CN101620854A - Method, system and device for frequency band expansion - Google Patents

Abstract

The invention discloses a frequency expansion method. The embodiment of the invention also provides corresponding system and device. Tone characteristic parameter and original high band signal are obtained; an excitation spectrum is generated according to the tone characteristic parameter; and a reconstructed high band signal is generated according to the excitation information and the contour information in the original high band signal. The technical scheme of the invention includes that key characteristic parameter of the original high band signal which is extracted at a coding terminal and a decoding terminal are adopted for reconstruction on a high band signal according to the tone characteristic parameter; the reconstructed high band signal is not only similar to the original high band signal in contour but also well matched in detail, and better hearing effect can be achieved by less bit resource consumption.

Description

The method of band spread, system and equipment

Technical field

The present invention relates to the signaling bearer technical field, be specifically related to method, system and the equipment of band spread.

Background technology

In the existing bearing technology, under the situation of broadband, ultra broadband part resource-constrained, usually can only encode to spectrum envelope, temporal envelope, the partial spectrum coefficient of broadband, ultra broadband part, carry out band spread in decoding end by these parametric techniques then, the signal of reconstruct broadband, ultra broadband part.

The method of existing band spread is representative with AMRWB+ (Adaptive multi-rate wide bandplus AMR-WB coding+) mainly, utilize the pumping signal of arrowband and the composite filter of high-band to generate high band signal, realize band spread.。

In research and practice process to prior art, the present inventor finds, prior art need consume more bit resource and describe high-band composite filter parameter, and it is many to exist consumption bit resource, the bad problem of effect inapplicable or reconstruct under the situation of bit resource-constrained.

Summary of the invention

The embodiment of the invention provides a kind of method, system and equipment of band spread, solves the many problems of consumption bit resource that prior art exists.

A kind of coding/decoding method comprises:

Obtain tonality feature parameter and original high band signal;

Generate excitation spectrum according to described tonality feature parameter;

Generate the high band signal of reconstruct according to the profile information in described excitation spectrum and the described original high band signal.

A kind of coding method comprises:

Frequency-region signal to the ultra broadband part carries out tonality feature analysis extraction tonality feature parameter;

Described tonality feature parameter is carried out quantization encoding.

A kind of band spread system comprises:

Coding side is used to send original high band signal and tonality feature parameter;

Decoding end is used for tonality feature parameter and original high band signal that the received code end sends; Generate excitation spectrum according to described tonality feature parameter; Generate the high band signal of reconstruct according to the profile information in described excitation spectrum and the described original high band signal.

A kind of decoding end comprises:

Receiving element is used to receive tonality feature parameter and original high band signal;

The excitation spectrum generation unit is used for generating excitation spectrum according to described tonality feature parameter;

High band signal generation unit is used for generating the high band signal of reconstruct according to the profile information of described excitation spectrum and described original high band signal.

A kind of coding side comprises:

The tonality feature parameter extraction unit is used for the frequency-region signal of ultra broadband part is carried out tonality feature analysis extraction tonality feature parameter;

Quantifying unit is used for described tonality feature parameter is carried out quantization encoding.

The embodiment of the invention adopts the key characterization parameter of extracting original high band signal at coding side: the tonality feature parameter, carry out coding transmission with few bit resource.In decoding end, generate excitation spectrum according to the tonality feature parameter, and finally generate the high band signal of reconstruct according to the profile information of original high band signal, realize band spread.Because when decoding end is carried out the reconstruct of high band signal, minutia based on original high band signal is operated, the high band signal of reconstruct except profile and original high band signal near, coupling is preferably arranged in detail, just can reach auditory effect preferably by less bit resource consumption.

Description of drawings

Fig. 1 is the process flow diagram of first embodiment of a kind of coding/decoding method of the present invention;

Fig. 2 is the process flow diagram of second embodiment of a kind of coding/decoding method of the present invention;

Fig. 3 is the process flow diagram of an embodiment of a kind of coding method of the present invention;

Fig. 4 is the structural drawing of an embodiment of a kind of band spread of the present invention system;

Fig. 5 is the structural drawing of an embodiment of a kind of band spread decoding end of the present invention;

Fig. 6 is the structural drawing of an embodiment of a kind of band spread coding side of the present invention.

Embodiment

See also the signaling process figure of Fig. 1 for first embodiment of a kind of coding/decoding method of the present invention.

The embodiment of the invention is used in the ultra broadband codec.The input signal of scrambler is the 32kHz sampling rate, and the processing frame length is 20ms.Input signal is carried out after branch frequency band, down-sampling handle, 320 sampled points are arranged, 320 sampled points are arranged corresponding to the signal of 8～16kHz frequency band corresponding to the signal of 0～8kHz frequency band.Wherein the signal in 0～8kHz frequency band is handled by core encoder, and core encoder can be a codec G.729.1, also can be other broadband signal codec.

For the time-domain signal in 8～16kHz frequency band y_hi (0), y_hi (1) ... y_hi (319) }, obtained one group of MDCT coefficient after the process MDCT conversion, i.e. the frequency-region signal of this frequency band correspondence { y_swb (0), y_swb (1),, y_swb (319) } because the ultra broadband part is only handled the signal in 8～14kHz frequency band, so frequency-region signal is only handled { y_swb (0), y_swb (1) ..., y_swb (239) }.This can be organized frequency-region signal at coding side and be divided into N subband, extract the frequency domain envelope and the quantification of each subband, obtain one group of frequency domain envelope { spec_env (0) after the quantification, spec_env (1),, spec_env (N-1) }, this group envelope arrives decoding end through coding transmission.

For the broadband core encoder G.729.1,7～8kHz part signal not within its process range, in order to ensure the continuity at decoding end decoded signal frequency spectrum, need extract the characteristic parameter of 7～8kHz part signal at coding side.Because G.729.1 scrambler has carried out the MDCT conversion to the signal of 4～8kHz, corresponding frequency-region signal y_wb (0), y_wb (1) ... y_wb (159) }, wherein the frequency-region signal of 7～8kHz correspondence be y_wb (120), y_wb (121) ... y_wb (159) }, at coding side it is divided into M subband, extracts the frequency domain envelope and the quantification of each subband, obtain one group of frequency domain envelope after the quantification

Spec_env_extra (0), spec_env_extra (1) ..., spec_env_extra (M-1) }, this group envelope arrives decoding end through coding transmission.N=15 in the present embodiment, M=3.

This frequency expansion method embodiment comprises coding method and coding/decoding method: extract the tonality feature parameter by coding side earlier, and send to decoding end, by the profile information reconstruct high band signal of decoding end according to tonality feature parameter and original high band signal.Specifically may further comprise the steps:

Step 101: coding side sends the tonality feature parameter of extraction to decoding end;

To the frequency-region signal of ultra broadband part spec (0), spec (1) ..., spec (239) } and carry out the tonality feature analysis, extract the tonality feature parameter, and described tonality feature parameter is carried out quantization encoding, coding transmission is arrived decoding end.

Concrete steps are as follows:

The autocorrelation function of 1a, calculating ultra broadband part frequency-region signal $\mathrm{Corr}\left[n\right]=\underset{i=0}{\overset{240-n}{\mathrm{\Σ}}}\mathrm{spec}\left(n\right)\×\mathrm{spec}(n+i),$ N ∈ [MIN, MAX], MIN=5MAX=70 in the present embodiment.

