CN104392725A - Method and device for hybrid coding/decoding of multi-channel lossless audios - Google Patents

Method and device for hybrid coding/decoding of multi-channel lossless audios Download PDF

Info

Publication number
CN104392725A
CN104392725A CN 201410721299 CN201410721299A CN104392725A CN 104392725 A CN104392725 A CN 104392725A CN 201410721299 CN201410721299 CN 201410721299 CN 201410721299 A CN201410721299 A CN 201410721299A CN 104392725 A CN104392725 A CN 104392725A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
coding
rice
golomb
entropy
module
Prior art date
Application number
CN 201410721299
Other languages
Chinese (zh)
Inventor
杨新辉
刘任化
Original Assignee
中科开元信息技术(北京)有限公司
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

Links

Abstract

The invention discloses a method and a device for hybrid coding/decoding of multi-channel lossless audios. In the method and the device, the coding/encoding of an algorithm and a Golomb-rice entropy is chosen according to sampling frequency, quantity of the channels, digitalization bits and computing complexity of a pre-treatment module and a predication module; and the hybrid coding/decoding of the algorithm and the Golomb-rice for multi-channel lossless audio signals is realized by the respective coding/decoding advantages of the algorithm and the Golomb-rice entropy. For the consideration of the balance between the entropy coding complexity and the coding/decoding efficiency, a hybrid entropy coding method applicable to the multi-channel lossless audio data is provided; the switching between the coding/decoding complexity and the coding/decoding speed can be realized by using the method; and wider application requirements can be satisfied according to the difference of the operational capabilities of processors of a coders and an encoder. The high-quality coding for the multi-channel lossless audios can be realized and the hardware cost is reduced effectively.

Description

多声道无损音频混合编解码方法及装置 Multi-channel lossless audio encoding and decoding method and apparatus for mixing

技术领域 FIELD

[0001] 本发明涉及数字信号处理技术,尤其是涉及数字音频信源编解码方法及装置。 [0001] The present invention relates to digital signal processing technology, and more particularly to a digital audio source codec method and apparatus.

背景技术 Background technique

[0002] 随着半导体技术,宽带网络技术和大容量存储技术的发展,人们对多声道,高采样频率的无损音频信号的需求越来越强烈。 [0002] With the development of semiconductor technology, broadband network technologies and mass storage technology, the demand for multi-channel lossless audio signal with a high sampling frequency is more and more intense. 但因为声道数量的增加,采用频率的提高,也带来了无损音频编解码计算复杂度的倍增。 However, because of the increased number of channels, using the frequency increases, also it brings Lossless Audio Codec computational complexity of the multiplication. 影响编解码计算复杂度的主要因素有:声道数量,采用频率,量化位数,编解码算法复杂度等;其中声道数量,采用频率,量化位数等因素是客观输入条件,编解码算法复杂度是可调节因素。 The main factors affecting the computational complexity of the encoding and decoding are: the number of channels, with a frequency, quantization bit number, the complexity of the encoding and decoding algorithms and the like; wherein the number of channels, with a frequency, quantization bit number is the objective input conditions and other factors, codec algorithm complexity is adjustable factor. 编解码算法大体可分为:预处理算法,预测算法,熵编码。 Codec algorithm can be divided into: preprocessing algorithm, prediction algorithms, entropy coding.

[0003] 多声道数字音频编码领域,为了去除数据冗余,需要使用熵编码算法实现数据压缩,主要的熵编码方法有算术编码,哈夫曼编码,Golomb-Rice编码,行程编码等。 [0003] The multi-channel digital audio coding, in order to remove redundant data, an entropy encoding algorithm for data compression required, the main coding method as arithmetic coding, Huffman coding, Golomb-Rice coding, run-length coding and the like. 这些熵编码算法适用于不同统计特性的数据源的压缩,而且时间与空间复杂度也有较大的差别。 The entropy coding algorithm suitable for compressing different statistical properties of the data source, and the time and space complexity is also a greater difference. 哈夫曼编码较难把概率模型建模与编码分开,在压缩率方面,扩展的哈夫曼编码能接近信源的熵,但其空间复杂度和时间复杂度都比较大;与哈夫曼编码比,算术编码较容易把概率模型建模与编码分开,它是对整个信源的编码,其压缩率也能接近信源的熵,但概率表的刷新比较耗时,时间复杂度比较大;Golomb-rice依据的是几何分布的概率模型对数据进行压缩,压缩率稍差,但其时间与空间复杂度较低;而行程编码只适用于连续重复的数据压缩, 缺乏通用性。 Huffman coding is more difficult to separate the probabilistic modeling and coding, the compression ratio, Huffman encoding can be extended closer to the entropy of the source, but the spatial complexity and temporal complexity is relatively large; and Huffman than coding, arithmetic coding easier to separate than the probability modeling and coding, it is the encoding of the entire source, the compression ratio can be close to the source of entropy, but the probability table refresh time-consuming, time complexity is relatively large ; Golomb-rice probability model based on the geometric distribution of the data compression, the compression ratio somewhat less, but the complexity of its low time and space; run-length coding is only applicable to the continuously repeated compression data, it lacks versatility.

发明内容 SUMMARY

[0004] 本发明的目的是从计算复杂度和压缩率两个角度考虑,提出适用范围更广的混合熵编码方法。 [0004] The object of the present invention is from computational complexity and compression two viewpoints, a broader scope of application proposed hybrid entropy coding method.

[0005] 本发明提出的基于计算复杂度分析的混合熵编解码方法,根据采样频率,声道数, 量化位数,预处理模块,预测模块的运算复杂度选择算术与Golomb-rice熵编解码,利用了算术与Golomb-rice熵编解码各自的优点,实现对多声道音频信号的编解码。 [0005] Entropy encoding and decoding method based on a mixed computational complexity of the analysis provided by the invention, according to the sampling frequency, the computational complexity of the number of channels, the quantization bit number, pre-processing module, selecting the prediction module and Golomb-rice arithmetic entropy encoding and decoding using the Golomb-rice arithmetic entropy encoding and decoding their own advantages to achieve encoding and decoding multi-channel audio signal.

[0006] 为了实现本发明的目的,提出一种多声道无损音频混合编解码方法,所述方法根据采样频率,声道数,量化位数,预处理模块及预测模块的运算复杂度选择算术与Golomb-rice熵编解码,利用算术与Golomb-rice熵编解码各自的优点,实现对多声道音频信号的算术与Golomb-rice的混合编解码。 [0006] To achieve the object of the present invention to provide a multi-channel lossless audio hybrid codec, the method according to the sampling frequency, number of channels, number of quantization bits, the computational complexity of the preprocessing module and the prediction module selection arithmetic Golomb-rice and entropy encoding and decoding, the Golomb-rice using an arithmetic entropy encoding and decoding their own advantages to achieve hybrid codec arithmetic and multi-channel audio signal of Golomb-rice.

