split index at least comprises a value of highest two bits.

9. The method according to claim 1, wherein the further splitting comprises: splitting the combined index IndSLF into T1 recombined indexes Indt0′ according to a set value range, wherein T1 is less than or equal to number of vectors generating the combined index, wherein a value range of at least one Indt0′ is greater than or equal a value range of the split index of a corresponding vector t, wherein the split index participates in a combination, wherein a value range of at least one Indt0′ is less than or equal the value range of the split index of the corresponding vector t, wherein the split index participates in another combination;combining each recombined index and an uncombined split index of a corresponding vector;encoding the combined each recombined index and an uncombined split index of a corresponding vector; andencoding an uncombined split index of the vector when the vector without having an allocated recombined index exists.

10. The method according to claim 9, wherein the splitting the combined index into T1 recombined indexes Indt0′ according to a set value range comprises splitting a total length KSLF of the bit sequence of the combined index into T1 sections according to a set length, wherein a value of each section corresponds to one Indt0′, wherein the KSLF is the length of the bit sequence of an IndSLF,max, and wherein the IndSLF,max denotes a maximal value of the IndSLF.

11. The method according to claim 10, wherein the splitting the Indt into the two split indexes Indt0 and Indt1 according to the set factor SLFt comprises: selecting in the bit sequence with the length being Kt,max, a value of Kt0 bits starting from highest bit of the Indt as the Indt0;selecting a value of remaining bits as the Indt1, wherein SLFt=2^(Kt), wherein Kt0+Kt=Kt,max, wherein the Kt,max is a length of a bit sequence of the Indt,max, wherein the Indt,max denotes a maximal value of the Indt,wherein Indt0 of a vector t that needs to provide a split index is selected as a base to obtain the combined index IndSLF,wherein the splitting the total length KSLF of the bit sequence of the combined index into the T1 sections according to the set length comprises splitting the KSLF according to the Kt0 value used by the vector t generating the combined index, andwherein number of bits split by each Indt0′ is less than or equal to the value of Kt0 used by corresponding vector t.

12. The method according to claim 1, wherein encoding the combined index IndSLF comprises: comparing the combined index IndSLF;adjusting a threshold value THR, wherein THR≤2^(KSLF)−IndSLF,max, wherein the KSLF is a length of a bit sequence of the IndSLF,max, and wherein the IndSLF,max denotes a maximal value of the IndSLF;encoding the IndSLF by using a first number of encoding bits when the IndSLF is less than the THR;encoding the IndSLF added with an offset value THR0 by using a second number of encoding bits when the IndSLF is not less than the THR, wherein THR≤THR0≤2^(KSLF)−IndSLF,max, wherein the first number is less than the second number, wherein the second number is less than or equal to the KSLF, and wherein the first number and the second number are both positive integers; andencoding other uncombined split indices.

13. A vector joint pulse encoder working on a CELP (code excited linear prediction) model for encoding a voice signal, comprising: a processor;a memory in communication with the processor containing computer instructions that when executed by the processor cause the processor to:calculate an encoding index Indt of each vector from a plurality of vectors, wherein the each vector is determined based on the voice signal, wherein a subscript t denotes a tth vector and t∈[0, T−1], and wherein T represents number of vectors and is an integer greater than or equal to 2;split the Indt at least once into two split indexes Indt0 and Indt1 according to a set factor SLFt, wherein the set factor SLFt is a set split factor SLF for the tth vector that varies according to the tth vector, wherein the set factor SLFt is a positive integer, wherein the Indt0 denotes a serial number of an interval to which the Indt belongs, wherein the Indt1 denotes a serial number of the Indt in the interval to which the Indt belongs, wherein a length of the interval is not greater than the SLFt, and wherein Indt≤Indt0×SLFt+Indt1; andobtain a combined index IndSLF based on a split index of at least one vector and a split index of other vector for a further splitting if any of the Indt0 and Indt1 is uncombined yet, and encoding the final output of the previous steps if all of the Indt0 and Indt1 is combined.

14. The encoder according to claim 13, wherein the computer instructions that when executed by the processor cause the encoder further to: split the combined index into T1 recombined indexes Indt0′ according to a set value range, wherein T1 is less than or equal to number of vectors generating the combined index, wherein a value range of at least one Indt0′ is greater than or equal a value range of the split index, participating in a combination, of a corresponding vector t, and wherein a value range of at least one Indt0′ is less than or equal the value range of the split index, participating in another combination, of the corresponding vector t; and combine each recombined index and an uncombined split index of a corresponding vector respectively and then perform encoding; andencode an uncombined split index of the vector when a vector without having an allocated recombined index exists.