The tonality feature parameter of 1b, ultra broadband part is $\mathrm{tona}=\frac{\min \left(\mathrm{Corr}\right[n\left]\right)}{\max \left(\mathrm{Corr}\right[n\left]\right)}.$

1c, tonality feature parametric t ona is quantized the tonality feature parametric t ona after obtaining quantizing ^q, with its coding transmission to decoding end.

Step 102: the decoding end decoding obtains the tonality feature parameter;

Decoding end obtains tonality feature parametric t ona after the coding that receives is decoded ^q

Step 103: decoding end generates excitation spectrum according to the tonality feature parameter;

Specifically can may further comprise the steps:

Step 103A: decoding end generates the initial excitation spectrum;

Be specially narrow-band spectrum and the wideband spectrum generation initial excitation spectrum of decoding end according to decoding and reconstituting; Can add random noise when generating the initial excitation spectrum and obtain E={excit (0), excit (1) ..., excit (279) }.

Step 103B: decoding end can be carried out albefaction to the initial excitation spectrum, makes the energy distribution equilibrium of initial excitation spectrum;

After decoding end is extracted the spectrum envelope of initial excitation spectrum E, the initial excitation spectrum is carried out albefaction.Concrete steps are as follows:

A is divided into 7 subbands with initial excitation spectrum, calculates the average energy of each subband, i.e. spectrum envelope, ${\mathrm{env}}_i=\sqrt{\frac{\underset{j=i*40}{\overset{(i+1)*40}{\mathrm{\Σ}}}{\mathrm{excit}\left(j\right)}^2}{40}}.$

B has better continuity in order to make the initial excitation spectrum after the albefaction, can also carry out level and smooth interpolation to spectrum envelope, obtain one group of normalized factor β (0), β (1) ... β (279) }.The process of level and smooth interpolation is as follows:

$\mathrm{\β}\left(j\right)=\left\{\begin{array}{c}{\mathrm{env}}_0,j=\mathrm{0,1},\·\·\·,19\\ {\mathrm{env}}_6,j=\mathrm{260,261},\·\·\·279\\ w\×{\mathrm{env}}_i+(1-w)\×{\mathrm{env}}_{i+1},j=i\×40-20,\·\·\·,i\×40+20,i=\mathrm{1,2,3,4,5}\end{array}\right.,$ Wherein w is one 40 point symmetry window function.

C carries out albefaction to the initial excitation spectrum, and the initial excitation spectrum after the albefaction is:

E '=and excit ' (0), excit ' (1) ... excit ' (279) }, the albefaction process is: excit ' (i)=excit (i)/β (i).

103C: decoding end is carried out the tone shaping according to the original frequency band signal spectrum to the initial excitation spectrum and is obtained excitation spectrum; Need to prove that if the initial excitation spectrum has been passed through the albefaction operation of 103B, the initial excitation spectrum of this step input is the initial excitation spectrum after the albefaction.

Concrete steps are as follows:

The puppet that is calculated the initial excitation spectrum by E ' is composed SE={s_excit (0), s_excit (1) ..., s_excit (279) }

B adopts following method to calculate the tonality feature parameter T of the puppet spectrum SE of initial excitation spectrum.

Calculate the autocorrelation function of pseudo-spectrum SE frequency-region signal $\mathrm{Corr}\left[n\right]=\underset{i=0}{\overset{240-n}{\mathrm{\Σ}}}\mathrm{spec}\left(n\right)\×\mathrm{spec}(n+i),$ N ∈ [MIN, MAX], MIN=5MAX=70 in the present embodiment.

The tonality feature parameter of the puppet spectrum SE of initial excitation spectrum $T=\frac{\min \left(\mathrm{Corr}\right[n\left]\right)}{\max \left(\mathrm{Corr}\right[n\left]\right)}.$

C calculates tone shaping factor-alpha=β * (tona ^q-T), r1=1-α, r2=α/max (SE).In the present embodiment β=-1.9/0.6.

D utilizes tone shaping factor pair initial excitation spectrum to adjust.Be specially: if α＞0, i.e. tona ^q＜T, the tone characteristic of promptly original ultra broadband frequency spectrum is eager to excel than the tone characteristic of initial excitation spectrum, then the initial excitation spectrum is carried out the sharpening adjustment, be Y=E ' * (SE * r2+r1), obtain the excitation spectrum Y={y_excit (0) after the tone shaping, y_excit (1) ... y_excit (279) }.

If α≤0, i.e. tona ^q〉=T a little less than the tone characteristic of the tone characteristic of promptly original ultra broadband frequency spectrum than the initial excitation spectrum, then composes adaptive adding noise to initial excitation, smoothly adjusts, and adjustment process is as follows:

Generate random noise Noise={noise (0), noise (1) ..., noise (279) }.

Calculate the energy of pumping signal and noise signal, energyE '=∑ excit ' (i) and energyN=∑ noise (i).

According to following formula pumping signal is carried out the tone adjustment,

$Y=E^{\′}\×\frac{1-{\mathrm{tona}}^q}{1-T}+\mathrm{Noise}\×\frac{{\mathrm{tona}}^q-T}{1-T}\×\sqrt{\frac{\mathrm{energy}{E}^{\′}}{\mathrm{energy}{N}^{\′}}},$ Obtain the excitation spectrum Y={y_excit (0) after the tone shaping, y_excit (1) ... y_excit (279) }.

The tone characteristic of the tone characteristic of the excitation spectrum after the tone shaping and original ultra-broadband signal spectrum is mated more.

Step 104: decoding end can be carried out the frequency band adjustment to the excitation spectrum of final generation, to obtain the accurate more excitation spectrum of frequency band;

Concrete steps are as follows:

4a Y_extra={y_extra (0), y_extra (1) ..., y_extra (39) }, y_extra (i)=y_excit (279-i) is corresponding to the pumping signal of 7～8kHz.

4b Y_swb={y_swb (0), y_swb (1) ..., y_swb (239) }, y_swb (i)=y_excit (START+i) is corresponding to the pumping signal of 8～14kHz.Wherein START is pumping signal reference position parameter, START=30 in the present embodiment.

Step 105: decoding end generates the high band signal of reconstruct according to the profile information (as the frequency domain envelope information) in described excitation spectrum and the original high band signal.Promptly excitation spectrum is carried out the adjustment of energy level coupling and obtain the high band signal of reconstruct; Concrete steps are as follows:

5a obtains the frequency domain envelope (being the frequency domain energy in the present embodiment) of excitation spectrum, promptly according to the sub-band division identical with coding side, extracts the sub belt energy of excitation spectrum, and the excitation spectrum sub belt energy of 7～8kHz part is:

{ excit_energy_extra (0), excit_energy_extra (1), excit_energy_extra (2) }, the excitation spectrum sub belt energy of 8～14kHz part is { excit_energy (0), excit_energy (1) ..., excit_energy (14) }.The computation process of sub belt energy is: energy=∑ y_n (i) ², promptly the spectral coefficient in each subband is asked quadratic sum.