[0007] 所述算术与Golomb-rice的混合熵编码方法包括具体步骤: [0007] The arithmetic coding method with hybrid Golomb-rice include the particular steps of:

[0008] 对原始音频输入信号进行预处理,减少信号之间的冗余度,提高压缩率; [0008] The pretreatment of the original audio input signal to reduce the redundancy between the signals, increase the compression ratio;

[0009] 对音频输入信号进行预测,去除原始信号前后数据的相关性; [0009] The audio input signal prediction, remove the correlation before and after the raw data signal;

[0010] 根据音频数据采样频率,声道数,预测模块复杂度参数,预处理模块复杂度参数, 音频数据特征参数组合信息,设置选择算术熵编码和Golomb-Rice熵编码的阈值; [0010] The audio data sampling frequency complexity, number of channels, the prediction module complexity parameter, the preprocessing module parameters, characteristic parameters of the audio combination information, select the set of arithmetic entropy encoding and entropy encoding the Golomb-Rice threshold;

[0011] 所述算术与Golomb-rice的混合熵解码方法包括具体步骤: [0011] The arithmetic and hybrid Golomb-rice specific entropy decoding method comprising the steps of:

[0012] 进行预测逆处理; [0012] The inverse prediction process;

[0013] 进行预处理逆处理; [0013] Inverse pretreatment process;

[0014] 从码流中读取熵编码算法标识,以决定调用算术还是Golomb-rice熵解码器,解码当前声道或者帧数据。 [0014] The entropy coding algorithm identifier read from the code stream to determine or call the Golomb-rice arithmetic entropy decoder, or decoding the current frame data channel.

[0015] 所述方法允许算术编码和Golomb-Rice熵编码在数据帧中独立存在或混合存在, 艮P,码流内全部数据采用算术编码方法、码流内全部数据采用Golomb-Rice编码方法或码流内数据采用算术编码和Golomb-Rice编码并存的方法。 [0015] The method allows arithmetic coding and Golomb-Rice entropy encoding or mixed independent existence in the data frame, Gen P, all data arithmetic coding method using the bit stream, all the data stream using Golomb-Rice coding method or the code stream data, and arithmetic coding method Golomb-Rice coding coexist.

[0016] 所述根据运算复杂度选择算术与Golomb-rice熵编解码,是通过对编解码技术中的各个处理模块进行归一化处理,为各编解码模块设定复杂度权重,根据计算复杂度决策要求,选定适当的处理模块及等级、实现压缩率与编解码速度之间的动态调节。 [0016] The computational complexity of the selection and arithmetic entropy encoding and decoding Golomb-rice according to the codec through in each processing module normalized, setting the weights for the complexity of the encoding and decoding module, according to the computational complexity requirements of decision, and decide on appropriate level processing module, dynamic compression ratio adjustment between the speed and the codec.

[0017] 所述预处理有以下这些形式: [0017] Pretreatment of the following forms:

[0018] 对原始输入信号的滤波整型处理、对原始输入信号的变换处理、对原始输入信号的去相关处理、对原始输入信号的联合处理和对原始输入信号的重组处理。 [0018] Integer filtering process the original input signal, for converting the original input signal processing related to processing of the original input signal, the combined treatment of recombinant and the original input signal processing of the original input signal.

[0019] 所述对音频输入信号进行预测有以下形式: [0019] The input audio signal is predicted to have the following form:

[0020] Levinson-DurbinLPC线性预测、LMS最小均方误差自适应预测、RLS递推最小二乘算法和LTPLong-term预测。 [0020] Levinson-DurbinLPC linear prediction, LMS adaptive minimum mean square error prediction, RLS recursive least squares algorithm and LTPLong-term prediction.

[0021] 所述设置选择算术熵编码和Golomb-Rice熵编码的阈值包括; [0021] The selected set of arithmetic entropy encoding and Golomb-Rice entropy coding threshold comprises;

[0022] 1)设置调用算术编码和Golomb-rice摘编码的米样频率阈值; [0022] 1) Set calls arithmetic coding and Golomb-rice rice sample code excerpts frequency threshold value;

[0023] 2)设置调用算术编码和Golomb-rice熵编码的声道数阈值; [0023] 2) Set calls Golomb-rice arithmetic coding and entropy coding the channel number threshold;

[0024] 3)设置调用算术编码和Golomb-rice熵编码的预测滤波器的最大阶数阈值; The maximum order of the predictive filter threshold [0024] 3) Set calls arithmetic coding and Golomb-rice entropy encoding;

[0025] 4)设置调用算术编码和Golomb-rice熵编码的预测模块的复杂度阈值。 [0025] 4) complexity threshold setting module calls the prediction arithmetic coding and Golomb-rice entropy encoding.

[0026] 所述预测模块的复杂度阈值选择LPC预测阶数作为决策因子: [0026] The complexity of the prediction selection threshold module LPC prediction order as a decision factor:

[0027] LPC>16,摘编码选择Golomb-Rice; [0027] LPC> 16, selection code excerpts Golomb-Rice;

[0028] LPC〈 = 16,熵编码选择算术编码。 [0028] LPC <= 16, arithmetic coding entropy code selection.

[0029] 为实现本发明的方法,还提出一种多声道无损音频混合编解码装置,包括混合编码装置和混合解码装置, [0029] The method of the present invention is achieved further provides a multi-channel lossless audio codecs mixing apparatus comprises a mixing apparatus and mixing coding decoding means

[0030] 所述混合编码装置包括: [0030] The hybrid coding apparatus comprising:

[0031] 预处理模块,对原始音频输入信号进行预处理,减少信号之间的冗余度,提高压缩率; [0031] The pre-processing module, the original audio input signal pre-processing, reducing the redundancy between signals, improve the compression ratio;

[0032] 预测模块,去除原始信号前后数据的相关性; [0032] The prediction module, before and after removal of the original signal correlation of data;

[0033] 复杂度决策模块,根据信号的采样频率,声道个数,预测滤波器的阶数以及预测处理模块的计算复杂度选择调用算术或者Golomb-rice熵编码模块: [0033] The complexity of the decision module, the calculation order signal sampling frequency, number of channels, the prediction filter processing module and a prediction selection call complexity Golomb-rice or arithmetic entropy coding module:

[0034] 熵编码模块,包括算术编码和Golomb-Rice熵编码的混合熵编码器; [0034] The entropy encoding module, an arithmetic coding and Golomb-Rice hybrid entropy encoding the entropy encoder;

[0035] 所述混合解码装置包括: [0035] mixing said decoding apparatus comprising:

[0036] 预测逆处理模块,对应预测处理的逆过程; [0036] The inverse prediction processing module, a corresponding inverse prediction processing procedure;

[0037] 预处理逆处理模块,对应预处理的逆过程; [0037] The inverse pre-processing module, the reverse process corresponding to pretreatment;

[0038] 熵解码模块,包括算术解码和Golomb-Rice熵解码的混合熵解码器。 [0038] The entropy decoding module, including arithmetic decoding and entropy decoding the Golomb-Rice decoder entropy of mixing.

[0039] 所述线性预测模块为基于LPC线性预测的128阶预测模块。 The [0039] first order linear prediction module 128 LPC prediction module based on linear prediction.

[0040] 本发明的有益效果: [0040] Advantageous effects of the invention:

[0041] 1、因本发明在考虑熵编码复杂度与编解码效率之间的平衡,提供了一种适用于多声道无损音频数据的混合熵编码方法,利用本方法可以实现在编解码复杂度和编解码速度之间的切换,可以根据编码设备或解码端设备的处理器运算能力差别(例如:需要实现高压缩率可以选择使用算术编码方法,需要实现高解码速度可以选择Golomb-Rice编码方法),选择适用的编解码模式,可以满足更广泛的应用需求。 [0041] 1, the present invention is due to a balance between entropy encoding considering complexity of the encoding and decoding efficiency, a suitable multi-channel hybrid lossless entropy encoding method of the audio data, the present method may be implemented using a codec complex and the degree of switching between encoding and decoding speed, processor power based on the difference encoding device or a decoding device side (for example: the need to achieve a high compression rate can be selected using the arithmetic coding method, the need to achieve a high speed decoding Golomb-Rice coding can be selected method), select the appropriate codec mode to meet a wider range of application requirements.

[0042] 2、本发明可以在受限的处理资源下实现高质量多声道的无损音频编码,有效降低硬件成本。 [0042] 2, the present invention can achieve high-quality multi-channel lossless audio coding process at the resource constrained, reduce hardware costs.

