CN103165134B - Coding and decoding device of audio signal high frequency parameter - Google Patents

Coding and decoding device of audio signal high frequency parameter Download PDF

Info

Publication number
CN103165134B
CN103165134B CN201310112757.5A CN201310112757A CN103165134B CN 103165134 B CN103165134 B CN 103165134B CN 201310112757 A CN201310112757 A CN 201310112757A CN 103165134 B CN103165134 B CN 103165134B
Authority
CN
China
Prior art keywords
frequency
module
high frequency
signal
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201310112757.5A
Other languages
Chinese (zh)
Other versions
CN103165134A (en
Inventor
胡瑞敏
杭波
李登实
李馨
涂卫平
王晓晨
李彩荣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University WHU
Original Assignee
Wuhan University WHU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN201310112757.5A priority Critical patent/CN103165134B/en
Publication of CN103165134A publication Critical patent/CN103165134A/en
Application granted granted Critical
Publication of CN103165134B publication Critical patent/CN103165134B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Abstract

The invention provides a coding and a decoding device of audio signal high frequency parameter. The coding and decoding device uses a plurality of modes to recover high frequency signals which can not pass a core layer code, thereby high frequency signals are rebuilt efficiently to improve output audio signal quality. The coding device of audio signal high frequency parameter comprises a frequency dividing module, a low frequency coding module, a time-frequency transforming module, a high frequency pumping module, a gain calculating module, a gain factor quantization module, a high frequency signal reconstructing module, a signal to noise ratio calculating module, a mode selecting module and a parameter coding module. The decoding device of audio signal high frequency parameter comprises a parameter allocation module, a low frequency decoding module, a decoding time-frequency transforming module, a high frequency pumping rebuilding module, an energy factor calculating module, a high frequency field signal rebuilding module and an inverse time-frequency transforming module.