The 5b decoding obtains original high-band signal frequency-domain envelope information:

{ spec_env_extra (0), spec_env_extra (1), spec_env_extra (2) } and spec_env (0), spec_env (1) ..., spec_env (14) }.Because the frequency domain envelope has been represented the average energy of this subband, can obtain original high-band signal frequency-domain energy so the frequency domain envelope be multiply by the number of the spectral coefficient in this subband: { orig_energy_extra (0), orig_energy_extra (1), orig_energy_extra (2) } and { orig_energy (0), orig_energy (1),, orig_energy (14) }.

5c can be to the frequency domain energy and the original high-band signal frequency-domain energy of excitation spectrum, tonality feature parameter according to original high band signal is carried out smoothing processing, frequency domain energy { the excit_energy_extra_sm (0) of the excitation spectrum after obtaining smoothly, excit_energy_extra_sm (1), excit_energy_extra_sm (2) }, { excit_energy_sm (0), excit_energy_sm (1) ..., excit_energy_sm (14) } and original high-band signal frequency-domain energy

{orig_energy_extra_sm(0)，orig_energy_extra_sm(1)，orig_energy_extra_sm(2)}、{orig_energy_sm(0)，orig_energy_sm(1)，……，orig_energy_sm(14)}。

5d according to the frequency domain energy after level and smooth calculate the energy level coupling gain factor ψ of each subband=χ (0), χ (1) ..., χ (17) }, $\mathrm{\χ}\left(i\right)=\left\{\begin{array}{c}\sqrt{\frac{\mathrm{orig}\_\mathrm{energy}\_\mathrm{extra}\_\mathrm{sm}\left(i\right)}{\mathrm{excit}\_\mathrm{energy}\_\mathrm{extra}\_\mathrm{sm}\left(i\right)}},i=\mathrm{0,1,2}\\ \sqrt{\frac{\mathrm{orig}\_\mathrm{energy}\_\mathrm{sm}\left(i\right)}{\mathrm{excit}\_\mathrm{energy}\_\mathrm{sm}\left(i\right)}},\mathrm{otherwise}\end{array}\right..$ Certainly, if do not carry out smoothing processing, input be frequency domain energy among 5a, the 5b.

5e carries out the shaping of energy level coupling to excitation spectrum: the spectral coefficient to the excitation spectrum of each subband multiply by the corresponding energy level coupling of this subband gain factor, i.e. y_re (i)=y_n (i) * χ _iObtain the high-band excitation spectrum of reconstruct, comprise reconstruct spectrum { y_re_extra (0), the y_re_extra (1) of 7～8kHz part, y_re_extra (39) }, reconstruct spectrum { y_re (0), the y_re (1) of 8～14kHz part, y_re (239) }, the high band signal of promptly final reconstruct is finished the expansion of frequency band.

Decoding end is carried out the adjustment of energy level coupling to excitation spectrum, and the energy envelope of synthetic spectrum and original ultra broadband spectrum are close.

The embodiment of the invention adopts the key characterization parameter of extracting original high band signal at coding side: the tonality feature parameter, carry out coding transmission with few bit resource.In decoding end, generate the excitation spectrum of the high band signal of control reconstruct according to the tonality feature parameter, and finally generate the high band signal of reconstruct according to the profile information (for example frequency domain envelope) of original high band signal, realize band spread.Because when decoding end is carried out the reconstruct of high band signal, minutia based on original high band signal is that the tonality feature parameter is operated, the high band signal of reconstruct except profile and original high band signal near, coupling is preferably arranged in detail, just can reach auditory effect preferably by less bit resource consumption.

See also the signaling process figure of Fig. 2 for second embodiment of a kind of coding/decoding method of the present invention.

The embodiment of the invention is used in the ultra broadband codec.The input signal of scrambler is the 32kHz sampling rate, and the processing frame length is 20ms.Input signal is carried out after branch frequency band, down-sampling handle, 320 sampled points are arranged, 320 sampled points are arranged corresponding to the signal of 8～16kHz frequency band corresponding to the signal of 0～8kHz frequency band.Wherein the signal in 0～8kHz frequency band is handled by core encoder, and core encoder can be a codec G.729.1, also can be other broadband signal codec.

For the time-domain signal in 8～16kHz frequency band y_hi (0), y_hi (1) ... y_hi (319) }, obtained one group of MDCT coefficient after the process MDCT conversion, i.e. the frequency-region signal of this frequency band correspondence { y_swb (0), y_swb (1),, y_swb (319) } because the ultra broadband part is only handled the signal in 8～14kHz frequency band, so frequency-region signal is only handled { y_swb (0), y_swb (1) ..., y_swb (239) }.This can be organized frequency-region signal at coding side and be divided into N subband, extract the frequency domain envelope and the quantification of each subband, obtain one group of frequency domain envelope { spec_env (0) after the quantification, spec_env (1),, spec_env (N-1) }, this group envelope arrives decoding end through coding transmission.

For the broadband core encoder G.729.1,7～8kHz part signal not within its process range, in order to ensure the continuity at decoding end decoded signal frequency spectrum, need extract the characteristic parameter of 7～8kHz part signal at coding side.Because G.729.1 scrambler has carried out the MDCT conversion to the signal of 4～8kHz, corresponding frequency-region signal y_wb (0), y_wb (1) ... y_wb (159) }, wherein the frequency-region signal of 7～8kHz correspondence be y_wb (120), y_wb (121) ... y_wb (159) }, at coding side it is divided into M subband, extracts the frequency domain envelope and the quantification of each subband, obtain one group of frequency domain envelope after the quantification

Spec_env_extra (0), spec_env_extra (1) ..., spec_env_extra (M-1) }, this group envelope arrives decoding end through coding transmission.N=15 in the present embodiment, M=3.

This frequency expansion method embodiment comprises coding method and coding/decoding method: extract the tonality feature parameter by coding side earlier, and send to decoding end, by the profile information reconstruct high band signal of decoding end according to tonality feature parameter and original high band signal.Specifically may further comprise the steps:

Step 201: coding side sends the tonality feature parameter of extraction to decoding end;

To the frequency-region signal of ultra broadband part spec (0), spec (1) ..., spec (239) } and carry out the tonality feature analysis, extract the tonality feature parameter, and, be transferred to decoding end this parameter quantification coding.

Concrete leaching process is as follows:

(1) calculates the tonality feature parameter of each subband

A (1) with frequency-region signal spec (0), spec (1) ..., spec (239) } and be divided into K subband, in the present embodiment, and K=4, i group frequency-region signal is:

S _i＝{spec(i×60+0)，spec(i×60+1)，……，spec(i×60+59)}，i＝0，1，2，3。

Maximal value peak in every group of frequency-region signal of B (1) search _i=maxS _i, and record maximal value position p _i

C (1) calculates p _iThe average of the frequency domain envelope the in ± L scope, that is:

${\mathrm{mean}}_i=\underset{j=p_i-L}{\overset{p_i+L}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/2\×L+1,$ L=7 in the present embodiment.