[0043] 3、因改进技术具有简单,高效的特点,很容易被应用到数字音频领域,提供无损音频编解码方案。 [0043] 3, due to improved techniques for simple, efficient characteristics, can easily be applied to the field of digital audio, there is provided a lossless audio codec scheme.

[0044] 4、作为无损音频技术的改善点,可以在中国自主的无损音频标准中获得应用。 [0044] 4-point improvement as lossless audio technology, can find application in China's own lossless audio standards.

附图说明 BRIEF DESCRIPTION

[0045] 图1是本发明的混合编码核心模块结构框图; [0045] FIG. 1 is a block diagram of a hybrid module coding the core structure of the present invention;

[0046] 图2是本发明的无损音频编码模块结构框图; [0046] FIG. 2 is a block diagram showing a lossless audio encoding module of the present invention;

[0047] 图3根据采样频率选择熵编码器示意图 [0047] FIG. 3 a schematic diagram of an entropy encoder to select the sampling frequency

[0048] 图4是根据声道数选择熵编码器示意图 [0048] FIG. 4 is a schematic diagram according to an entropy encoder the number of selected channels

[0049] 图5是根据数据的量化位数选择熵编码器示意图 [0049] FIG. 5 is selected entropy encoder the number of bits according to the quantization data of a schematic

[0050] 图6是根据预测滤波器的阶数选择熵编码器示意图 [0050] FIG. 6 is a schematic view of the order of the prediction filter selected according to the entropy encoder

[0051] 图7是根据预测处理模块的复杂度选择熵编码器示意图 [0051] FIG. 7 is selected according to the complexity of the entropy coder schematic prediction processing module

[0052] 图8是根据组合信息选择熵编码器示意图 [0052] FIG. 8 is a schematic view of selecting an entropy encoder according to combination information

[0053] 图9是本发明的混合解码核心模块结构框图; [0053] FIG. 9 is a block diagram of hybrid core decoding module according to the invention;

[0054] 图10是本发明的无损音频解码模块结构框图; [0054] FIG. 10 is a block diagram showing the structure of a lossless audio decoding module of the present invention;

[0055] 图11是本发明的混合熵编码模式示例1 ; [0055] FIG. 11 is a hybrid entropy coding mode according to the present invention example 1;

[0056] 图12是本发明的混合熵编码模式示例2 [0056] FIG. 12 is a hybrid entropy coding mode according to the present invention Example 2

[0057] 图13是本发明的混合熵编码模式示例3 [0057] FIG. 13 is a hybrid entropy coding mode according to the present invention Example 3

[0058] 图14是本发明的混合熵编码模式示例4。 [0058] FIG. 14 is a hybrid entropy coding mode example 4 of the present invention.

具体实施方式 detailed description

[0059] 为使本发明的目的、技术方案和优点更加清楚明白,以下结合具体实施例,对本发明进一步详细说明。 [0059] To make the objectives, technical solutions, and advantages of the present invention will become more apparent hereinafter with reference to specific embodiments, the present invention is further described in detail.

[0060] 本发明关键技术包括: [0060] The key technology of the present invention comprises:

[0061] (1)通过使用算术编码和Golomb-Rice编码的混合熵编码方法来提高编解码速度并适当提高压缩率,本发明允许算术编码和Golomb-Rice编码在数据帧中独立存在或混合存在,即,码流内全部数据采用算术编码方法、码流内全部数据采用Golomb-Rice编码方法或码流内数据米用算术编码和Golomb-Rice编码并存的方法; [0061] (1) improved by using mixing entropy coding arithmetic coding and Golomb-Rice coding method of the codec rate and appropriately increase the compression ratio, the present invention allows the arithmetic coding and Golomb-Rice coding of independent existence in the data frame or mixed , i.e., all of the data stream using the arithmetic coding method, all the data streams using the method of Golomb-rice coding method or the code stream data with m arithmetic coding and Golomb-rice coding coexist;

[0062] (2)利用计算复杂度决策扩展无损音频编码适用范围,实现对更多声道数量和更高采样频率的支持;通过对编解码技术中的各个处理模块进行归一化处理,为各编解码模块设定复杂度权重,根据计算复杂度决策要求,选定适当的处理模块及等级。 [0062] (2) using the calculated decision complexity scalable lossless audio coding scope, to implement support for more channels and a higher sampling frequency number; normalizing process by each codec art processing module for the each codec module sets weights complexity, computational complexity decision according to claim, and a processing module selected appropriate level.

[0063] (3)利用简单的计算复杂度决策实现压缩率与编解码速度之间的动态调节; [0063] (3) a simple computational complexity decision dynamically adjusting the compression ratio between the speed of the codec;

[0064] 1、算术与Golomb-rice的混合编码方法 [0064] 1, arithmetic coding method and hybrid Golomb-rice of

[0065] 1. 1混合编码核心框图 [0065] 1.1 hybrid coding a block diagram of the core

[0066] 图1是本发明的混合编码核心模块结构框图;如图所示,本发明允许算术编码和Golomb-Rice编码在数据帧中独立存在或混合存在。 [0066] FIG. 1 is a block diagram of a hybrid module coding the core structure of the present invention; As shown, the present invention allows for arithmetic coding and Golomb-Rice coding independently of the presence or in the presence of mixed data frames.

[0067] 1. 2无损音频编码框图 [0067] lossless audio coding block diagram 1.2

[0068] 图2是本发明的无损音频编码模块结构框图;显示无损音频编码模块的一个实施例,下面具体说明各个模块的结构特点。 [0068] FIG. 2 is a block diagram showing a lossless audio encoding module of the present invention; lossless audio encoding module displays one embodiment, the following detailed description of the structural characteristics of each module.

[0069] 1. 3预处理模块 [0069] The preprocessing module 1.3

[0070] 对原始音频输入信号进行一些特殊的处理,减少信号之间的冗余度,提高压缩率。 [0070] The original audio signal input some special processing to reduce redundancy between the signals, increase the compression ratio. 可能有以下这些形式: These may have the following form:

[0071] 1)对原始输入信号的滤波处理(如滤波整型)。 [0071] 1) filtering processing on the original input signal (such as filtering integer).

[0072] 2)对原始输入信号的变换处理(如小波变换)。 [0072] 2) process for the conversion of the original input signal (e.g., wavelet transform).

[0073] 3)对原始输入信号的去相关处理(如声道去相关)。 [0073] 3) de-correlation processing on the original input signal (such as channel decorrelation).

[0074] 4)对原始输入信号的联合处理(如联合声道编码) [0074] 4) joint processing of the original input signal (such as joint channel coding)

[0075] 5)对原始输入信号的重组处理(如声道数据的交叉重组) [0075] 5) treatment of recombinant original input signal (e.g., cross-channel data recombinant)

[0076] 1. 4预测模块 [0076] 1.4 prediction module

[0077] 预测模块是用FIR(Finiteimpulseresponsefilters),IIR(Infiniteimpulse responsefilters)滤波器技术去除原始信号前后数据的相关性,可能有以下这些形式: [0077] The prediction module is FIR (Finiteimpulseresponsefilters), IIR (Infiniteimpulse responsefilters) filter techniques before and after removal of the correlation of the original signal data, may have the following forms:

[0078] 1)LPC(LinearPrediction)线性预测; [0078] 1) LPC (LinearPrediction) linear prediction;

[0079] 2)LMS(LeastMeanSquare)最小均方误差自适应预测; [0079] 2) LMS (LeastMeanSquare) minimum mean square error adaptive prediction;

[0080] 3)RLS(RecursiveLeastSquare)递推最小二乘算法; [0080] 3) RLS (RecursiveLeastSquare) recursive least squares algorithm;

[0081] 4)LTP(long-termPrediction)Long-term预测; [0081] 4) LTP (long-termPrediction) Long-term prediction;

[0082] 上面列举的四种预测方法用的都是FIR滤波器技术,实现的算法都是: [0082] The four prediction methods listed above are used in the FIR filter technology, algorithms are:

Figure CN104392725AD00071

,其中N为FIR滤波器的阶数,w为FIR滤波器的系数向量,x为输入向量,y为输出向量。 , Where N is the order of the FIR filter, w is the FIR filter coefficient vector, x is the input vector, y is the output vector.