Description

Sound signal high-frequency parameter coding and decoding device
Technical field
The invention belongs to Audiotechnica field, particularly relate to sound signal high-frequency parameter coding and decoding device.
Background technology
Show the research of human auditory system, people's ear is higher for low frequency signal susceptibility, and the patient quantization error of institute is less, and the susceptibility for high-frequency signal is then lower, and tolerable quantization error is larger.This characteristic is widely used in audio compression techniques.When limited code rate, available bits can be distributed to low frequency part and ignore HFS by audio coder, because the radio-frequency component in sound signal represents the sophistication of sound, lose radio-frequency component sound can be made to sound heavy vexed ﹑ does not become clear, the radio-frequency component thus in sound signal is the requisite important component of high quality audio that people pursue.But because the feature of the HFS of signal with different type is different, so the encoding and decoding technique applied to also is distinguished to some extent.If input signal band is wider than the frequency band range of core layer coding, signal frequency cannot be encoded by core layer higher than the high-frequency signal of core layer coding range.In order to efficiently high-frequency signal can be rebuild, be necessary research high-frequency parameter encoding and decoding technique.
Summary of the invention
The object of the invention is the high-frequency signal for existing signal with different type, sound signal high-frequency parameter under particular demands coding and decoding device is provided, make during digital audio decode, to utilize various modes to recover the high-frequency signal cannot encoded by core layer, thus efficiently can rebuild high-frequency signal.
Technical scheme provided by the invention is a kind of sound signal high-frequency parameter code device, comprises with lower part:
Frequency division module, for becoming low-frequency time-domain signal and high frequency time-domain signal by the original wideband signal frequency split of input;
Low frequency coding module, encodes for the low-frequency time-domain signal obtained according to distribution module, output low frequency parameter and low-frequency excitation time-domain signal, and wherein low-frequency parameter outputs to code stream;
Time-frequency conversion module, for making time-frequency conversion to high frequency time-domain signal, obtains high frequency frequency-region signal; Low-frequency excitation time-domain signal is made time-frequency conversion, obtains low-frequency excitation frequency-region signal; White noise time-domain signal is made time-frequency conversion, obtains white noise frequency-region signal;
High frequency pumping module, high frequency pumping frequency-region signal and corresponding incentive mode index is rebuild for obtaining four kinds respectively according to four kinds of mapping relations between the reconstruct mode preset and incentive mode index, comprise and the high frequency that folds in whole or in part of time-frequency conversion module gained low-frequency excitation frequency-region signal is obtained a kind of reconstruction high frequency pumping frequency-region signal, corresponding incentive mode index is I 0; The high frequency of moving in whole or in part of time-frequency conversion module gained low-frequency excitation frequency-region signal is obtained a kind of reconstruction high frequency pumping frequency-region signal, and corresponding incentive mode index is I 1; So that time-frequency conversion module gained white noise frequency-region signal is rebuild high frequency pumping frequency-region signal as one, corresponding incentive mode index is I 2; Select the optimum reconstructed high frequency frequency-region signal of module gained former frame to rebuild high frequency pumping frequency-region signal as the one of present frame in mode, corresponding incentive mode index is I 3;
Gain calculation module, for rebuilding high frequency pumping frequency-region signal and time-frequency conversion module gained high frequency frequency-region signal according to high frequency pumping module gained four kinds, uniformly or non-uniformly be divided into n subband respectively and calculate four groups of high-frequency gain factors, n is preset value;
Gain factor quantization modules, for according to gain calculation module gained four groups of high-frequency gain factors, obtains four groups and quantizes rear high-frequency gain Summing Factor four gain index;
High-frequency signal reconstructed module, rebuilds high frequency pumping frequency-region signal for quantizing rear high-frequency gain Summing Factor high frequency pumping module gained four kinds according to gain factor quantization modules gained, obtains four kinds of reconstructed high frequency frequency-region signals;
Snr computation module, for according to time-frequency conversion module gained high frequency frequency-region signal and high-frequency signal reconstructed module gained four kinds of reconstructed high frequency frequency-region signals, obtains four snr values;
Mode selection module, for according to snr computation module gained four snr values and high-frequency signal reconstructed module gained four kinds of reconstructed high frequency frequency-region signals, by the gain index in snr value corresponding to maximal value and incentive mode index corresponding to high frequency pumping module gained, obtain optimum gain index, optimal excitation mode index and optimum reconstructed high frequency frequency-region signal;
Parameter coding module, encodes for the optimum gain index that obtains according to mode selection module and optimal excitation mode index, and is input to code stream.
The present invention is also corresponding provides a kind of sound signal high-frequency parameter decoding device, comprises with lower part:
Parametric distribution module, for isolating optimum gain index and optimal excitation mode index in code stream;
Low frequency decoder module, carries out low frequency decoding for the low-frequency parameter will obtained in code stream, obtains rebuilding low-frequency excitation time-domain signal;
Decoding time-frequency conversion module, makes time-frequency conversion for low frequency decoder module gained is rebuild low-frequency excitation time-domain signal, obtains rebuilding low-frequency excitation frequency-region signal; White noise time-domain signal is made time-frequency conversion, obtains white noise frequency-region signal;