The tonality feature parameter of this subband of D (1) is fluc _i=peak _i/ mean _i

(2) comprehensively extract G tonality feature parameter, the G=2 in the present embodiment that characterizes whole ultra broadband partial tone feature.Be specially ${\mathrm{tona}}_i=\left\{\begin{array}{c}({\mathrm{fluc}}_0+{\mathrm{fluc}}_1)/2,i=0\\ ({\mathrm{fluc}}_2+{\mathrm{fluc}}_3)/2,i=1\end{array}\right.,$ Tona wherein ₀Corresponding to spec (0), spec (1) ..., spec (119) } part tonality feature, tona ₁Corresponding to spec (120), spec (121) ..., spec (239) } part tonality feature.

(3) with tonality feature parametric t ona _iQuantize the tonality feature parametric t ona after obtaining quantizing _i ^q, with its coding transmission to decoding end.

Coding side is to tona _iDuring quantification, can be limited within 1.125～4.5 scopes its span.

When coding side extracted the tonality feature parameter, the parameter that uses can have multiple source, and concrete CALCULATION OF PARAMETERS also can have multiple computing method.For example:

A is the frequency-region signal { spec (0) of ultra broadband part wherein, spec (1),, spec (239) } can directly take absolute value to original frequency-region signal and obtain | y_swb (0) |, | y_swb (1) |, | y_swb (239) | }, also can be pseudo-spectrum signal { s (0), the s (1) that calculates through following formula (1) by original frequency-region signal,, s (239) }:

$s\left(i\right)=\left\{\begin{array}{c}\sqrt{y\_{\mathrm{swb}}^2\left(0\right)+y\_{\mathrm{swb}}^2\left(1\right)},i=0\\ \sqrt{y\_{\mathrm{swb}}^2\left(239\right)+y\_{\mathrm{swb}}^2\left(238\right)},i=239\\ \sqrt{y\_{\mathrm{swb}}^2\left(i\right)+(y\_\mathrm{swb}(i+1)-y\_\mathrm{swb}(i+1){)}^2},\mathrm{otherwise}\end{array}\right.---\left(1\right)$

B is in step (1), and the calculating of average also can be to calculate according to following formula ${\mathrm{mean}}_i=\underset{j=0}{\overset{59}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/60.$

In step (1), subband tonality feature CALCULATION OF PARAMETERS also can be to calculate according to following formula ${\mathrm{fluc}}_i=\sqrt[60]{\underset{j=0}{\overset{59}{\mathrm{\Π}}}\mathrm{spec}(i\×60+j)}/(\underset{j=0}{\overset{59}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/60).$

C in step (2), comprehensively also can the calculating of tonality feature parameter according to following formula ${\mathrm{tona}}_i=\left\{\begin{array}{c}\max ({\mathrm{fluc}}_0,{\mathrm{fluc}}_1),i=0\\ \max ({\mathrm{fluc}}_2,{\mathrm{fluc}}_3),i=1\end{array}\right..$

In step (2), comprehensively also can the calculating of tonality feature parameter according to following formula ${\mathrm{tona}}_i=\left\{\begin{array}{c}\sqrt{{\mathrm{fluc}}_0\×{\mathrm{fluc}}_1},i=0\\ \sqrt{{\mathrm{fluc}}_2\×{\mathrm{fluc}}_3},i=1\end{array}\right..$

Step 202: the decoding end decoding obtains the tonality feature parameter;

Decoding tonality feature parameter obtains tona ₀ ^qAnd tona ₁ ^qIf at coding side to tona _iDuring quantification, its span is limited within 1.125～4.5 scopes, can carries out nonlinear enhancing to it by following formula in decoding end, ${\mathrm{tona}}_i={\mathrm{tona}}_i^q+{\mathrm{tona}}_i^q\×{\mathrm{tona}}_i^q\×\mathrm{Tonality}\_\mathrm{TILT},$ Tonality_TILT=2.5/22 in the present embodiment.By the tonality feature nonlinearity in parameters is strengthened, can better control the synthetic of ultra broadband spectrum.Can certainly not carry out non-linear enhancing.

Step 203: decoding end generates excitation spectrum according to the tonality feature parameter; Specifically can may further comprise the steps:

Step 203A: decoding end generates the initial excitation spectrum;

Can generate excitation spectrum E={excit (0) in several ways in decoding end, excit (1) ..., excit (V-1) }, can adopt for example following mode to obtain excitation spectrum:

Time-domain signal in 0～4kHz frequency band of decoding end reconstruct is done the MDCT conversion, obtain frequency-region signal { y_lo (0), y_lo (1) in 0～4kHz frequency band,, y_lo (159) }, and the frequency-region signal { y_h (0) in the 4～7kHz frequency band that obtains of decoding, y_h (1) ..., y_h (119) }, with this two parts spectrum constitute original excitation spectrum y_lo (0), y_lo (1) ... y_lo (159), y_h (0), y_h (1),, y_h (119) }.

Owing to need reconstruct the frequency spectrum of 7～8kHz in decoding end, so in excitation spectrum, need to comprise the excitation information of 7～8kHz, V=280 in the present embodiment.

Step 203B is because excitation spectrum itself has the tone characteristic, for the tone characteristic of the tone characteristic that makes excitation spectrum and original ultra-broadband signal spectrum is complementary, can carry out the tone shaping to the initial excitation spectrum, and concrete steps are as follows:

A calculates the puppet spectrum of excitation spectrum according to formula (1):

SE＝{s_excit(0)，s_excit(1)，……，s_excit(279)}。

B adopts the tonality feature parameter T that obtains two subbands of pseudo-spectrum with coding side step (1), (2) identical method respectively ₀And T ₁

C is divided into two subbands with excitation spectrum and pseudo-spectrum,

E ₀＝{excit(0)，excit(1)，……，excit(159)}，E ₁＝{excit(160)，excit(161)，……，excit(279)}；

SE ₀＝{s_excit(0)，s_excit(1)，……，s_excit(159)}，

SE ₁＝{s_excit(160)，s_excit(161)，……，s_excit(279)}，

Calculate the tone energy parameter of pseudo-spectrum:

A _i=∑ SE _i/ (MAX _i* n _i), MAX wherein _i=maxSE _i, $n_i=\left\{\begin{array}{c}160,i=0\\ 120,i=1\end{array}\right.;$

D calculates the tone shaping factor: ${\mathrm{\α}}_i=\frac{T_i/{\mathrm{tona}}_i-1}{A_i\×T_i/{\mathrm{MAX}}_i-1}.$

C utilizes tone shaping factor pair initial excitation spectrum to adjust.Be specially: make r1 _i=1-α _i, r2 _i=α _i/ MAX _i, according to the tone characteristic T of excitation spectrum _iTone characteristic tona with original ultra broadband spectrum _iBetween relation excitation spectrum is adjusted, $Y_i=\left\{\begin{array}{c}{E}_i\×({\mathrm{SE}}_i\×{r2}_i+{r1}_i),T_i>{\mathrm{tona}}_i\\ E_i\×({\mathrm{SE}}_i\×{r2}_i+1),\mathrm{otherwise}\end{array}\right.,$ Obtain the excitation spectrum Y={Y after the tone shaping ₀, Y ₁}={ y_excit (0), y_excit (1) ... y_excit (279) }.