[0083] 所不同的是LPC是使代价函数误差平方和 [0083] The difference is that LPC square error cost function

Figure CN104392725AD00072

最小,LMS是使代价函数均方误差最/]' Minimum, LMS is the mean square error cost function most /] '

Figure CN104392725AD00073

,RLS是使代价函数指数加权的误差平方和 , RLS is a cost function exponentially weighted squared error

Figure CN104392725AD00074

最小计算出最优的滤波器系数,并通过滤波器预测去除数据之间的相关性;而LTP是通过滤波器预测和延迟时间来去除声道之间的相关性,与其它三种相比,它增加了对延迟时间的考虑,根据采样频率的不同选择不同的延迟采样数对输入数据进行去相关处理。 Minimum calculate the optimal filter coefficient, and to remove the correlation between the data predicted by the filter; LTP is the delay time and to remove the correlation between the channels, compared with the other three by the filter prediction, it is considered to increase the delay time for correlation processing to the input data according to different choice of sampling frequency number of samples delay. 其中M为处理单位的长度(如帧长),A为遗忘因子。 Wherein M is the length of the processing unit (e.g., frame length), A is a forgetting factor.

[0084] 1. 5复杂度决策模块 [0084] 1.5 complexity decision module

[0085] 对于某种特定的熵编码算法,系统的计算复杂度主要跟数字信号的采样频率,声道个数,数据的量化位数,预处理滤波器的阶数,预处理模块的复杂度大小有关,随着这些参数从小到大的变化,会带来整个系统计算复杂度的增加。 [0085] For the complexity of the computational complexity of the order of a particular entropy coding algorithm, the main system with a sampling frequency quantization bit number of the digital signals, number of channels, data preprocessing filter, preprocessing module about the size, complexity increases as these parameters vary from small to large, it will bring the entire system is calculated. 在以上一致的条件下,算术编码往往表现出较高的压缩率,GolombRice表现较好的压缩速度。 In any of the above conditions, the arithmetic coding tend to exhibit a higher compression ratio, GolombRice performed better compression rate.

[0086] 根据信号的采样频率,声道个数,数据的量化位数,预测滤波器的阶数以及预测处理模块的复杂度选择算术或者Golomb-rice熵编码方法,通过限定编解码的峰值运算量,从而使编解码器能很好的应用于资源受限与不受限的环境中。 [0086] The quantization bit number signal sampling frequency, number of channels, the data, the complexity of the processing module and the prediction order prediction filter selection Golomb-rice or arithmetic coding method, codec defined by the peak computing amount, so that the codec can be well applied to the resource-constrained and non-constrained environments. 决策的方法如下: Decision-making methods are as follows:

[0087] 1)根据采样频率FS选择熵编码器:根据实际应用中处理器资源的情况设置采样频率阈值fsuimh,当输入信号源的采样频率大于等于所设置的阈值fsuimi!时,使用Golomb-rice熵编解码器;当输入信号的采样频率小于所设置的阈值FSLimen时,使用算术熵编解码器。 [0087] 1) FS selected according to the sampling frequency of the entropy encoder:! Set the sampling frequency threshold value fsuimh according to the actual application processor resources, the threshold value fsuimi when the input signal source sampling frequency is greater than equal to the set, using the Golomb-rice the entropy codec; when the threshold sampling frequency of the input signal is less than the set FSLimen, using the arithmetic entropy codec. 采样频率的单位为Hz,KHz等。 The unit sampling frequency is Hz, KHz like. 采样频率的取值范围:0HZ〈FS〈192000Hz(根据应用不同,采样频率范围可以进行调整)。 Sampling frequency ranges: 0HZ <FS <192000Hz (depending on the application, the sampling frequency range may be adjusted). 如图3所示。 As shown in Figure 3.

[0088] 2)根据声道数CH选择熵编码器:根据实际应用中处理器资源的情况设置声道数阈倌CHLimen,当输入信号源的声道数大于等于所设置的阈倌CHLimen时,使用Golomb-rice熵编解码器;当输入信号的声道数小于所设置的阈值FSLimen时,使用算术熵编解码器,声道个数的单位为个。 [0088] 2) The number of channels selected entropy encoder CH: setting a threshold number of channels groom CHLimen according to the actual application processor resources, when the number of channels of the input signal source is greater than the threshold set equal to groom CHLimen, using Golomb-rice entropy codec; FSLimen threshold when the input signal is less than the number of channels provided, the decoder using the arithmetic entropy coding, the number of channels as a unit. 声道数取值范围:〇〈CH〈X(X可以根据应用取值不同,例如用于家庭娱乐,可以取值为8,对于影院娱乐,可以取值32或更高)。 Number of channels in the range: square <CH <X (X may be different values ​​depending on the application, for example for home entertainment value of 8 can be, for theater entertainment can take the values ​​32 or higher). 如图4所示。 As shown in Figure 4.

[0089] 3)根据数据的量化位数QBIT选择熵编码器:根据实际应用中处理器资源的情况设置量化位数阈值QBITLimen,当数据的量化位数大于等于所设置的阈值QBITLimen时,用Golomb-rice熵编解码器,当数据的量化位数小于所设置的阈值QBITLimen时,用算术熵编解码器。 [0089] 3) selecting an entropy encoder the number of bits according to the quantization data QBIT: setting a threshold value of quantization bits according to the actual application QBITLimen processor resources, when the threshold number of data bits QBITLimen than or equal to the quantized provided with Golomb -rice entropy codec, when the threshold number of bits quantized data is less than the set QBITLimen, arithmetic entropy codec. 0811'的取值范围0〈0811'〈32(根据处理器的字长不同,0811'可以改变),如图5所/_J、i〇 0811 'is in the range 0 <0811' <32 (depending on the word length of the processor, 0811 'can be changed), as shown in FIG 5 / _J, i〇

[0090] 4)根据预测滤波器的阶数0RD选择熵编码器:根据实际应用中处理器资源的情况设置预测滤波器的阶数阈值ORDUmm,当预测滤波器的阶数大于等于所设置的阈值ORDLimen时,用Golomb-rice熵编解码器;当预测滤波器的阶数小于所设置的阈值ORDLimen时,用算术熵编解码器。 [0090] 4) 0RD selected entropy coder according to the order of the prediction filter: the prediction filter is provided according to the actual application processor resource order thresholds ORDUmm, when the order of the prediction filter is greater than the set threshold value is equal to when ORDLimen using Golomb-rice entropy codec; when the threshold ORDLimen prediction filter of order less than the set, entropy arithmetic codec. 阶数0RD取值范围1〈0RD〈1024.如图6所示。 Order 0RD range 1 <0RD <1024. As shown in Fig.