High frequency pumping rebuilds module, for according to parametric distribution module gained optimal excitation mode index, and obtains corresponding reconstruction high frequency pumping frequency-region signal, if comprising incentive mode index is I according to kind of the mapping relations of four between incentive mode index and reconstruct mode 0, decoding time-frequency conversion module gained reconstruction low-frequency excitation frequency-region signal is folded into high frequency and obtains rebuilding high frequency pumping frequency-region signal; If incentive mode index is I 1, decoding time-frequency conversion module gained reconstruction low-frequency excitation frequency-region signal is moved high frequency and obtains rebuilding high frequency pumping frequency-region signal; If incentive mode index is I 2, using time-frequency conversion module gained white noise frequency-region signal of decoding as reconstruction high frequency pumping frequency-region signal; If incentive mode index is I 3, the optimum rebuilding module gained former frame using high frequency frequency-region signal rebuilds the reconstruction high frequency pumping frequency-region signal of high frequency frequency-region signal as present frame;
Energy factors computing module, for obtaining energy factors according to parametric distribution module gained optimum gain index;
High frequency frequency-region signal rebuilds module, high frequency pumping frequency-region signal and energy factors computing module gained energy factors is rebuild for rebuilding module gained according to high frequency pumping, synthesis obtains rebuilding high frequency frequency-region signal, and this reconstruction high frequency frequency-region signal rebuilds high frequency frequency-region signal as the optimum of present frame;
Inverse time is conversion module frequently, does to obtain rebuilding high frequency time-domain signal against time-frequency conversion for rebuilding high frequency frequency-region signal according to high frequency pumping reconstruction module.
The invention provides a kind of sound signal high-frequency parameter coding and decoding scheme under particular demands, comprise coding side device and decoding end device.Calculate four kinds of reconstructed high frequency frequency-region signals and high frequency frequency-region signal snr value at coding side, get optimum gain index corresponding to maximal value wherein and corresponding optimal excitation mode index puts into code stream; Decoding end device can lead to extraction optimum gain index and optimal excitation mode index, and reconstruction high frequency pumping frequency-region signal corresponding to optimal excitation mode index and obtained rebuilding high frequency frequency-region signal by optimum gain index gained energy factors, remake inverse time-frequency conversion and obtain rebuilding high frequency time-domain signal.The feature that the present invention utilizes high-frequency signal exclusive, the high-frequency signal generated by more multiple reconstruct mode and the correlativity of high-frequency signal, and the mode selecting correlativity maximum obtains high-quality reconstruction high frequency time-domain signal.
Accompanying drawing explanation
Fig. 1 is the structured flowchart of embodiment of the present invention code device;
Fig. 2 is the structured flowchart of embodiment of the present invention decoding device.
Embodiment
See Fig. 1, the high-frequency parameter encoding and decoding technique scheme that the embodiment of the present invention provides comprises a kind of sound signal high-frequency parameter code device, comprises with lower part:
Frequency division module, for becoming low-frequency time-domain signal and high frequency time-domain signal by the original wideband signal frequency split of input;
Low frequency coding module, encodes for the low-frequency time-domain signal obtained according to distribution module, output low frequency parameter and low-frequency excitation time-domain signal, and wherein low-frequency parameter outputs to code stream; During concrete enforcement, low frequency coding module can adopt any existing low frequency encoder techniques to realize, and the high-frequency coding part designed with the present invention is combined; According to the concrete low frequency encoder techniques used, the low-frequency excitation time-domain signal of output is one of following four kinds,
1, non-quantized low-frequency time-domain signal, namely the low-frequency time-domain signal of frequency division module gained directly exports as low-frequency excitation time-domain signal by low frequency coding module;
2, the low frequency coding module output time-domain signal of low frequency scrambler realization arbitrarily, the low-frequency time-domain signal acquired results after low frequency coding module coding by frequency division module gained exports as low-frequency excitation time-domain signal, comprises one of three kinds of situations below:
2.1 rebuild low-frequency time-domain signal;
2.2 do not quantize low frequency residual error time-domain signal;
2.3 quantize rear low frequency residual error time-domain signal;
Time-frequency conversion module, for making time-frequency conversion to high frequency time-domain signal, obtains high frequency frequency-region signal; Low-frequency excitation time-domain signal is made time-frequency conversion, obtains low-frequency excitation frequency-region signal; White noise time-domain signal is made time-frequency conversion, obtains white noise frequency-region signal; High frequency pumping module, high frequency pumping frequency-region signal and corresponding incentive mode index is rebuild: the high frequency that folds in whole or in part of time-frequency conversion module gained low-frequency excitation frequency-region signal is obtained a kind of reconstruction high frequency pumping frequency-region signal, and corresponding incentive mode index is I for obtaining four kinds respectively according to four kinds of mapping relations between the reconstruct mode preset and incentive mode index 0; The high frequency of moving in whole or in part of time-frequency conversion module gained low-frequency excitation frequency-region signal is obtained a kind of reconstruction high frequency pumping frequency-region signal, and corresponding incentive mode index is I 1; So that time-frequency conversion module gained white noise frequency-region signal is rebuild high frequency pumping frequency-region signal as one, corresponding incentive mode index is I 2; Select the optimum reconstructed high frequency frequency-region signal of module gained former frame to rebuild high frequency pumping frequency-region signal as the one of present frame in mode, corresponding incentive mode index is I 3, (the first frame signal does not adopt this reconstruct mode); Gain calculation module, for rebuilding high frequency pumping frequency-region signal and time-frequency conversion module gained high frequency frequency-region signal according to high frequency pumping module gained four kinds, uniformly or non-uniformly be divided into n subband respectively and calculate four groups of high-frequency gain factors, n is preset value, can be specified in advance as the case may be during enforcement by those skilled in the art;
Gain factor quantization modules, for according to gain calculation module gained four groups of high-frequency gain factors, obtains four groups and quantizes rear high-frequency gain Summing Factor four gain index;