Step 203C: decoding end can be carried out albefaction to the excitation spectrum after the tone shaping, makes the energy distribution equilibrium of excitation spectrum, helps follow-up energy level coupling and adjusts;

The spectrum envelope that decoding end is extracted the excitation spectrum Y after the tone shaping carries out albefaction to excitation spectrum.Concrete steps are as follows:

A is divided into 7 subbands with excitation spectrum, calculates the average energy of each subband, i.e. spectrum envelope, $Y\_{\mathrm{env}}_i=\sqrt{\frac{\underset{j=i*40}{\overset{(i+1)*40}{\mathrm{\Σ}}}y\_\mathrm{excit}{\left(j\right)}^2}{40}}.$

B has better continuity in order to make the spectrum after the albefaction, and spectrum envelope is carried out level and smooth interpolation, obtain one group of normalized factor β (0), β (1) ... β (279) }.The process of level and smooth interpolation is as follows:

$\mathrm{\β}\left(j\right)=\left\{\begin{array}{c}Y\_{\mathrm{env}}_0,j=\mathrm{0,1},\·\·\·,19\\ Y\_{\mathrm{env}}_6,j=\mathrm{260,261},\·\·\·279\\ w\×{Y\_\mathrm{env}}_i+(1-w)\×{Y\_\mathrm{env}}_{i+1},j=i\×40-20,\·\·\·,i\×40+20,i=\mathrm{1,2,3,4,5}\end{array}\right.,$ Wherein w is one 40 point symmetry window function.

C carries out albefaction to excitation spectrum, and the excitation spectrum after the albefaction is:

Y′＝{y_excit′(0)，y_excit′(1)，…y_excit′(279)}，

The albefaction process is:

y_excit′(i)＝y_excit(i)/β(i)。

At this moment, just generated final excitation spectrum.Can certainly not carry out albefaction, directly use excitation spectrum after the tone shaping as final excitation spectrum.

Step 204: decoding end can be carried out the frequency band adjustment to the excitation spectrum of final generation, to obtain the accurate more excitation spectrum of frequency band;

Decoding end is carried out the frequency band adjustment to excitation spectrum and is adopted following formula,

4a Y_extra={y_extra (0), y_extra (1) ..., y_extra (39) }, y_extra (i)=y_excit ' (279-i), corresponding to the pumping signal of 7～8kHz.

4b Y_swb={y_swb (0), y_swb (1) ..., y_swb (239) }, y_swb (i)=y_excit ' (START+i), corresponding to the pumping signal of 8～14kHz.Wherein START is pumping signal reference position parameter, START=30 in the present embodiment.

Step 205: decoding end can add noise spectrum in excitation spectrum;

In order to make the frequency spectrum of excitation spectrum more approaching original ultra broadband on the spectrum details, under the tonality feature parameter control of original ultra-broadband signal, add noise spectrum, obtain adding the excitation spectrum Y_extra_n={y_extra_n (0) after making an uproar, y_extra_n (1) ..., y_extra_n (39) }, Y_swb_n={y_swb_n (0), y_swb_n (1) ..., y_swb_n (239) }.Detailed process is as follows:

5a with the tonality feature parameter maps of original ultra-broadband signal in 0～1 interval range, tona _i=tona _i/ (4.5+Tonality_TILT * 22).

5b calculating energy gain factor ener, in the present embodiment

$\mathrm{ener}=0.25+\frac{1-({\mathrm{tona}}_0+{\mathrm{tona}}_1)\×0.5}{6}.$

5c adds noise spectrum to the pumping signal of 7～8kHz: $y\_\mathrm{extra}\_n\left(i\right)=y\_\mathrm{extra}\left(i\right)\×{\mathrm{tona}}_0+\mathrm{ener}\×\sqrt{1-{\mathrm{tona}}_0^2}\×\mathrm{noise},$ Wherein noise is the random number within ± 0.5 scope, i.e. a noise spectrum.

5d adds noise spectrum to the pumping signal of 8～14kHz:

$y\_\mathrm{swb}\_n\left(i\right)=y\_\mathrm{seb}\left(i\right)\×\mathrm{tona}+\mathrm{ener}\×\sqrt{1-{\mathrm{tona}}^2}\×\mathrm{noise},$ Wherein noise is the random number within ± 0.5 scope, i.e. a noise spectrum; Tona is comprehensive tonality feature parameter, is provided with as follows:

$\mathrm{tona}=\left\{\begin{array}{c}{\mathrm{tona}}_0,i=\mathrm{0,1},\·\·\·95\\ {\mathrm{tona}}_1,i=\mathrm{144,145},\·\·\·239\\ {\mathrm{tona}}_0\×(1-\mathrm{\μ})+{\mathrm{tona}}_1\×\mathrm{\μ},\mathrm{\μ}=\frac{i-96}{47},\mathrm{othersie}\end{array}\right..$

Step 206: decoding end generates the high band signal of reconstruct according to the profile information (as the frequency domain envelope information) in described excitation spectrum and the original high band signal.Be that decoding end is carried out the high band signal that the adjustment of energy level coupling obtains reconstruct to excitation spectrum.

Excitation spectrum is carried out the energy level coupling adjust, the energy envelope of synthetic spectrum and original ultra broadband spectrum are close.Concrete steps are as follows:

6a obtains the frequency domain envelope (being the frequency domain energy in the present embodiment) of excitation spectrum, promptly according to the sub-band division identical with coding side, extract the sub belt energy of excitation spectrum, the excitation spectrum sub belt energy of 7～8kHz part is { excit_energy_extra (0), excit_energy_extra (1), excit_energy_extra (2) }, the excitation spectrum sub belt energy of 8～14kHz part is

{excit_energy(0)，excit_energy(1)，……，excit_energy(14)}。The computation process of sub belt energy is: energy=∑ y_n (i) ², promptly the spectral coefficient in each subband is asked quadratic sum.

The 6b decoding obtains original high-band signal frequency-domain envelope information:

{ spec_env_extra (0), spec_env_extra (1), spec_env_extra (2) } and spec_env (0), spec_env (1) ..., spec_env (14) }.Because the frequency domain envelope has been represented the average energy of this subband, can obtain original high-band signal frequency-domain energy so the frequency domain envelope be multiply by the number of the spectral coefficient in this subband: { orig_energy_extra (0), orig_energy_extra (1), orig_energy_extra (2) } and { orig_energy (0), orig_energy (1),, orig_energy (14) }.

6c can be to the frequency domain energy and the original high-band signal frequency-domain energy of excitation spectrum, tonality feature parameter according to original high band signal is carried out smoothing processing, frequency domain energy { the excit_energy_extra_sm (0) of the excitation spectrum after obtaining smoothly, excit_energy_extra_sm (1), excit_energy_extra_sm (2) }, { excit_energy_sm (0), excit_energy_sm (1) ..., excit_energy_sm (14) } and original high-band signal frequency-domain energy

{orig_energy_extra_sm(0)，orig_energy_extra_sm(1)，orig_energy_extra_sm(2)}、{orig_energy_sm(0)，orig_energy_sm(1)，……，orig_energy_sm(14)}。Concrete processing is as follows:

Frequency domain energy to 7～8kHz part: energy_extra_sm (i)=energy_extra (i) * γ, wherein $\mathrm{\γ}=\frac{e_0}{e_1},$ e ₀＝∑energy_extra(i)， $e_1=\mathrm{\Σ}({\mathrm{tona}}_0\×e_0\×\frac{N_i}{40}+(1-{\mathrm{tona}}_0)\×\mathrm{energy}\_\mathrm{extra}\left(i\right)),$ N _iBeing the number of the spectral coefficient in each subband, is 8,16,16 in the present embodiment.