[0091] 5)根据预测处理模块的复杂度PRCC0选择熵编码器:根据实际应用中处理器资源的情况设置预测处理模块的复杂度阈值PRCCOUimn,当预测处理模块的复杂度大于等于所设置的阈值PRCCOLimen时,使用Golomb-rice熵编解码器;当预测处理模块的复杂度小于所设置的阈值PRCCOLimen时,使用算术熵编解码器,复杂度单位可以是CYCLE, MIPS(MHz)等,取值范围0〈PR0CC0〈x(x可以根据应用允许的最大处理量内设定适当的值, 例如5MIPS)。 [0091] 5) PRCC0 selected entropy coder according to the complexity of the prediction processing module: setting complexity threshold PRCCOUimn prediction processing module according to the actual application processor resources, when the complexity of the prediction processing module is not less than the set threshold value when PRCCOLimen using Golomb-rice entropy codec; when the threshold PRCCOLimen complexity prediction processing module is less than the set, using the arithmetic entropy codec complexity units may be CYCLE, MIPS (MHz) and the like, in the range 0 <PR0CC0 <x (x may be appropriately set according to the value of the maximum processing amount of the application allows, e.g. 5MIPS). 如图7所示。 As shown in FIG.

[0092] 6)根据以上决策因子的组合信息选择熵编码器:组合方法有两种,第一种可以采用并列组合,即各个因子同时满足时,选择某种熵编码算法;例如:当采样频率参数FSM8KHZ,并且声道数参数CH>2声道,并且数据的量化位数QBIT>16时,并且预测模块复杂度参数0RD>14,并且预处理模块复杂度参数PR0CO5MIPS,使用Golomb-Rice编码,否则使用算术编码。 [0092] 6) a combination of the above decision factor selection information entropy encoder according to: There are two methods in combination, a first composition may be used in parallel, i.e. simultaneously satisfy various factors, some entropy coding algorithm selection; for example: when the sampling frequency parameters FSM8KHZ, the number of channels and the parameter CH> 2 channel, and the data of quantization bits QBIT> 16, the prediction module and complexity parameter 0RD> 14, and the preprocessing module complexity parameter PR0CO5MIPS, the use of Golomb-Rice coding, otherwise, using arithmetic coding. 以上组合可以根据实际需要,在某几个参数中进行组合条件限定。 Combination of the above may be combined in defined conditions according to the actual needs of a few parameters.

[0093] 第二种是根据采样频率参数,声道数参数,数据的量化位数参数,预测模块复杂度参数,预处理模块复杂度参数等组合值C0MP选择熵编码器:根据实际应用中处理器资源的情况设置组合阈值COMPLimen。 [0093] The second entropy coder to select the parameters according to the sampling frequency, number of channels parameter, quantization parameter data bits, prediction module complexity parameter, the preprocessing module complexity parameter values ​​like composition C0MP: according to the actual application resource combination is provided where the threshold COMPLimen.

[0094] 组合阈值是通过把采样频率FS,声道数CH,量化位数阈值,预测滤波器的阶数0RD 转换为以CYCLE,MIPS等为单位阈值,与预测处理模块的复杂度阈值求加权平均值得到,并使其跟实际应用中的处理器资源匹配。 [0094] The compositions threshold value is determined by the sampling frequency of the FS, the number of channels CH, quantization threshold number of bits, the prediction filter order 0RD convert to CYCLE, MIPS and other units of the threshold value, the complexity threshold prediction processing module A weighted get the average, and with it the practical application processor resources to match.

[0095] 组合阈值计算公式: [0095] The combination of threshold value calculation formula:

[0096]COMPLimen=ff〇*FSLimen•Ag+ffj*CHLimen•A1+ff2*QBITLimen*A2+ff3 *0RDLimen*A3++ff4 •PROCCLimen [0096] COMPLimen = ff〇 * FSLimen • Ag + ffj * CHLimen • A1 + ff2 * QBITLimen * A2 + ff3 * 0RDLimen * A3 ++ ff4 • PROCCLimen

[0097] 求组合阈值同样的方法求输入数据的组合值,当输入数据的组合值大于等于组合阈值时,用Golomb-rice熵编解码器,当输入数据的组合值大于组合阈值时小于组合阈值时,用算术熵编解码器。 [0097] The compositions find the value of a combination of the threshold value in the same manner of input data, when the input data of the combined value is greater than equal to the combined threshold value, using Golomb-rice entropy codec, less than the combined threshold value when the input data of the combined value is greater than the combination of the threshold value when codec arithmetic entropy.

[0098] 组合值计算公式: [0098] combined value calculation formula:

[0099]COMP=ff〇•FS•Ao+ffi•CH• •QBIT•A2ff3 • 0RD•A3++ff4 •PR0CC [0099] COMP = ff〇 • FS • Ao + ffi • CH • • QBIT • A2ff3 • 0RD • A3 ++ ff4 • PR0CC

[0100] 其中1(|,11,12,1 3,14为采样频率?5,声道数〇1,数据的量化位数0811',预测滤波器的阶数〇RD和预测处理模块的复杂度PR0CC及阈值的加权系数,ApA2,A3为采样频率FS, 声道数CH,数据的量化位数QBIT,预测滤波器的阶数0RD,及阈值转换为以CYCLE,MIPS等为单位的乘积因子。 [0100] where 1 (|, 11,12,1 3,14 5 is the sampling frequency, number of channels 〇1, quantization bits of data 0811 ', and the complex 〇RD prediction processing order of the prediction filter module? PR0CC of weighting coefficients and threshold, ApA2, A3 the FS is the sampling frequency, number of channels CH, QBIT quantized data bits, 0RD order prediction filter, and a threshold value is converted to CYCLE, MIPS as the multiplication factor of units .

[0101] 组合值及组合阈值的单位为CYCLE,MIPS等,如图8所示。 [0101] Unit composition values ​​and combinations threshold as CYCLE, MIPS, etc., as shown in FIG.

[0102] 1.6算术熵编码模块 [0102] 1.6 arithmetic entropy coding module

[0103] 算术熵编码模块的方法如下: [0103] The arithmetic entropy coding modules as follows:

[0104] 1)信源符号序列{^,a2〜am},[0, 1)为当前分析区间,按信源符号的概率序列将当前分析区间划分为m个区间,m个区间与m个信源符号一一映射: [0104] 1) the source symbol sequence {^, a2~am}, [0, 1) for the current analysis section, according to the probability of the source symbol sequence analysis of the current interval is divided into m intervals, and intervals of m m to one mapping the source symbols:

Figure CN104392725AD00091

[0106] 其中Fx(k)为前k个符号的累积概率 [0106] wherein Fx (k) is the cumulative probability of k symbols before

Figure CN104392725AD00092

[0108] 其中Ph)为符号的概率 [0108] wherein Ph) is the probability of a symbol of

[0109] 2)检索信源符号序列,找到当前信源符号ak在当前分析区间所对应的区间[Fx(k-1),Fx (k)),将此区间作为新的当前分析区间。 [0109] 2) to retrieve the source sequence of symbols, symbols ak find the current source in the current analysis interval corresponding to the interval [Fx (k-1), Fx (k)), this section as a new current analysis section.

[0110] 3)按照信源符号的概率序列将当前分析区间划分为m个区间 [0110] 3) the source sequence according to the probability of the current symbol interval is divided into analysis intervals m

Figure CN104392725AD00093

[0112] (如果是自适应或者上下文自适应模型,必须更新概率或者条件概率表),读取下一个消息符号作为当前符号,然后重复第二、三步。 [0112] (or if adaptive context adaptive model must be updated probability or conditional probability table), read the next message symbol as a current symbol, and then repeating the second, three steps. 直到信源符号序列检索完毕。 Until complete source symbol string search.

[0113] 4)最后的当前分析区间中的任何一个数都可以作为信源符号序列{apa^aj的标识。 [0113] 4) Finally, any of a number of the current interval of analysis can be used as a source symbol sequence {apa ^ aj identification.