High-frequency signal reconstructed module, rebuilds high frequency pumping frequency-region signal for quantizing rear high-frequency gain Summing Factor high frequency pumping module gained four kinds according to gain factor quantization modules gained, obtains four kinds of reconstructed high frequency frequency-region signals;
Snr computation module, for according to time-frequency conversion module gained high frequency frequency-region signal and high-frequency signal reconstructed module gained four kinds of reconstructed high frequency frequency-region signals, obtains four snr values;
Mode selection module, for according to snr computation module gained four snr values and high-frequency signal reconstructed module gained four kinds of reconstructed high frequency frequency-region signals, by the gain index in snr value corresponding to maximal value and incentive mode index corresponding to high frequency pumping module gained, obtain optimum gain index, optimal excitation mode index and optimum reconstructed high frequency frequency-region signal, wherein the optimum reconstructed high frequency frequency-region signal of present frame will obtain a kind of foundation of rebuilding high frequency pumping frequency-region signal of next frame as high frequency pumping module; Parameter coding module, encodes for the optimum gain index that obtains according to mode selection module and optimal excitation mode index, and is input to code stream.
Wherein, the high frequency time-domain signal that frequency division module exports is input to time-frequency conversion module, and low-frequency time-domain signal is input to low frequency coding module; Low frequency coding module output low frequency parameter is to code stream; The low-frequency excitation time-domain signal that low frequency coding module exports can be low-frequency time-domain signal, or the low-frequency time-domain signal rebuild after low frequency coding module coding, or do not quantize low frequency residual error time-domain signal, or low frequency residual error time-domain signal after quantizing, and white noise time-domain signal is input to time-frequency conversion module; The low-frequency excitation frequency-region signal that time-frequency conversion module exports, the optimum reconstructed high frequency frequency-region signal of the former frame that white noise frequency-region signal and mode selection module export is input to high frequency pumping module; Four kinds of reconstruction high frequency pumping frequency-region signals that the high frequency frequency-region signal that time-frequency conversion module exports and high frequency pumping module export are input to gain calculation module; Four groups of high-frequency gain factors that gain calculation module exports are input to gain factor quantization modules; Gain factor quantization modules exports four groups and quantizes rear high-frequency gain Summing Factor four gain index, rebuilds high frequency pumping frequency-region signal and is input to high-frequency signal reconstructed module for four kinds that wherein after four groups of quantifications, high-frequency gain Summing Factor high frequency pumping module exports; Four kinds of reconstructed high frequency frequency-region signals that high-frequency signal reconstructed module exports and the high frequency frequency-region signal that time-frequency conversion module exports are input to snr computation module; Four snr values that snr computation module exports and high-frequency signal reconstructed module gained four kinds of reconstructed high frequency frequency-region signals, high frequency pumping module gained four kinds of incentive mode indexes, gain factor quantization modules gained gain index are input to mode selection module; Mode selection module exports the optimal excitation mode index of optimum gain index corresponding to maximum signal to noise ratio value and correspondence and optimum reconstructed high frequency frequency-region signal, the optimum gain index that mode selection module exports and optimal excitation mode index are input to parameter coding module, and parameter coding module is encoded to optimum gain index and optimal excitation mode index and is input to code stream for transmission.
As table 1
Table 1 reconstruct mode and incentive mode index corresponding relation
Gain calculation module, for rebuilding high frequency pumping frequency-region signal, according to time-frequency conversion module according to high frequency pumping module gained four kinds
Each module is implemented as prior art; during concrete enforcement; except providing sound signal high-frequency parameter code device in hardware, adopting software mode to realize following flow process by computing machine and belonging to equivalents, also should in application claims protection domain.
Coding work flow process comprises the following steps:
Step 1, obtains low-frequency time-domain signal and high frequency time-domain signal to the wideband audio signal frequency division of input;
Step 2, time-frequency convert is carried out to gained high frequency time-domain signal in step 1, obtain high frequency frequency-region signal, low frequency coding is carried out to gained low-frequency time-domain signal in step 1, obtain low-frequency parameter and low-frequency excitation time-domain signal, wherein low-frequency excitation time-domain signal can be non-quantized low-frequency time-domain signal, or rebuild low-frequency time-domain signal, do not quantize low frequency residual error time-domain signal after low frequency coding module coding, quantize after low frequency residual error time-domain signal;
Step 3, outputs to code stream by gained low-frequency parameter in step 2; To low-frequency excitation time-domain signal, and white noise time-domain signal carries out time-frequency conversion respectively and obtains low-frequency excitation frequency-region signal and white noise frequency-region signal;
Step 4, by gained low-frequency excitation frequency-region signal in step 3, white noise frequency-region signal and the last optimum reconstructed high frequency frequency-region signal performing step 9 gained former frame obtain four kinds by high frequency pumping and rebuild high frequency pumping frequency-region signal and corresponding incentive mode index, concrete operations are as follows: if obtained rebuilding high frequency pumping frequency-region signal by the high frequency that folds in whole or in part of low-frequency excitation frequency-region signal, corresponding incentive mode index is I 0; If obtained rebuilding high frequency pumping frequency-region signal by the high frequency of moving in whole or in part of low-frequency excitation frequency-region signal, corresponding incentive mode index is I 1; If using white noise frequency-region signal as reconstruction high frequency pumping frequency-region signal, corresponding incentive mode index is I 2; If using the optimum reconstructed high frequency frequency-region signal of former frame as the reconstruction high frequency pumping frequency-region signal of this frame, corresponding incentive mode index is I 3, the first frame signal does not adopt this reconstruct mode;