Frequency domain energy to 8～14kHz part: energy_sm (i)=energy (i) * γ, wherein $\mathrm{\γ}=\frac{e_0}{e_1},$ e ₀＝∑energy(i)， $e_1=\mathrm{\Σ}(\frac{{\mathrm{tona}}_0+{\mathrm{tona}}_1}{2}\×{\mathrm{\η}}_i+(1-\frac{{\mathrm{tona}}_0+{\mathrm{tona}}_1}{2})\×\mathrm{energy}\left(i\right)),$ η _i＝0.15×energy(i-2)+0.2×energy(i-1)+0.3×energy(i)+0.2×energy(i+1)+0.15×energy(i+2)

6d calculates the energy level coupling gain factor of each subband according to the frequency domain energy after level and smooth

ψ＝{χ(0)，χ(1)，…，χ(17)}， $\mathrm{\χ}\left(i\right)=\left\{\begin{array}{c}\sqrt{\frac{\mathrm{orig}\_\mathrm{energy}\_\mathrm{extra}\_\mathrm{sm}\left(i\right)}{\mathrm{excit}\_\mathrm{energy}\_\mathrm{extra}\_\mathrm{sm}\left(i\right)}},i=\mathrm{0,1,2}\\ \sqrt{\frac{\mathrm{orig}\_\mathrm{energy}\_\mathrm{sm}\left(i\right)}{\mathrm{excit}\_\mathrm{energy}\_\mathrm{sm}\left(i\right)}},\mathrm{otherwise}\end{array}\right..$ Certainly, if do not carry out smoothing processing, input be frequency domain energy among 6a, the 6b.

6e carries out the shaping of energy level coupling to excitation spectrum: the spectral coefficient to the excitation spectrum of each subband multiply by the corresponding energy level coupling of this subband gain factor, i.e. y_re (i)=y_n (i) * x _iObtain the high-band excitation spectrum of reconstruct, comprise reconstruct spectrum { y_re_extra (0), the y_re_extra (1) of 7～8kHz part, y_re_extra (39) }, reconstruct spectrum { y_re (0), the y_re (1) of 8～14kHz part, y_re (239) }, the high band signal of promptly final reconstruct is finished the expansion of frequency band.

The embodiment of the invention adopts the key characterization parameter of extracting original high band signal at coding side: the tonality feature parameter, carry out coding transmission with few bit resource.In decoding end, generate the excitation spectrum of the high band signal of control reconstruct according to the tonality feature parameter, and finally generate the high band signal of reconstruct according to the profile information (for example frequency domain envelope) of original high band signal, realize band spread.Because when decoding end is carried out the reconstruct of high band signal, minutia based on original high band signal is that the tonality feature parameter is operated, the high band signal of reconstruct except profile and original high band signal near, coupling is preferably arranged in detail, just can reach auditory effect preferably by less bit resource consumption.

See also the structural drawing of Fig. 3 for an embodiment of a kind of coding method of the present invention;

Step 301: the frequency-region signal to the ultra broadband part carries out tonality feature analysis extraction tonality feature parameter;

Concrete steps are as follows:

301A with frequency-region signal spec (0), spec (1) ..., spec (239) } and be divided into K subband, K=4 in the present embodiment, i group frequency-region signal is:

S _i＝{spec(i×60+0)，spec(i×60+1)，……，spec(i×60+59)}，i＝0，1，2，3。

301B searches for the maximal value peak in each subband frequency-region signal _i=maxS _i, and record maximal value position p _i

301C calculates p _iThe average of the frequency domain envelope the in ± L scope, that is:

${\mathrm{mean}}_i=\underset{j=p_i-L}{\overset{p_i+L}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/2\×L+1,$ L=7 in the present embodiment.

The calculating of average also can be to calculate according to following formula ${\mathrm{mean}}_i=\underset{j=0}{\overset{59}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/60.$

The tonality feature parameter of this subband of 301D is fluc _i=peak _i/ mean _i

Subband tonality feature CALCULATION OF PARAMETERS also can be to calculate according to following formula

${\mathrm{fluc}}_i=\sqrt[60]{\underset{j=0}{\overset{59}{\mathrm{\Π}}}\mathrm{spec}(i\×60+j)}/(\underset{j=0}{\overset{59}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/60).$

Comprehensive each the subband tonality feature parameter of 301E obtains the tonality feature parameter.

Extract G tonality feature parameter, G=2 in the present embodiment characterizing whole ultra broadband partial tone feature.Be specially ${\mathrm{tona}}_i=\left\{\begin{array}{c}({\mathrm{fluc}}_0+{\mathrm{fluc}}_1)/2,i=0\\ ({\mathrm{fluc}}_2+{\mathrm{fluc}}_3)/2,i=1\end{array}\right.,$ Tona wherein ₀Corresponding to spec (0), spec (1) ..., spec (119) } part tonality feature, tona ₁Corresponding to spec (120), spec (121) ..., spec (239) } part tonality feature.

Comprehensively also can calculating of tonality feature parameter according to following formula ${\mathrm{tona}}_i=\left\{\begin{array}{c}\max ({\mathrm{fluc}}_0,{\mathrm{fluc}}_1),i=0\\ \max ({\mathrm{fluc}}_2,{\mathrm{fluc}}_3),i=1\end{array}\right..$

Step 301 can also adopt following manner to finish,

Calculate the autocorrelation function of ultra broadband part frequency-region signal;

Calculate the autocorrelation function of ultra broadband part frequency-region signal $\mathrm{Corr}\left[n\right]=\underset{i=0}{\overset{240-n}{\mathrm{\Σ}}}\mathrm{spec}\left(n\right)\×\mathrm{spec}(n+i),$ N ∈ [MIN, MAX], MIN=5MAX=70 in the present embodiment.

The tonality feature parameter is $\mathrm{tona}=\frac{\min \left(\mathrm{Corr}\right[n\left]\right)}{\max \left(\mathrm{Corr}\right[n\left]\right)}.$

Step 302: described tonality feature parameter is carried out quantization encoding.

See also the structural drawing of Fig. 4 for an embodiment of a kind of band spread of the present invention system;

Coding side 401 is used to send original high band signal and tonality feature parameter;

The frequency-region signal of 401 pairs of ultra broadband parts of coding side spec (0), spec (1) ..., spec (239) } and carry out the tonality feature analysis, extract the tonality feature parameter, and described pitch parameters is carried out quantization encoding, coding transmission is arrived decoding end.

Concrete steps are as follows:

The autocorrelation function of a, calculating ultra broadband part frequency-region signal $\mathrm{Corr}\left[n\right]=\underset{i=0}{\overset{240-n}{\mathrm{\Σ}}}\mathrm{spec}\left(n\right)\×\mathrm{spec}(n+i),$ N ∈ [MIN, MAX], MIN=5MAX=70 in the present embodiment.