[0114]算术熵解码方法是算术熵编码的逆过程,它从码流中读取信源符号序列的标识,并根据当前区(初始区间为[0, 1))计算出累积概率区间数组{[FX(0),FX(1)),[FX(1),FX(2),…[Fx(kl),Fx(k))}的索引i,由此解码出与区间[Fx(il),Fx(i))对应的信源符号^,并根据解码的符号序列缩放区间,最终解码出原始信源符号序列,对非自适应的算术编码,概率表固定不变,编码一个符号后不会刷新概率表, 而对于自适应的算术编码,概率表是实时变化的,每编码一个符号后会刷新概率表,下一个编码符号将根据刷新后的概率表进行编码。 [0114] The entropy decoding arithmetic arithmetic entropy encoding method is the reverse process, it reads the source identifier information symbol sequence from the code stream, and calculates the cumulative probability interval array according to the current region (an initial interval [0, 1)) { [FX (0), FX (1)), [FX (1), FX (2), ... [Fx (kl), Fx (k))} the index i, thereby decoding the interval [Fx (il ), Fx (i)) corresponding to the source symbols ^ and scaled symbol sequence in accordance with the decoded interval, the original source finally decoded symbol sequence, the non-adaptive arithmetic coding, fixed probability table, a symbol encoding is not refreshed probability table, and for the adaptive arithmetic coding, the probability table changes in real time, after each encoded symbol will refresh probability table, the next coded symbol will be coded according to the probability table after the refresh.

[0115] 1. 7Golomb-rice摘编码模块 [0115] 1. 7Golomb-rice excerpts code module

[0116]Golomb-rice熵编码的方法如下: [0116] Golomb-rice entropy coding method is as follows:

[0117] l)m值的选取,在理论情况下最优m值如下: [0117] l) m selecting values ​​in the theoretical optimal value of m as follows:

Figure CN104392725AD00101

[0119] 2)分割符号的整数值n:将符号的整数值n分割为前缀9 =pr和余数r= n&(2m_l)两部份,对前缀q采用一元码编码,余数采用m个二进制位进行编码。 [0119] 2) divided signed integer values ​​n: integer value of n symbols is divided into a prefix 9 = pr and a remainder r = n & (2m_l) two parts, a prefix q using unary code encoding, the remainder using the m bits encoded.

[0120] 3)读取下一个信源符号,重复第二步,直到信源符号序列编码完毕。 [0120] 3) read the next source symbols, the second step is repeated until the complete sequence of symbols coded signal source.

[0121] 对非自适应的Golomb-rice编码中,一般以符号块为单位选取m值,以块中所有符号的整数值的平均值作为输入,用上面的公式计算m值,块中每符号的整数值都以相同的m 进行分割。 [0121] The non-adaptive Golomb-rice encoding, symbol block units generally selected value of m, the average value of the block to an integer value as input all of the symbols, the value of m is calculated using the above formula, each symbol block integer values ​​are divided by the same m.

[0122] 而对于自适应的Golomb-rice编码,每个符号的整数值选择的m值都可以不相同(而具体自适应的方法有多种,这里不再描述)。 [0122] For the adaptive Golomb-rice encoding, an integer value of each symbol of the selected m values ​​may not be the same (but with a variety of specific adaptive methods, not described herein).

[0123] 1.8比特流格式化模块 [0123] 1.8 bitstream formatting module

[0124] 格式化特征信息比特与熵编码输出比特,输出格式化的编码码流。 [0124] Format information bits and wherein the output bits of the entropy encoding, the encoded stream output formats.

[0125] 2、算术与Golomb-rice的混合解码方法 [0125] 2, and Golomb-rice arithmetic decoding method of mixing

[0126] 图3是本发明的混合解码核心模块结构框图;图4是本发明的无损音频解码模块结构框图。 [0126] FIG. 3 is a block diagram showing the structure of a hybrid decoding core module of the present invention; FIG. 4 is a block diagram showing the structure of a lossless audio decoding module of the present invention. 其中解码框图中预测逆处理模块及预处理逆处理模块是编码框图中的对应处理的逆过程。 Wherein a block diagram of a decoding prediction inverse pre-processing module and a reverse processing module corresponding to the process is the reverse process of coding the block diagram.

[0127] 2. 1读取熵编码算法标识 [0127] 2.1 entropy encoding algorithm identifier read

[0128] 从码流中读取熵编码算法标识,以决定调用算术还是Golomb-rice熵解码器解码当前声道或者帧数据。 [0128] The entropy coding algorithm identifier read from the code stream to determine or call the Golomb-rice arithmetic entropy decoder or decoding the current frame data channel.

[0129] 2. 2算术熵解码模块 [0129] 2.2 arithmetic entropy decoding module

[0130] 算术熵解码模块的方法如下: [0130] Entropy decoding module arithmetic is as follows:

[0131] 1)初始化当前分析区间,[0, 1)为当前分析区间,按信源符号的概率序列将当前分析区间划分为m个区间,m个区间与m个信源符号一一映射: [0131] 1) Initialize the current interval analysis, [0, 1) for the current analysis section, according to the probability of the source symbol sequence analysis of the current interval is divided into m intervals, m intervals and the m source symbols one mapping:

Figure CN104392725AD00102

[0133]其中Fx(k)为前k个符号的累积概率 [0133] wherein Fx (k) is the cumulative probability of k symbols before

Figure CN104392725AD00103

[0135] 其中Ph)为符号的概率 [0135] wherein Ph) is the probability of a symbol of

[0136] 并从码流中读取信源符号序列的当前标识。 [0136] and reads the source code symbol sequence from the current stream identifier.

[0137] 2)根据当前分析区及当前标识计算出出累积概率区间数组{[FX(0),FX(1)),[FX(1),FX(2),…[Fx(kl),Fx(k))}的索引i,由此解码出与区间 [0137] 2) calculate the cumulative probability interval array {[FX (0) according to the current analysis zone and current identity, FX (1)), [FX (1), FX (2), ... [Fx (kl), Fx (k))} the index i, and thereby decoding section

[Fx(il),Fx(i))对应的信源符号%。 [Fx (il), Fx (i)) corresponding to the source symbol%.

[0138] 3)按照信源符号的概率序列更新当前分析区为[Fx(i_l),Fx(i)),并更新当前标示(如果是自适应或者上下文自适应模型,必须更新概率或者条件概率表)。 [0138] 3) the source symbols according to the probability of updating the current sequence analysis region is [Fx (i_l), Fx (i)), and updates the current label (or if adaptive context adaptive model must be updated probability or conditional probability table).

[0139] 4)重复第二、三步,直到解码出信源符号序列。 [0139] 4) repeating the second, three-step, until decoded source symbol sequence.

[0140] 2. 3Golomb_rice摘解码模块 [0140] 2. 3Golomb_rice Abstract decoding module

[0141]Golomb-rice熵解码模块的方法如下: [0141] Golomb-rice entropy decoding module:

[0142] 1)从码流读取或者自适应计算m值 [0142] 1) read from the code stream or the value of m is calculated adaptively

[0143] 2) -元解码前缀q,从码流中读取m位二制作为余数,计算出信源符号值n= q*2m|r。 [0143] 2) - the elementary decoding prefix q, two m-bit read from the code stream production for the remainder, the source symbol values ​​calculated n = q * 2m | r.

[0144] 3)重复第一、二步,直到解码出信源符号序列。 [0144] 3) repeating the first and second step until the decoded source symbol sequence.