Step 5, rebuilds high frequency frequency domain pumping signal by gained four kinds in the high frequency frequency-region signal of gained in step 2 and step 4, calculates four groups of high-frequency gain factors by gain;
Step 6, is quantized by gain factor by the four groups of high-frequency gain factors of gained in step 5, obtains four groups and quantizes rear high-frequency gain Summing Factor four gain index;
Step 7, after being quantized by gained in step 6 four groups, in high-frequency gain Summing Factor step 4, gained four kinds is rebuild high frequency pumping frequency-region signal and is obtained four kinds of reconstructed high frequency frequency-region signals by high-frequency signal reconstruct;
Step 8, by gained four kinds of reconstructed high frequency frequency-region signals in gained high frequency frequency-region signal in step 2 and step 7, obtains four snr values by snr computation module;
Step 9, model selection is passed through by gained four kinds of reconstructed high frequency frequency-region signals in four snr values of gained in step 8 and step 7, select the maximal value in four snr values, obtain optimum gain index, optimal excitation mode index and optimum reconstructed high frequency frequency-region signal by the gain index wherein corresponding to maximal value and corresponding step 4 gained incentive mode index;
Step 10, carries out parameter coding by gained optimum gain index in step 9 and optimal excitation mode index, and is input in code stream.
See Fig. 2, the sound signal high-frequency parameter decoding device that the embodiment of the present invention provides, comprises with lower part:
Parametric distribution module, for isolating optimum gain index and optimal excitation mode index in code stream;
Low frequency decoder module, low-frequency parameter for obtaining in code stream carries out low frequency decoding, obtain rebuilding low-frequency excitation time-domain signal, the low-frequency excitation time-domain signal rebuilding low-frequency excitation time-domain signal and sound signal high-frequency parameter code device end is corresponding, when low-frequency excitation time-domain signal is selected do not quantize low-frequency time-domain signal or rebuild low-frequency time-domain signal, it can only be rebuild low-frequency time-domain signal after low frequency decoder module coding that decoding end rebuilds low-frequency excitation time-domain signal; When low-frequency excitation is not for quantizing low frequency residual error time-domain signal or quantizing rear low frequency residual error time-domain signal, decoding end correspondence can only be low frequency residual error time-domain signal after quantizing;
Decoding time-frequency conversion module, makes time-frequency conversion for low frequency decoder module gained is rebuild low-frequency excitation time-domain signal, obtains rebuilding low-frequency excitation frequency-region signal; White noise time-domain signal is made time-frequency conversion, obtains white noise frequency-region signal;
High frequency pumping rebuilds module, for according to parametric distribution module gained optimal excitation mode index, and obtains corresponding reconstruction high frequency pumping frequency-region signal according to kind of the mapping relations of four between incentive mode index and reconstruct mode: if incentive mode index is I 0, decoding time-frequency conversion module gained reconstruction low-frequency excitation frequency-region signal is folded into high frequency and obtains rebuilding high frequency pumping frequency-region signal; If incentive mode index is I 1, decoding time-frequency conversion module gained reconstruction low-frequency excitation frequency-region signal is moved high frequency and obtains rebuilding high frequency pumping frequency-region signal; If incentive mode index is I 2, using time-frequency conversion module gained white noise frequency-region signal of decoding as reconstruction high frequency pumping frequency-region signal; If incentive mode index is I 3, the optimum rebuilding module gained former frame using high frequency frequency-region signal rebuilds the reconstruction high frequency pumping frequency-region signal of high frequency frequency-region signal as present frame, and the first frame signal does not adopt this reconstruct mode;
Energy factors computing module, for obtaining energy factors according to parametric distribution module gained optimum gain index;
High frequency frequency-region signal rebuilds module, high frequency pumping frequency-region signal and energy factors computing module gained energy factors is rebuild for rebuilding module gained according to high frequency pumping, synthesis obtains rebuilding high frequency frequency-region signal, this reconstruction high frequency frequency-region signal rebuilds high frequency frequency-region signal as the optimum of present frame, obtains reconstruction high frequency pumping frequency-region signal corresponding to next frame by being used for high frequency pumping reconstruction module; Inverse time is conversion module frequently, does to obtain rebuilding high frequency time-domain signal against time-frequency conversion for rebuilding high frequency frequency-region signal according to high frequency pumping reconstruction module.
Wherein, code stream input parameter distribution module and low frequency decoder module; Low frequency decoder module gained is rebuild low-frequency excitation time-domain signal, can be the reconstruction low-frequency time-domain signal after low frequency decoder module coding, or low frequency residual error time-domain signal after quantizing, and white noise time-domain signal is input to decoding time-frequency conversion module; The optimal excitation mode index that parametric distribution module exports and the reconstruction low-frequency excitation frequency-region signal that time-frequency conversion module of decoding exports, the optimum of the former frame that white noise frequency-region signal and high frequency frequency-region signal reconstruction module export is rebuild high frequency frequency-region signal and is input to high frequency pumping reconstruction module; The optimum gain index that parametric distribution module exports is input to energy factors computing module; The energy factors of reconstruction high frequency pumping frequency-region signal and the output of energy factors computing module that high frequency pumping rebuilds module output is input to high frequency frequency-region signal and rebuilds module; The reconstruction high frequency frequency-region signal that high frequency frequency-region signal rebuilds module output is input to inverse time conversion module frequently, and high frequency time-domain signal is rebuild in inverse time conversion module output frequently.
Each module is implemented as prior art; during concrete enforcement; except providing sound signal high-frequency parameter decoding device in hardware, adopting software mode to realize following flow process by computing machine and belonging to equivalents, also should in application claims protection domain.