The tonality feature parameter of b, ultra broadband part is $\mathrm{tona}=\frac{\min \left(\mathrm{Corr}\right[n\left]\right)}{\max \left(\mathrm{Corr}\right[n\left]\right)}.$

C, tonality feature parametric t ona is quantized the tonality feature parametric t ona after obtaining quantizing ^q, with its coding transmission to decoding end.

Decoding end 402 is used for tonality feature parameter and original high band signal that received code end 401 sends; Generate excitation spectrum according to described tonality feature parameter; Generate the high band signal of reconstruct according to the profile information in described excitation spectrum and the described original high band signal.

Decoding end 402 further can comprise:

Receiving element 40201 is used for tonality feature parameter and original high band signal that received code end 401 sends;

Excitation spectrum generation unit 40202 is used for the excitation spectrum according to the high band signal of described tonality feature parameter generation reconstruct;

High band signal generation unit 40203 is used for generating the high band signal of reconstruct according to the profile information of described excitation spectrum and described original high band signal;

Excitation spectrum generation unit 40202 further can comprise: excitation spectrum shaping unit 40202a, be used to generate the initial excitation spectrum, and described initial excitation is composed the excitation spectrum that carries out after shaping obtains the tone shaping.

Excitation spectrum generation unit 40202 further can comprise:

Albefaction unit 40202b is used for the excitation spectrum after described initial excitation spectrum or the tone shaping is carried out excitation spectrum after albefaction obtains albefaction.

Decoding end 402 further can comprise:

Frequency band adjustment unit 40204 is used for the excitation spectrum that generates is carried out the frequency band adjustment.

Noise adds unit 40205, is used for adding the noise spectrum that makes the more approaching original high-band signal spectrum structure of excitation spectrum at described excitation spectrum.

Described high band signal generation unit 40203 further can also comprise:

Spectrum envelope extraction unit 40203a is used to extract the spectrum envelope of excitation spectrum;

Smooth unit 40203b is used for the spectrum envelope and the original high-band signal decoding spectrum envelope of excitation spectrum are carried out the continuous frequency domain envelope of level and smooth interpolation generation;

Gain factor computing unit 40203c is used for calculating the gain factor that energy level mates according to described frequency domain envelope;

Generation unit 40203d is used for according to described gain factor excitation spectrum being carried out shaping, generates the high band signal of reconstruct.

See also the structural drawing of Fig. 5 for an implementation column of a kind of band spread decoding end of the present invention; Comprise:

Receiving element 501 is used for tonality feature parameter and original high band signal that the received code end sends;

Excitation spectrum generation unit 502 is used for the excitation spectrum according to the high band signal of described tonality feature parameter generation reconstruct;

High band signal generation unit 503 is used for generating reconstruct high-band signal frequency-domain information according to the profile information of described excitation spectrum and described original high band signal.

Described excitation spectrum generation unit 502 further comprises:

Excitation spectrum shaping unit 502a is used for according to described Pood's property the excitation spectrum that generates being carried out shaping.

Albefaction unit 502b is used for the excitation spectrum that generates is carried out albefaction.

Described decoding end can also comprise:

Frequency band adjustment unit 504 is used for excitation spectrum is carried out the frequency band adjustment.

Noise adds unit 505, is used for adding the noise spectrum that makes the more approaching original high-band signal spectrum structure of excitation spectrum at described excitation spectrum.

Described high band signal generation unit 403 further can comprise:

Spectrum envelope extraction unit 503a is used to extract the spectrum envelope of excitation spectrum;

Smooth unit 503b is used for the spectrum envelope and the original high-band signal decoding spectrum envelope of excitation spectrum are carried out the continuous frequency domain envelope of level and smooth interpolation generation;

Gain factor computing unit 503c is used for calculating the gain factor that energy level mates according to described frequency domain envelope;

Generation unit 503d is used for according to described gain factor excitation spectrum being carried out shaping, generates the high band signal of reconstruct.

See also the structural drawing of Fig. 6 for an embodiment of a kind of band spread coding side of the present invention; This coding side can be applied in the system shown in Figure 4, and it comprises:

Tonality feature parameter extraction unit 601 is used for the frequency-region signal of ultra broadband part is carried out tonality feature analysis extraction tonality feature parameter;

Concrete steps are as follows:

With frequency-region signal spec (0), spec (1) ..., spec (239) } and be divided into K subband, K=4 in the present embodiment, i group frequency-region signal is:

S _i＝{spec(i×60+0)，spec(i×60+1)，……，spec(i×60+59)}，i＝0，1，2，3。

Search for the maximal value peak in each subband frequency-region signal _i=maxS _i, and record maximal value position p _i

Calculate p _iThe average of the frequency domain envelope the in ± L scope, that is:

${\mathrm{mean}}_i=\underset{j=p_i-L}{\overset{p_i+L}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/2\×L+1,$ L=7 in the present embodiment.

The calculating of average also can be to calculate according to following formula ${\mathrm{mean}}_i=\underset{j=0}{\overset{59}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/60.$

The tonality feature parameter of this subband is fluc _i=peak _i/ mean _i

Subband tonality feature CALCULATION OF PARAMETERS also can be to calculate according to following formula $\mathrm{flu}{c}_i=\sqrt[60]{\underset{j=0}{\overset{59}{\mathrm{\Π}}}\mathrm{spec}(i\×60+j)}/(\underset{j=0}{\overset{59}{\mathrm{\Σ}}}\mathrm{spec}(i\×60+j)/60).$

Comprehensive each subband tonality feature parameter obtains the tonality feature parameter.

Extract G tonality feature parameter, G=2 in the present embodiment characterizing whole ultra broadband partial tone feature.Body is ${\mathrm{tona}}_i=\left\{\begin{array}{c}({\mathrm{fluc}}_0+{\mathrm{fluc}}_1)/2,i=0\\ ({\mathrm{fluc}}_2+{\mathrm{fluc}}_3)/2,i=1\end{array}\right.,$ Tona wherein ₀Corresponding to spec (0), spec (1) ..., spec (119) } part tonality feature, tona ₁Corresponding to spec (120), spec (121) ..., spec (239) } part tonality feature.

Comprehensively also can calculating of tonality feature parameter according to following formula ${\mathrm{tona}}_i=\left\{\begin{array}{c}\max ({\mathrm{fluc}}_0,{\mathrm{fluc}}_1),i=0\\ \max ({\mathrm{fluc}}_2,{\mathrm{fluc}}_3),i=1\end{array}\right..$

Can also adopt following manner to finish:

Calculate the autocorrelation function of ultra broadband part frequency-region signal;

Calculate the autocorrelation function of ultra broadband part frequency-region signal $\mathrm{Corr}\left[n\right]=\underset{i=0}{\overset{240-n}{\mathrm{\Σ}}}\mathrm{spec}\left(n\right)\×\mathrm{spec}(n+i),$ N ∈ [MIN, MAX], MIN=5MAX=70 in the present embodiment.

The tonality feature parameter is $\mathrm{tona}=\frac{\min \left(\mathrm{Corr}\right[n\left]\right)}{\max \left(\mathrm{Corr}\right[n\left]\right)}.$

Quantifying unit 602 is used for that described tonality feature parameter is carried out quantization encoding and sends described tonality feature parameter.

Wherein, tonality feature parameter extraction unit 601 can comprise:

Autocorrelation function unit 601a is used for obtaining the tonality feature parameter by the autocorrelation function that calculates ultra broadband part frequency-region signal.

Tonality feature parameter extraction unit 601 can comprise:

Computing unit 601b is used to calculate each subband frequency-region signal tonality feature parameter;

Extraction unit 601c comprehensively extracts the tonality feature parameter that characterizes whole ultra broadband partial tone feature.

One of ordinary skill in the art will appreciate that all or part of step in the whole bag of tricks of the foregoing description is to instruct relevant hardware to finish by program, this program can be stored in the computer-readable recording medium, and storage medium can comprise: ROM, RAM, disk or CD etc.

More than frequency expansion method that the embodiment of the invention provided and system and equipment are described in detail, used specific case herein principle of the present invention and embodiment are set forth, the explanation of above embodiment just is used for helping to understand method of the present invention and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to thought of the present invention, the part that all can change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.

Claims (20)

1, a kind of coding/decoding method is characterized in that, comprising:

Obtain tonality feature parameter and original high band signal;

Generate excitation spectrum according to described tonality feature parameter;

Generate the high band signal of reconstruct according to the profile information in described excitation spectrum and the described original high band signal.

2, coding/decoding method according to claim 1 is characterized in that, describedly generates excitation spectrum according to described tonality feature parameter and comprises: generate the initial excitation spectrum; Described initial excitation is composed the excitation spectrum that carries out after the tone shaping obtains the tone shaping.

3, coding/decoding method according to claim 2 is characterized in that, describedly described initial excitation spectrum is carried out the excitation spectrum that the tone shaping obtains after the tone shaping comprises:

Calculate the puppet spectrum of initial excitation spectrum;

Calculate the tonality feature parameter of the puppet spectrum of initial excitation spectrum;

The tonality feature calculation of parameter tone shaping factor according to described tonality feature parameter and described pseudo-spectrum;

Utilize described tone shaping factor pair initial excitation spectrum to adjust.

4, coding/decoding method according to claim 2 is characterized in that, describedly generates excitation spectrum according to described tonality feature parameter and further comprises: the excitation spectrum after described initial excitation spectrum or the tone shaping is carried out excitation spectrum after albefaction obtains albefaction.

5, according to claim 1,2 or 4 arbitrary described coding/decoding methods, it is characterized in that, described according to further comprising behind the described tonality feature parameter generation excitation spectrum: that the described excitation spectrum that generates is carried out the frequency band adjustment.

6, according to claim 1,2 or 4 arbitrary described coding/decoding methods, it is characterized in that, described according to further comprising behind the described tonality feature parameter generation excitation spectrum: as in described excitation spectrum, to add noise spectrum.

7, according to claim 1,2 or 4 arbitrary described coding/decoding methods, it is characterized in that, describedly generate reconstruct high-band signal according to the profile information in excitation spectrum and the original high band signal and comprise:

Obtain the frequency domain envelope and the original high-band signal frequency-domain envelope of described excitation spectrum;

Calculate the gain factor of energy level coupling according to described frequency domain envelope;

According to described gain factor described excitation spectrum is carried out shaping, generate the high band signal of reconstruct.

8, coding/decoding method according to claim 7 is characterized in that, describedly further comprises after obtaining described excitation spectrum and original high-band signal frequency-domain envelope:

Frequency domain envelope and described original high-band signal frequency-domain envelope to described excitation spectrum carry out smoothing processing, obtain the frequency domain envelope and the original high-band signal frequency-domain envelope of the excitation spectrum after level and smooth.

9, a kind of coding method is characterized in that:

Frequency-region signal to the ultra broadband part carries out tonality feature analysis extraction tonality feature parameter;

Described tonality feature parameter is carried out quantization encoding.

10, coding method according to claim 9, it is characterized in that described frequency-region signal to the ultra broadband part carries out tonality feature analysis extraction tonality feature parameter and comprises: obtain described tonality feature parameter by the autocorrelation function that calculates ultra broadband part frequency-region signal.

11, coding method according to claim 9 is characterized in that, described frequency-region signal to the ultra broadband part carries out tonality feature analysis extraction tonality feature parameter and comprises:

Calculate each subband frequency-region signal tonality feature parameter;

Comprehensively extract the tonality feature parameter that characterizes whole ultra broadband partial tone feature.

12, coding method according to claim 11 is characterized in that, described calculating subband frequency-region signal tonality feature parameter comprises:

Frequency-region signal is divided into a plurality of subband frequency-region signals;

Search for the maximal value and the position of each subband frequency-region signal;

Calculate the frequency domain envelope mean value in each subband frequency-region signal maximal value setting range;

The tonality feature parameter of subband frequency-region signal is the ratio of subband frequency-region signal maximal value and frequency domain envelope mean value.

13, a kind of band spread system is characterized in that, comprising:

Coding side is used to send original high band signal and tonality feature parameter;

Decoding end is used for tonality feature parameter and original high band signal that the received code end sends; Generate excitation spectrum according to described tonality feature parameter; Generate the high band signal of reconstruct according to the profile information in described excitation spectrum and the described original high band signal.

14, a kind of decoding end is characterized in that comprising:

Receiving element is used to receive tonality feature parameter and original high band signal;

The excitation spectrum generation unit is used for generating excitation spectrum according to described tonality feature parameter;

High band signal generation unit is used for generating the high band signal of reconstruct according to the profile information of described excitation spectrum and described original high band signal.

15, band spread decoding end according to claim 14 is characterized in that, described excitation spectrum generation unit comprises:

The excitation spectrum shaping unit is used to generate the initial excitation spectrum, described initial excitation is composed the excitation spectrum that carries out after the tone shaping obtains the tone shaping.

16, band spread decoding end according to claim 15 is characterized in that, the described excitation spectrum generation unit of separating comprises:

The albefaction unit is used for the excitation spectrum after described initial excitation spectrum or the tone shaping is carried out excitation spectrum after albefaction obtains albefaction.

17, according to claim 14,15 or 16 arbitrary described band spread decoding end, it is characterized in that described decoding end further comprises:

The frequency band adjustment unit is used for the described excitation spectrum that generates is carried out the frequency band adjustment.

18, according to claim 14 or 16 described band spread decoding end, it is characterized in that described decoding end further comprises:

Noise adds the unit, is used for adding noise spectrum at described excitation spectrum.

19, a kind of coding side is characterized in that, comprising:

The tonality feature parameter extraction unit is used for the frequency-region signal of ultra broadband part is carried out tonality feature analysis extraction tonality feature parameter;

Quantifying unit is used for described tonality feature parameter is carried out quantization encoding.

20, coding side according to claim 19 is characterized in that, described tonality feature parameter extraction unit comprises:

The autocorrelation function unit is used for obtaining described tonality feature parameter by the autocorrelation function that calculates ultra broadband part frequency-region signal.

21, coding side according to claim 19 is characterized in that, described tonality feature parameter extraction unit comprises:

Computing unit is used to calculate each subband frequency-region signal tonality feature parameter;

Extraction unit comprehensively extracts the tonality feature parameter that characterizes whole ultra broadband partial tone feature.

Images