[0145] 实施例 [0145] Example

[0146] 下面通过一个具体的例子来说明对本发明的实施方法,对于基于无损音频混合编解码技术,具体实现步骤如下: [0146] below with a specific example to illustrate the method of the embodiment of the present invention, for mixing based lossless audio codec, specific implementation steps are as follows:

[0147] 1)参考本发明框图实现无损音频编解码器,基本模块如下: [0147] 1) of the present invention with reference to a block diagram of a lossless audio codec, the following basic modules:

[0148] 声道去相关模块 [0148] channel decorrelation module

[0149] 线性预测模块(基于LPC线性预测128阶) [0149] Linear prediction module (LPC stage 128 based on linear prediction)

[0150] 复杂度决策模块 [0150] the complexity of the decision-making module

[0151] 摘编码模块(包括算术编码和Golomb-Rice编码混合编码器) [0151] Adapted code module (including arithmetic coding and Golomb-Rice coding hybrid encoder)

[0152] 2)复杂度决策设计(例) [0152] 2) complexity of design decisions (Example)

[0153] 参考本发明方法,可以选择LPC预测阶数作为决策因子: [0153] The method of the present invention, reference may be selected as the LPC prediction order decision factors:

[0154]LPC>16,摘编码选择Golomb-Rice [0154] LPC> 16, select the Golomb-Rice code excerpts

[0155]LPC〈 = 16,熵编码选择算术编码 [0155] LPC <= 16, arithmetic coding entropy code selection

[0156] 本发明的方法与纯的算术熵编解码比,在压缩性能不恶化的前提下,提高了编解码的速度,而与纯的Golomb-rice熵编解码比,提高了压缩性能。 The method [0156] according to the present invention with pure arithmetic entropy encoding and decoding ratio, under the premise of not deteriorating the compression performance, increase the speed of the encoding and decoding, and the pure entropy codec Golomb-rice ratio, improved compression performance. 它通过限制熵编解码器的峰值性能,使得用这种方法的无损音频编解码器能很好的应用于资源受限与不受限的环境中。 By limiting the peak entropy codec performance, so that a lossless audio codec using this method can be well applied in resource-constrained and non-constrained environments.

[0157] 以上所述的具体实施例,对本发明的目的、技术方案和有益效果进行了进一步的详细说明,所应理解的是,以上所述仅为本发明的具体实施例而已,并不用于限制本发明, 凡在本发明的精神和原则之内,所做的任何修改、等同替换、改进等,均应包含在本发明的保护范围之内。 [0157] The foregoing specific embodiments of the object, technical solutions, and advantages of the invention will be further described in detail It should be understood that the above descriptions are merely embodiments of the present invention, but not intended limit the present invention, all within the spirit and principle of the present invention, any changes made, equivalent substitutions and improvements should be included within the scope of the present invention.

Claims (10)

  1. 1. 一种多声道无损音频混合编解码方法,其特征在于,所述方法根据采样频率,声道数,量化位数,预处理模块及预测模块的运算复杂度选择算术与Golomb-rice熵编解码,利用算术与Golomb-rice熵编解码各自的优点,实现对多声道音频信号的算术与Golomb-rice的混合编解码。 A multi-channel audio mixing lossless encoding and decoding method, characterized in that the method according to the sampling frequency, number of channels, the computational complexity of quantization bits, the preprocessing module and the prediction module selection entropy arithmetic and Golomb-rice codec, and Golomb-rice using an arithmetic entropy encoding and decoding their own advantages to achieve hybrid codec multichannel audio signal, the arithmetic and Golomb-rice.
  2. 2. 根据权利要求1的混合编解码方法,其特征在于: 所述算术与Golomb-rice的混合熵编码方法包括具体步骤: 对原始音频输入信号进行预处理,减少信号之间的冗余度,提高压缩率; 对音频输入信号进行预测,去除原始信号前后数据的相关性; 根据音频数据采样频率,声道数,预测模块复杂度参数,预处理模块复杂度参数,音频数据特征参数组合信息,设置选择算术熵编码和Golomb-Rice熵编码的阈值; 所述算术与Golomb-rice的混合熵解码方法包括具体步骤: 进行预测逆处理; 进行预处理逆处理; 从码流中读取熵编码算法标识,以决定调用算术还是Golomb-rice熵解码器,解码当前声道或者帧数据。 2. The hybrid codec method according to claim 1, wherein: said arithmetic entropy coding methods and hybrid Golomb-rice include the particular steps of: preprocessing an original audio input signal to reduce the redundancy between the signals, improve the compression ratio; predicted audio input signal, the correlation before and after the removal of the original data signal; a frequency, number of channels, the prediction module complexity parameter, the preprocessing module complexity parameter, characteristic parameters of audio information according to a combination of audio data samples, set select arithmetic entropy encoding and entropy encoding Golomb-Rice threshold; said arithmetic entropy decoding with hybrid Golomb-rice method includes specific steps: inverse prediction process; pretreatment inverse process; entropy encoding algorithm read from the code stream identification, to determine or call the Golomb-rice arithmetic entropy decoder, or decoding the current frame data channel.
  3. 3. 根据权利要求1的方法,其特征在于,所述方法允许算术编码和Golomb-Rice熵编码在数据帧中独立存在或混合存在,即,码流内全部数据采用算术编码方法、码流内全部数据米用Golomb-Rice编码方法或码流内数据米用算术编码和Golomb-Rice编码并存的方法。 3. The method according to claim 1, wherein said arithmetic coding method allows entropy encoding and Golomb-Rice or mixed independent existence in the data frame, i.e., all data arithmetic coding method using the bit stream, the stream the method of rice all data m and an arithmetic coding Golomb-rice coding coexist with Golomb-rice coding method or the data stream.
  4. 4. 根据权利要求1的方法,其特征在于,所述根据运算复杂度选择算术与Golomb-rice 熵编解码,是通过对编解码技术中的各个处理模块进行归一化处理,为各编解码模块设定复杂度权重,根据计算复杂度决策要求,选定处理模块及等级、实现压缩率与编解码速度之间的动态调节。 4. A method according to claim 1, wherein said selecting and Golomb-rice arithmetic entropy encoding and decoding according to the computational complexity of the codec by each of the processing techniques normalization processing module, for the codec module sets weights complexity, computational complexity decision according to claim level is selected and the processing module, dynamic compression ratio adjustment between the speed and the codec.
  5. 5. 根据权利要求1的方法,其特征在于,所述预处理有以下这些形式: 对原始输入信号的滤波整型处理、对原始输入信号的变换处理、对原始输入信号的去相关处理、对原始输入信号的联合处理和对原始输入信号的重组处理。 The method according to claim 1, wherein said pretreatment has the following form: Integer filtering the original input signal processing, conversion processing of the original input signal, correlation processing to the original input signal, to joint processing and restructuring the original input signal processing on the original input signal.
  6. 6. 根据权利要求1的方法,其特征在于,所述对音频输入信号进行预测有以下形式: Levinson-Durbin LPC线性预测、LMS最小均方误差自适应预测、RLS递推最小二乘算法和LTP Long-term 预测。 6. The method according to claim 1, wherein said prediction of the audio input signal in the following form: Levinson-Durbin LPC linear prediction, the LMS adaptive minimum mean square error prediction, recursive least squares algorithm and the RLS LTP Long-term forecast.
  7. 7. 根据权利要求1的方法,其特征在于,所述设置选择算术熵编码和Golomb-Rice熵编码的阈值包括; 1) 设置调用算术编码和Golomb-rice熵编码的米样频率阈值; 2) 设置调用算术编码和Golomb-rice熵编码的声道数阈值; 3) 设置调用算术编码和Golomb-rice熵编码的预测滤波器的最大阶数阈值; 4) 设置调用算术编码和Golomb-rice熵编码的预测模块的复杂度阈值。 7. The method according to claim, wherein said setting selection arithmetic entropy encoding and Golomb-Rice entropy coding threshold comprises; 1) disposed meter sample frequency threshold call arithmetic coding and Golomb-rice entropy coding; 2) set calls arithmetic coding and Golomb-rice entropy coding of the number of channels a threshold value; 3) set the maximum order of the threshold values ​​of the prediction filter is called arithmetic coding and Golomb-rice entropy coding; 4) is provided to call an arithmetic coding and Golomb-rice entropy encoding complexity threshold prediction module.
  8. 8. 根据权利要求6的方法,其特征在于,所述预测模块的复杂度阈值选择LPC预测阶数作为决策因子: LPC > 16,摘编码选择Golomb-Rice ; LPC〈= 16,熵编码选择算术编码。 8. The method according to claim 6, wherein the prediction complexity threshold module selection order LPC predictor as a decision factor: LPC> 16, selection code excerpts Golomb-Rice; LPC <= 16, arithmetic entropy code selection coding.
  9. 9. 一种多声道无损音频混合编解码装置,包括混合编码装置和混合解码装置,其特征在于, 所述混合编码装置包括: 预处理模块,对原始音频输入信号进行预处理,减少信号之间的冗余度,提高压缩率; 预测模块,去除原始信号前后数据的相关性; 复杂度决策模块,根据信号的采样频率,声道个数,预测滤波器的阶数以及预测处理模块的计算复杂度选择调用算术或者Golomb-rice熵编码模块: 熵编码模块,包括算术编码和Golomb-Rice熵编码的混合熵编码器; 所述混合解码装置包括: 预测逆处理模块,对应预测处理的逆过程; 预处理逆处理模块,对应预处理的逆过程; 熵解码模块,包括算术解码和Golomb-Rice熵解码的混合熵解码器。 A multi-channel lossless audio codecs mixing apparatus comprises a mixing apparatus and mixing coding decoding apparatus, characterized in that said hybrid coding apparatus comprising: a preprocessing module for preprocessing the original audio input signal to reduce the signal between redundancy, increasing the compression ratio; prediction module, remove the correlation before and after the raw data signal; complexity decision module calculates the order of the signal sampling frequency, number of channels, and the prediction filter in accordance with the prediction processing module or choose to call the arithmetic complexity of Golomb-rice entropy encoding module: entropy encoding module, an arithmetic coding and Golomb-Rice encoding of the entropy encoder entropy of mixing; mixing said decoding apparatus comprising: an inverse prediction processing module, a corresponding inverse prediction processing procedure ; preconditioning reverse processing modules, corresponding to an inverse process of pretreatment; entropy decoding module, including arithmetic decoding and entropy decoding the Golomb-Rice decoder entropy of mixing.
  10. 10.根据权利要求8混合编解码装置,其特征在于,所述线性预测模块为基于LPC线性预测的128阶预测模块。 10.8 hybrid codec apparatus according to claim, characterized in that said linear prediction module 128 is a module based on LPC prediction order linear prediction.
CN 201410721299 2014-12-02 2014-12-02 Method and device for hybrid coding/decoding of multi-channel lossless audios CN104392725A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201410721299 CN104392725A (en) 2014-12-02 2014-12-02 Method and device for hybrid coding/decoding of multi-channel lossless audios