Decoding effort flow process comprises the following steps:
Step 1, from code stream, optimum gain index and optimal excitation mode index is isolated by parametric distribution, carry out low frequency decoding to the low-frequency parameter obtained in code stream to obtain rebuilding low-frequency excitation time-domain signal, can be rebuild low-frequency time-domain signal after low frequency decoder module coding, or low frequency residual error time-domain signal after quantizing;
Step 2, is input to step 1 gained reconstruction low-frequency excitation time-domain signal and white noise time-domain signal and carries out time-frequency conversion reconstruction low-frequency excitation frequency-region signal and white noise frequency-region signal respectively;
Step 3, by gained optimal excitation mode index in step 1, low-frequency excitation frequency-region signal is rebuild with gained in step 2, white noise frequency-region signal and the last optimum performing step 5 gained former frame are rebuild high frequency frequency-region signal and are rebuild by high frequency pumping, obtain corresponding reconstruction high frequency pumping frequency-region signal;
Step 4, calculates energy factors by step 1 gained optimum gain index by energy factors;
Step 5, rebuilds high frequency pumping frequency-region signal by gained energy factors in step 4 and step 3 gained and obtains rebuilding high frequency frequency-region signal;
Step 6, is obtained rebuilding high frequency time-domain signal by inverse time-frequency conversion by the reconstruction high frequency frequency-region signal obtained in step 5.
For the ease of implementing, the concrete implementation step that the invention provides embodiment describes in detail for reference.AVS-S input signal sampling rate is 16kHz, and signal bandwidth is 8kHz, and the frequency band range of AVS-S core layer coding is below 6.4kHz, and the high-frequency signal of 6.4-8kHz cannot be encoded by core layer.Described sound signal high-frequency parameter coding and decoding device is applied to the high-frequency signal coding of AVS-S, efficiently can rebuild the high-frequency signal of 6.4-8kHz.Details are as follows for specific implementation process:
(1) sound signal high-frequency parameter coding
At coding side, the wideband audio signal of input obtains 0 ~ 6.4kHz low-frequency time-domain signal and 6.4 ~ 12.8kHz high frequency time-domain signal by frequency division module, its medium-high frequency time-domain signal does 288 FFT conversion by time-frequency conversion module, by conversion after frequency band division be four subbands, the high frequency frequency-region signal only getting low 72 6.4 ~ 8kHz deals with object, is designated as HF_ori; By the low-frequency excitation time-domain signal of 4.8 ~ 6.4kH and white noise time-domain signal after low frequency coding module coding respectively by FFT time-frequency conversion, gained 4.8 ~ 6.4kH low-frequency excitation frequency-region signal and white noise frequency-region signal, and the optimum reconstructed high frequency frequency-region signal of 6.4 ~ 8kHz former frame of high frequency frequency-region signal reconstruction module output obtains by high frequency pumping module reconstruction high frequency pumping frequency-region signal HF_reb [the i] (i=0 that four kinds of frequency band ranges are 6.4 ~ 8kHz, 1,2,, and corresponding 2bit incentive mode index n 3) i(i=0,1,2,3).As table 2:
Reconstruct mode and incentive mode index corresponding relation in table 2 embodiment
Index coding 11,01,10,00 represents incentive mode index I respectively 0, I 1, I 2, I 3.
High frequency pumping frequency-region signal HF_reb [i] (i=0 will be rebuild respectively by gain calculation module, 1,2,3) and high frequency frequency-region signal HF_ori be divided into 8 subbands equably, the energy ratio of both calculating and high-frequency gain factor gain [i] (i=0,1,2,3), and the vector being divided into two groups of four-dimension is by gain factor quantization modules, obtain four groups and quantize rear high-frequency gain factor gain_q [i] (i=0,1,2,3) and four gain index g_index [i] (i=0,1,2,3);
Four groups that gain factor quantization modules are obtained quantize rear high-frequency gain factor gain_q [i] (i=0,1,2,3) high frequency pumping frequency-region signal HF_reb [i] (i=0,1,2 are rebuild for four kinds that obtain with high frequency pumping module respectively, 3) be multiplied by high-frequency signal reconstructed module and obtain four kinds of reconstructed high frequency frequency-region signal HF_reb_new [i] (i=0,1,2,3);
High frequency frequency-region signal HF_ori and four kind of reconstructed high frequency frequency-region signal HF_reb_new [i] (i=0 is calculated by snr computation module, 1,2,3) signal to noise ratio snr [i] (i=0,1,2,3) value, selects the optimum gain index g_index [i corresponding to maximal value by mode selection module 0] and corresponding optimal excitation mode index carry out parameter coding with 14bits and 2bits respectively, then reach bit stream multiplexer.I in first frame 0for I 0, I 1, I 2one of, i afterwards 0for I 0, I 1, I 2, I 3one of.
(2) sound signal high-frequency parameter decoding device
4.8 ~ 6.4kH after low frequency decoder module coding is rebuild low-frequency excitation time-domain signal, white noise time-domain signal is by FFT time-frequency conversion, obtain 4.8 ~ 6.4kH and rebuild low-frequency excitation frequency-region signal, the optimum that white noise frequency-region signal and high frequency frequency-region signal rebuild 6.4 ~ 8kHz former frame that module exports rebuilds high frequency frequency-region signal.From high frequency code stream, optimum gain index g_index [i is isolated by parametric distribution module 0] and optimal excitation mode index by optimal excitation mode index four kinds of mapping relations (as table 3) between the incentive mode index provided by high frequency pumping reconstruction module and reconstruct mode, obtain corresponding reconstruction high frequency pumping frequency-region signal HF_reb [i 0].I in first frame 0for I 0, I 1, I 2one of, i afterwards 0for I 0, I 1, I 2, I 3one of.
The corresponding relation of incentive mode index and reconstruct mode in table 3 embodiment
The optimum gain index extracted by parametric distribution module energy factors gain [i is obtained by energy factors computing module 0], high frequency pumping frequency-region signal HF_reb [i will be rebuild 0] and energy factors gain [i 0] obtain rebuilding high frequency frequency-region signal HF_new_f by high frequency frequency-region signal reconstruction module; Obtain rebuilding high frequency time-domain signal HF_new by inverse FFT conversion module.
Specific embodiment described herein is only to the explanation for example of the present invention's spirit.Those skilled in the art can make various amendment or supplement or adopt similar mode to substitute to described specific embodiment, but can't depart from spirit of the present invention or surmount the scope that appended claims defines.