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201410721299 CN104392725A (en) 2014-12-02 2014-12-02 Method and device for hybrid coding/decoding of multi-channel lossless audios

Publications (1)

Publication Number Publication Date
CN104392725A true true CN104392725A (en) 2015-03-04

Family

ID=52610618

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201410721299 CN104392725A (en) 2014-12-02 2014-12-02 Method and device for hybrid coding/decoding of multi-channel lossless audios

Country Status (1)

Country Link
CN (1) CN104392725A (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101052972A (en) * 2004-09-22 2007-10-10 液滴技术有限公司 Multiple technique entropy coding system and method
CN101243497A (en) * 2005-07-11 2008-08-13 Lg电子株式会社 Apparatus and method of coding and decoding an audio signal
CN101601087A (en) * 2006-11-16 2009-12-09 弗劳恩霍夫应用研究促进协会 Device for encoding and decoding
CN101809653A (en) * 2007-12-06 2010-08-18 Lg电子株式会社 A method and an apparatus for processing an audio signal
WO2011162723A1 (en) * 2010-06-21 2011-12-29 Agency For Science, Technology And Research Entropy encoder arrangement and entropy decoder arrangement
CN102368385A (en) * 2011-09-07 2012-03-07 中科开元信息技术(北京)有限公司 Backward block adaptive Golomb-Rice coding and decoding method and apparatus thereof
CN103748886A (en) * 2011-06-16 2014-04-23 弗兰霍菲尔运输应用研究公司 Entropy coding of motion vector differences

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101052972A (en) * 2004-09-22 2007-10-10 液滴技术有限公司 Multiple technique entropy coding system and method
CN101243497A (en) * 2005-07-11 2008-08-13 Lg电子株式会社 Apparatus and method of coding and decoding an audio signal
CN101601087A (en) * 2006-11-16 2009-12-09 弗劳恩霍夫应用研究促进协会 Device for encoding and decoding
CN101809653A (en) * 2007-12-06 2010-08-18 Lg电子株式会社 A method and an apparatus for processing an audio signal
WO2011162723A1 (en) * 2010-06-21 2011-12-29 Agency For Science, Technology And Research Entropy encoder arrangement and entropy decoder arrangement
CN103748886A (en) * 2011-06-16 2014-04-23 弗兰霍菲尔运输应用研究公司 Entropy coding of motion vector differences
CN102368385A (en) * 2011-09-07 2012-03-07 中科开元信息技术(北京)有限公司 Backward block adaptive Golomb-Rice coding and decoding method and apparatus thereof

Similar Documents

Publication Publication Date Title
US6125348A (en) Lossless data compression with low complexity
US6043763A (en) Lossless data compression with low complexity
US6028541A (en) Lossless data compression with low complexity
US20030195742A1 (en) Encoding device and decoding device
US20050192799A1 (en) Lossless audio decoding/encoding method, medium, and apparatus
US6118392A (en) Lossless data compression with low complexity
US20020049586A1 (en) Audio encoder, audio decoder, and broadcasting system
US20110001643A1 (en) System and method for providing high throughput entropy coding using syntax element partitioning
US7433824B2 (en) Entropy coding by adapting coding between level and run-length/level modes
US7822601B2 (en) Adaptive vector Huffman coding and decoding based on a sum of values of audio data symbols
US20070043575A1 (en) Apparatus and method for encoding audio data, and apparatus and method for decoding audio data
JP2003044097A (en) Method for encoding speech signal and music signal
CN101110214A (en) Speech coding method based on multiple description lattice type vector quantization technology
CN1681213A (en) Lossless audio coding/decoding method and apparatus
JPH09106299A (en) Coding and decoding methods in acoustic signal conversion
Wang et al. Context-based adaptive arithmetic coding in time and frequency domain for the lossless compression of audio coding parameters at variable rate
CN1420487A (en) Method for quantizing one-step interpolation predicted vector of 1kb/s line spectral frequency parameter
CN101266796A (en) A quantified coding method and device
CN101015216A (en) Context-based signal coding and decoding
KR20080070026A (en) Encoder-assisted frame loss concealment techniques for audio coding
US20050075888A1 (en) Fast codebook selection method in audio encoding
CN101849258A (en) Technique for encoding/decoding of codebook indices for quantized MDCT spectrum in scalable speech and audio codecs
CN1984336A (en) Binary method and device
JP2009501943A (en) Selective use of multiple entropy models in adaptive coding and decoding
CN101931414A (en) Pulse coding method and device, and pulse decoding method and device

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
RJ01