Claims (2)

1. a sound signal high-frequency parameter code device, is characterized in that: comprise with lower part,
Frequency division module, for becoming low-frequency time-domain signal and high frequency time-domain signal by the original wideband signal frequency split of input;
Low frequency coding module, encodes for the low-frequency time-domain signal obtained according to distribution module, output low frequency parameter and low-frequency excitation time-domain signal, and wherein low-frequency parameter outputs to code stream;
Time-frequency conversion module, for making time-frequency conversion to high frequency time-domain signal, obtains high frequency frequency-region signal; Low-frequency excitation time-domain signal is made time-frequency conversion, obtains low-frequency excitation frequency-region signal; White noise time-domain signal is made time-frequency conversion, obtains white noise frequency-region signal;
High frequency pumping module, high frequency pumping frequency-region signal and corresponding incentive mode index is rebuild for obtaining four kinds respectively according to four kinds of mapping relations between the reconstruct mode preset and incentive mode index, comprise and the high frequency that folds in whole or in part of time-frequency conversion module gained low-frequency excitation frequency-region signal is obtained a kind of reconstruction high frequency pumping frequency-region signal, corresponding incentive mode index is I 0; The high frequency of moving in whole or in part of time-frequency conversion module gained low-frequency excitation frequency-region signal is obtained a kind of reconstruction high frequency pumping frequency-region signal, and corresponding incentive mode index is I 1; So that time-frequency conversion module gained white noise frequency-region signal is rebuild high frequency pumping frequency-region signal as one, corresponding incentive mode index is I 2; Select the optimum reconstructed high frequency frequency-region signal of module gained former frame to rebuild high frequency pumping frequency-region signal as the one of present frame in mode, corresponding incentive mode index is I 3;
Gain calculation module, for rebuilding high frequency pumping frequency-region signal and time-frequency conversion module gained high frequency frequency-region signal according to high frequency pumping module gained four kinds, be divided into n subband respectively and calculate four groups of high-frequency gain factors, n is preset value;
Gain factor quantization modules, for according to gain calculation module gained four groups of high-frequency gain factors, obtains four groups and quantizes rear high-frequency gain Summing Factor four gain index;
High-frequency signal reconstructed module, rebuilds high frequency pumping frequency-region signal for quantizing rear high-frequency gain Summing Factor high frequency pumping module gained four kinds according to gain factor quantization modules gained, obtains four kinds of reconstructed high frequency frequency-region signals;
Snr computation module, for according to time-frequency conversion module gained high frequency frequency-region signal and high-frequency signal reconstructed module gained four kinds of reconstructed high frequency frequency-region signals, obtains four snr values;
Mode selection module, for according to snr computation module gained four snr values and high-frequency signal reconstructed module gained four kinds of reconstructed high frequency frequency-region signals, by the gain index in snr value corresponding to maximal value and incentive mode index corresponding to high frequency pumping module gained, obtain optimum gain index, optimal excitation mode index and optimum reconstructed high frequency frequency-region signal;
Parameter coding module, encodes for the optimum gain index that obtains according to mode selection module and optimal excitation mode index, and is input to code stream.
2. a sound signal high-frequency parameter decoding device, is characterized in that: comprise with lower part,
Parametric distribution module, for isolating optimum gain index and optimal excitation mode index in code stream;
Low frequency decoder module, carries out low frequency decoding for the low-frequency parameter will obtained in code stream, obtains rebuilding low-frequency excitation time-domain signal;
Decoding time-frequency conversion module, makes time-frequency conversion for low frequency decoder module gained is rebuild low-frequency excitation time-domain signal, obtains rebuilding low-frequency excitation frequency-region signal; White noise time-domain signal is made time-frequency conversion, obtains white noise frequency-region signal;
High frequency pumping rebuilds module, for according to parametric distribution module gained optimal excitation mode index, and obtains corresponding reconstruction high frequency pumping frequency-region signal, if comprising incentive mode index is I according to kind of the mapping relations of four between incentive mode index and reconstruct mode 0, decoding time-frequency conversion module gained reconstruction low-frequency excitation frequency-region signal is folded into high frequency and obtains rebuilding high frequency pumping frequency-region signal; If incentive mode index is I 1, decoding time-frequency conversion module gained reconstruction low-frequency excitation frequency-region signal is moved high frequency and obtains rebuilding high frequency pumping frequency-region signal; If incentive mode index is I 2, using time-frequency conversion module gained white noise frequency-region signal of decoding as reconstruction high frequency pumping frequency-region signal; If incentive mode index is I 3, the optimum rebuilding module gained former frame using high frequency frequency-region signal rebuilds the reconstruction high frequency pumping frequency-region signal of high frequency frequency-region signal as present frame;
Energy factors computing module, for obtaining energy factors according to parametric distribution module gained optimum gain index;
High frequency frequency-region signal rebuilds module, high frequency pumping frequency-region signal and energy factors computing module gained energy factors is rebuild for rebuilding module gained according to high frequency pumping, synthesis obtains rebuilding high frequency frequency-region signal, and this reconstruction high frequency frequency-region signal rebuilds high frequency frequency-region signal as the optimum of present frame;
Inverse time is conversion module frequently, does to obtain rebuilding high frequency time-domain signal against time-frequency conversion for rebuilding high frequency frequency-region signal according to high frequency pumping reconstruction module.
CN201310112757.5A 2013-04-02 2013-04-02 Coding and decoding device of audio signal high frequency parameter Active CN103165134B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310112757.5A CN103165134B (en) 2013-04-02 2013-04-02 Coding and decoding device of audio signal high frequency parameter

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310112757.5A CN103165134B (en) 2013-04-02 2013-04-02 Coding and decoding device of audio signal high frequency parameter

Publications (2)

Publication Number Publication Date
CN103165134A CN103165134A (en) 2013-06-19
CN103165134B true CN103165134B (en) 2015-01-14

Family

ID=48588158

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310112757.5A Active CN103165134B (en) 2013-04-02 2013-04-02 Coding and decoding device of audio signal high frequency parameter

Country Status (1)

Country Link
CN (1) CN103165134B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104517611B (en) 2013-09-26 2016-05-25 华为技术有限公司 A kind of high-frequency excitation signal Forecasting Methodology and device
CN103714822B (en) * 2013-12-27 2017-01-11 广州华多网络科技有限公司 Sub-band coding and decoding method and device based on SILK coder decoder
CN106023999B (en) * 2016-07-11 2019-06-11 武汉大学 For improving the decoding method and system of three-dimensional audio spatial parameter compression ratio
CN108417219B (en) * 2018-02-22 2020-10-13 武汉大学 Audio object coding and decoding method suitable for streaming media

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5987406A (en) * 1997-04-07 1999-11-16 Universite De Sherbrooke Instability eradication for analysis-by-synthesis speech codecs
CN101089951A (en) * 2006-06-16 2007-12-19 徐光锁 Band spreading coding method and device and decode method and device
CN101192409A (en) * 2006-11-21 2008-06-04 华为技术有限公司 Method and device for selecting self-adapting codebook excitation signal
CN101458930A (en) * 2007-12-12 2009-06-17 华为技术有限公司 Excitation signal generation in bandwidth spreading and signal reconstruction method and apparatus

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5987406A (en) * 1997-04-07 1999-11-16 Universite De Sherbrooke Instability eradication for analysis-by-synthesis speech codecs
CN101089951A (en) * 2006-06-16 2007-12-19 徐光锁 Band spreading coding method and device and decode method and device
CN101192409A (en) * 2006-11-21 2008-06-04 华为技术有限公司 Method and device for selecting self-adapting codebook excitation signal
CN101458930A (en) * 2007-12-12 2009-06-17 华为技术有限公司 Excitation signal generation in bandwidth spreading and signal reconstruction method and apparatus

Also Published As

Publication number Publication date
CN103165134A (en) 2013-06-19

Similar Documents

Publication Publication Date Title
JP6321072B2 (en) Audio / voice encoding method and audio / voice decoding method
US20190122680A1 (en) Adaptive transition frequency between noise fill and bandwidth extension
JP6019266B2 (en) Stereo audio encoder and decoder
CN102194457B (en) Audio encoding and decoding method, system and noise level estimation method
JP2013015598A (en) Audio coding/decoding method, system and noise level estimation method
WO2009029036A1 (en) Method and device for noise filling
CN103165134B (en) Coding and decoding device of audio signal high frequency parameter
JP6148811B2 (en) Low frequency emphasis for LPC coding in frequency domain
CN102194458B (en) Spectral band replication method and device and audio decoding method and system
CN104978970A (en) Noise signal processing and generation method, encoder/decoder and encoding/decoding system
CN101281748B (en) Method for filling opening son (sub) tape using encoding index as well as method for generating encoding index
US20130006644A1 (en) Method and device for spectral band replication, and method and system for audio decoding
JP5308519B2 (en) Multi-mode scheme for improved audio coding
RU2633097C2 (en) Methods and devices for signal coding and decoding
CN105957533B (en) Voice compression method, voice decompression method, audio encoder and audio decoder
CN102074243B (en) Bit plane based perceptual audio hierarchical coding system and method
KR101786863B1 (en) Frequency band table design for high frequency reconstruction algorithms
CN104380377B (en) For the method and apparatus of scalable low encoding complexity/decoding
KR102353050B1 (en) Signal reconstruction method and device in stereo signal encoding
CN105261373A (en) Self-adaptive grid construction method and device used for bandwidth extended coding
CN105336334B (en) Multi-channel sound signal coding method, decoding method and device
JP2013019968A (en) Frequency band copy method, device and audio decoding method, and system
CN108630212B (en) Perception reconstruction method and device for high-frequency excitation signal in non-blind bandwidth extension
Hansen et al. Fine-grain scalable audio coding based on envelope restoration and the SPIHT algorithm
CN102479514A (en) Coding method, decoding method, apparatus and system thereof

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant