CN101373594A - Method and device for correcting audio signal - Google Patents

Abstract

Translated fromChinese

本发明属于通信技术领域，本发明公开了一种修正音频信号的方法及装置。本发明方法包括：在音频信号数据帧边界附近截取一段信号；对截取的所述信号进行线性处理，得到新的信号；计算该新的信号的测评指标，当计算所得的测评指标小于预置的测评指标时，继续进行线性处理，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。本发明具有以下效果：计算量小，通过修正音频信号，使音频信号的时域波形变的平滑且有周期性；并且保证了信号的相位信息，减小了频谱发散，使频谱趋于平滑，进而消除边界效应。

The invention belongs to the technical field of communication, and discloses a method and a device for correcting audio signals. The method of the present invention includes: intercepting a section of signal near the boundary of the audio signal data frame; performing linear processing on the intercepted signal to obtain a new signal; calculating the evaluation index of the new signal, when the calculated evaluation index is less than the preset When evaluating the index, the linear processing is continued until the evaluation index of the linearly processed signal is greater than or equal to the preset evaluation index. The present invention has the following effects: the amount of calculation is small, and the time-domain waveform of the audio signal becomes smooth and periodic by correcting the audio signal; and the phase information of the signal is guaranteed, the spectrum divergence is reduced, and the spectrum tends to be smooth. Thereby eliminating the boundary effect.

Description

Translated fromChinese

修正音频信号的方法及装置Method and device for correcting audio signal

技术领域technical field

本发明涉及通信技术领域，尤其涉及一种修正音频信号的方法及装置。The invention relates to the field of communication technology, in particular to a method and device for correcting audio signals.

背景技术Background technique

变换域编码是当今音频编码标准普遍采用的压缩技术之一，属于频域编码，通过降低信号中各分量间的相关性，并对变换后的系数进行量化和编码，以达到信息压缩的目的，充分利用人耳在频域上的听觉特性，如掩蔽效应和临界频带，来实现对音频信号的压缩。在实际应用中，通常把音频信号分成若干独立的数据帧进行FFT(Fast Fourier Transform，快速傅立叶变换)或DCT(Discrete Cosine Transform，离散余弦变换)，但是不能保证每一帧信号在边缘是连续的，也不能保证信号能够连续延拓成周期信号序列，信号在数据块边缘的跳变会使信号的能量谱发散而不集中，从而产生大量的高频信号；另外，对FFT或DCT的系数进行量化编码，这样不可避免地产生量化误差，这种量化所带来的误差在合成音频信号时被综合窗作用后放大许多倍，使合成音频信号严重失真，即产生边界效应。Transform domain coding is one of the compression technologies commonly used in today's audio coding standards. It belongs to frequency domain coding. It reduces the correlation between components in the signal and quantizes and codes the transformed coefficients to achieve the purpose of information compression. Make full use of the auditory characteristics of the human ear in the frequency domain, such as masking effects and critical frequency bands, to achieve compression of audio signals. In practical applications, the audio signal is usually divided into several independent data frames for FFT (Fast Fourier Transform, Fast Fourier Transform) or DCT (Discrete Cosine Transform, Discrete Cosine Transform), but it cannot be guaranteed that each frame signal is continuous at the edge , and there is no guarantee that the signal can be continuously extended into a periodic signal sequence. The jump of the signal at the edge of the data block will cause the energy spectrum of the signal to diverge and not concentrate, thereby generating a large number of high-frequency signals; in addition, the coefficients of FFT or DCT Quantization coding, which inevitably produces quantization errors, and the errors brought about by this quantization are amplified many times by the synthesis window when the audio signal is synthesized, which seriously distorts the synthesized audio signal, that is, produces boundary effects.

边界效应是由音频信号的数据帧间不连续造成的，使音频信号的自然度和可懂度受到严重的影响，影响了编码器的效果，使音频质量严重下降；并且使音频信号听起来带有明显的周期性的“嘟嘟”噪声，在频谱图上表现为：出现明显的间隔性的“噪声竖线”。The boundary effect is caused by the discontinuity between data frames of the audio signal, which seriously affects the naturalness and intelligibility of the audio signal, affects the effect of the encoder, and seriously reduces the audio quality; and makes the audio signal sound There are obvious periodic "beep" noises, which are manifested in the spectrogram: obvious interval "noise vertical lines".

现有技术中，为了消除边界效应，通常采用MDCT(Modified DiscreteCosine，修正离散余弦变换)作为时频变换工具，利用50％的样点重叠和时域混叠消除滤波器组，在不降低变换编码性能的情况下，克服FFT、DCT处理运算中的边界效应。相对于DCT而言，MDCT采取了50％的数据重叠技术，即：当前数据块的前一半数据与其前相邻数据块的后一半数据重叠，而后一半数据与其后相邻数据块的前一半数据重叠。In the prior art, in order to eliminate boundary effects, MDCT (Modified DiscreteCosine, Modified Discrete Cosine Transform) is usually used as a time-frequency transformation tool, using 50% sample point overlap and time-domain aliasing elimination filter bank, without reducing the transformation coding In the case of performance, overcome the boundary effect in FFT and DCT processing operations. Compared with DCT, MDCT adopts a 50% data overlapping technology, that is, the first half of the data of the current data block overlaps the second half of the data of the previous adjacent data block, and the second half of the data overlaps with the first half of the next adjacent data block. overlapping.

MDCT变换的正变换定义如下：The forward transformation of the MDCT transformation is defined as follows:

${\mathrm{<span><span>X</span>x</span>}}_{\mathrm{<span><span>i</span>i</span>}}<span><span>(</span>(</span>\mathrm{<span><span>k</span>k</span>}<span><span>)</span>)</span><span><span>=</span>=</span>\underset{\mathrm{<span><span>n</span>no</span>}<span><span>=</span>=</span>\mathrm{<span><span>0</span>0</span>}}{\overset{\mathrm{<span><span>N</span>N</span>}<span><span>-</span>-</span>\mathrm{<span><span>1</span>1</span>}}{\mathrm{<span><span>\&Sigma;</span>\&Sigma;</span>}}}{\mathrm{<span><span>w</span>w</span>}}_{\mathrm{<span><span>a</span>a</span>}}<span><span>(</span>(</span>\mathrm{<span><span>n</span>no</span>}<span><span>)</span>)</span>{\mathrm{<span><span>x</span>x</span>}}_{\mathrm{<span><span>i</span>i</span>}}<span><span>(</span>(</span>\mathrm{<span><span>n</span>no</span>}<span><span>)</span>)</span>\mathrm{<span><span>cos</span>cos</span>}<span><span>[</span>[</span>\frac{<span><span>(</span>(</span>\mathrm{<span><span>2</span>2</span>}\mathrm{<span><span>k</span>k</span>}<span><span>+</span>+</span>\mathrm{<span><span>1</span>1</span>}<span><span>)</span>)</span>\mathrm{<span><span>\&pi;</span>\&pi;</span>}}{\mathrm{<span><span>N</span>N</span>}}<span><span>(</span>(</span>\mathrm{<span><span>n</span>no</span>}<span><span>+</span>+</span>{\mathrm{<span><span>n</span>no</span>}}_{\mathrm{<span><span>0</span>0</span>}}<span><span>)</span>)</span><span><span>]</span>]</span><span><span>,</span>,</span>\mathrm{<span><span>k</span>k</span>}<span><span>=</span>=</span>\mathrm{<span><span>0,1</span>0,1</span>}<span><span>,</span>,</span><span><span>\&CenterDot;</span>\&Center\; Dot;</span><span><span>\&CenterDot;</span>\&Center\; Dot;</span><span><span>\&CenterDot;</span>\&Center\; Dot;</span>\frac{\mathrm{<span><span>N</span>N</span>}}{\mathrm{<span><span>2</span>2</span>}}<span><span>-</span>-</span>\mathrm{<span><span>1</span>1</span>}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>1</span>1</span>}<span><span>)</span>)</span>$

其中，$n_0=\frac{N/2+1}{2}$是MDCT的相位变量。从MDCT定义可知，进行变换的数据块长度N必须为偶数，MDCT是用N个音频时域样本变换得到N/2个频域样本。in,${\mathrm{no}}_0=\frac{N/2+1}{2}$ is the phase variable of MDCT. From the definition of MDCT, it can be seen that the length N of the data block to be transformed must be an even number, and MDCT uses N audio time domain samples to transform to obtain N/2 frequency domain samples.

MDCT变换的逆变换定义如下：The inverse transform of the MDCT transform is defined as follows:

${\mathrm{<span><span>y</span>the\; y</span>}}_{\mathrm{<span><span>i</span>i</span>}}<span><span>(</span>(</span>\mathrm{<span><span>n</span>no</span>}<span><span>)</span>)</span><span><span>=</span>=</span>{\mathrm{<span><span>w</span>w</span>}}_{\mathrm{<span><span>s</span>the\; s</span>}}<span><span>(</span>(</span>\mathrm{<span><span>n</span>no</span>}<span><span>)</span>)</span>\frac{\mathrm{<span><span>4</span>4</span>}}{\mathrm{<span><span>N</span>N</span>}}\underset{\mathrm{<span><span>n</span>no</span>}<span><span>=</span>=</span>\mathrm{<span><span>0</span>0</span>}}{\overset{\mathrm{<span><span>N</span>N</span>}<span><span>/</span>/</span>\mathrm{<span><span>2</span>2</span>}<span><span>-</span>-</span>\mathrm{<span><span>1</span>1</span>}}{\mathrm{<span><span>\&Sigma;</span>\&Sigma;</span>}}}{\mathrm{<span><span>X</span>x</span>}}_{\mathrm{<span><span>i</span>i</span>}}<span><span>(</span>(</span>\mathrm{<span><span>k</span>k</span>}<span><span>)</span>)</span>\mathrm{<span><span>cos</span>cos</span>}<span><span>[</span>[</span>\frac{<span><span>(</span>(</span>\mathrm{<span><span>2</span>2</span>}\mathrm{<span><span>k</span>k</span>}<span><span>+</span>+</span>\mathrm{<span><span>1</span>1</span>}<span><span>)</span>)</span>\mathrm{<span><span>\&pi;</span>\&pi;</span>}}{\mathrm{<span><span>N</span>N</span>}}<span><span>(</span>(</span>\mathrm{<span><span>n</span>no</span>}<span><span>+</span>+</span>{\mathrm{<span><span>n</span>no</span>}}_{\mathrm{<span><span>0</span>0</span>}}<span><span>)</span>)</span><span><span>]</span>]</span><span><span>,</span>,</span>\mathrm{<span><span>n</span>no</span>}<span><span>=</span>=</span>\mathrm{<span><span>0,1</span>0,1</span>}<span><span>,</span>,</span><span><span>\&CenterDot;</span>\&Center\; Dot;</span><span><span>\&CenterDot;</span>\&Center\; Dot;</span><span><span>\&CenterDot;</span>\&CenterDot;</span>\frac{\mathrm{<span><span>N</span>N</span>}}{\mathrm{<span><span>2</span>2</span>}}<span><span>-</span>-</span>\mathrm{<span><span>1</span>1</span>}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>2</span>2</span>}<span><span>)</span>)</span>$

MDCT逆变换是用N/2个频域信号样本计算N个时域音频样本。The MDCT inverse transform uses N/2 frequency domain signal samples to calculate N time domain audio samples.

当信号样本被分成相对独立的数据帧后进行时频变换处理，在数据块的边缘会发生畸变，解决这一有效的方法是在相邻数据帧间采取数据重叠技术。通过上述可知，MDCT采用50％的数据重叠，并且利用分析、综合窗w_a(n)、w_s(n)进一步减弱了数据帧间的不连续性。因此，MDCT一定程度上消减了边界效应，改善了编码音频的可懂度，提高了编码质量。When the signal samples are divided into relatively independent data frames and then undergo time-frequency transformation processing, distortion will occur at the edge of the data block. An effective method to solve this problem is to adopt data overlapping technology between adjacent data frames. From the above, it can be seen that MDCT uses 50% data overlap, and uses analysis and synthesis windows w_a (n) and w_s (n) to further weaken the discontinuity between data frames. Therefore, MDCT reduces the boundary effect to a certain extent, improves the intelligibility of encoded audio, and improves the encoding quality.

然而，MDCT系数不可避免地产生量化误差，这种误差会影响帧间的连续性，因此MDCT不能完全消除边界效应带来的影响，当信号变化比较剧烈，特别是处理能量较强的音频信号后的边界效应尤为明显；并且在处理多声道编码技术上，在解码端对能量进行重新分配会使各个声道不连续，出现更为严重的边界效应。However, MDCT coefficients inevitably produce quantization errors, which will affect the continuity between frames. Therefore, MDCT cannot completely eliminate the impact of boundary effects. When the signal changes sharply, especially after processing audio signals with strong energy The boundary effect is particularly obvious; and in dealing with multi-channel coding technology, redistribution of energy at the decoding end will make each channel discontinuous, resulting in more serious boundary effects.

发明内容Contents of the invention

本发明实施例提供一种修正音频信号的方法及装置，能够有效消除边界效应。Embodiments of the present invention provide a method and device for correcting an audio signal, which can effectively eliminate boundary effects.

一种修正音频信号的方法，包括：A method of modifying an audio signal comprising:

在音频信号数据帧边界附近截取一段信号；Intercept a section of signal near the boundary of audio signal data frame;

对截取的所述信号进行线性处理，得到新的信号；performing linear processing on the intercepted signal to obtain a new signal;

计算所述新的信号的测评指标，当所述测评指标小于预置的测评指标时，继续进行线性处理，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。Calculate the evaluation index of the new signal, and when the evaluation index is less than the preset evaluation index, continue the linear processing until the evaluation index of the linearly processed signal is greater than or equal to the preset evaluation index.

一种修正音频信号的装置，包括：A device for modifying an audio signal, comprising:

截取信号单元，用于在音频信号数据帧边界附近截取一段信号；The intercept signal unit is used to intercept a section of signal near the boundary of the audio signal data frame;

线性处理单元，用于对接收的信号进行线性处理，得到新的信号；a linear processing unit, configured to linearly process the received signal to obtain a new signal;

计算单元，用于计算所述新的信号的测评指标；a calculation unit, configured to calculate the evaluation index of the new signal;

比较单元，用于从计算单元接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性处理单元，直到接收的测评指标大于或等于所述预置的测评指标。A comparison unit, configured to receive the evaluation index from the calculation unit, compare the size of the evaluation index with a preset evaluation index, and send the new signal to the linear processing unit when the evaluation index is smaller than the preset evaluation index until the received evaluation index is greater than or equal to the preset evaluation index.

通过上述技术方案可知，由于音频信号的不连续发生在相邻数据帧边界附近，因此在音频信号数据帧边界附近截取一段信号，对截取的所述信号进行线性处理，用线性处理后的信号替换原来发生跳变或不连续的信号，并计算经过线性处理后的新的信号的测评指标，当经过线性处理后的新的信号的测评指标小于预置的测评指标时，继续进行线性处理，进一步降低相邻数据帧边界附近信号的不连续性，直到线性处理后的信号的测评指标大于或等于预置的测评指标，使得相邻数据帧边界附近的信号具有连续性，实现了消除边界效应的目的。It can be seen from the above technical solution that since the discontinuity of the audio signal occurs near the boundary of adjacent data frames, a section of signal is intercepted near the boundary of the audio signal data frame, the intercepted signal is linearly processed, and replaced by the linearly processed signal The original jump or discontinuous signal occurs, and the evaluation index of the new signal after linear processing is calculated. When the evaluation index of the new signal after linear processing is smaller than the preset evaluation index, linear processing is continued, and further Reduce the discontinuity of the signal near the border of adjacent data frames until the evaluation index of the linearly processed signal is greater than or equal to the preset evaluation index, so that the signal near the border of adjacent data frames has continuity and realizes the elimination of border effects. Purpose.

附图说明Description of drawings

图1为本发明实施例提供的方法流程图；Fig. 1 is the flow chart of the method provided by the embodiment of the present invention;

图2为本发明实施例提供的装置示意图；Fig. 2 is the schematic diagram of the device provided by the embodiment of the present invention;

图3为本发明实施例一提供的装置示意图；Fig. 3 is a schematic diagram of the device provided byEmbodiment 1 of the present invention;

图4为本发明实施例二提供的装置示意图；Fig. 4 is a schematic diagram of the device provided by Embodiment 2 of the present invention;

图5为本发明实施例三提供的装置示意图；Fig. 5 is a schematic diagram of the device provided by Embodiment 3 of the present invention;

图6为本发明实施例四提供的装置示意图；Fig. 6 is a schematic diagram of the device provided by Embodiment 4 of the present invention;

图7为本发明实施例五提供的装置示意图；Fig. 7 is a schematic diagram of the device provided by Embodiment 5 of the present invention;

图8为本发明实施例六提供的装置示意图；Fig. 8 is a schematic diagram of the device provided by Embodiment 6 of the present invention;

图9为本发明实施例七提供的装置示意图；Fig. 9 is a schematic diagram of the device provided by Embodiment 7 of the present invention;

图10为本发明实施例八提供的装置示意图。Fig. 10 is a schematic diagram of the device provided by Embodiment 8 of the present invention.

具体实施方式Detailed ways

本发明实施例提供了一种修正音频信号的方法及装置，用于修正音频信号相邻数据帧间的不连续性，使得修正后音频信号的波形平滑，进而实现消除边界效应的目的，为了使本发明实施例的技术方案更清楚，详细，下面列举实施例进行说明。Embodiments of the present invention provide a method and device for correcting an audio signal, which are used to correct the discontinuity between adjacent data frames of the audio signal, so that the waveform of the corrected audio signal is smooth, thereby achieving the purpose of eliminating boundary effects. The technical solutions of the embodiments of the present invention are clearer and more detailed, and the following examples are listed for illustration.

首先，对本发明实施例提供的方法进行总体描述。First, the method provided by the embodiment of the present invention is generally described.

参见图1，为本发明实施例提供的方法流程图：Referring to Fig. 1, the flow chart of the method provided for the embodiment of the present invention:

11)：在音频信号数据帧边界附近截取一段信号，例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；11): intercepting a section of signal near the border of the audio signal data frame, for example, X point is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, then starting from X point and intercepting 128 points forward respectively, to After intercepting 128 points, the intercepted 256 points form a frame signal;

12)：对截取的所述信号进行线性处理，得到新的音频信号；12): performing linear processing on the intercepted signal to obtain a new audio signal;

13)：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，判断所述测评指标小于预置的测评指标时，返回步骤12)，继续进行线性处理，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标，否则，结束线性处理。13): Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score, and when it is judged that the evaluation index is less than the preset evaluation index, return to step 12), and continue to perform linear processing until The evaluation index of the linearly processed signal is greater than or equal to the preset evaluation index; otherwise, the linear processing ends.

其中，所述预置的测评指标为消除边界效应所对应的音频信号的信噪比、频谱失真度、平均意见分。Wherein, the preset evaluation indicators are signal-to-noise ratio, spectral distortion degree, and average opinion score of the audio signal corresponding to eliminating the boundary effect.

下面针对本发明实施例提供的方法分别列举实施例进行详细说明：In the following, the methods provided by the embodiments of the present invention will be described in detail by enumerating the embodiments respectively:

实施例一：Embodiment one:

101：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；；101: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

102：对截取的信号进行LP(Linear Prediction，线性预测)分析，得到预测系数，然后利用公式$s\&prime;\left(n\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_{i}s(n-i)$进行线性预测，所得的预测值代替数据帧边界附近的跳变值，得到新的音频信号；102: Perform LP (Linear Prediction, linear prediction) analysis on the intercepted signal to obtain the prediction coefficient, and then use the formula$\mathrm{the\; s}\&prime;\left(\mathrm{no}\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_i\mathrm{the\; s}(\mathrm{no}-i)$ Carry out linear prediction, and the obtained predicted value replaces the jump value near the data frame boundary to obtain a new audio signal;

其中，s’(n)代表预测值，p代表预测阶数，a_i代表预测系数；Among them, s'(n) represents the predicted value, p represents the prediction order, and a_i represents the prediction coefficient;

其中，LP分析是最有效的语音分析技术之一，通过一个语音信号的抽样值可以用过去若干个取样值的线性组合来逼近，比较常用杜宾(Durbin)推算法求解线性预测系数。Among them, LP analysis is one of the most effective speech analysis techniques. The sampling value of a speech signal can be approximated by the linear combination of several sampling values in the past. The Durbin method is commonly used to solve the linear prediction coefficient.

103：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回102步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。103: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 102 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

该实施例中，对数据帧边界附近的音频信号通过LP计算得到相关性较大的数据来代替原来相关性较小的不连续或跳变的数据，使音频信号具有连续性。In this embodiment, the audio signal near the boundary of the data frame is calculated by LP to replace the original discontinuous or jumping data with less correlation, so that the audio signal has continuity.

实施例二：Embodiment two:

201：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；201: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

202：对截取的信号进行LP(Linear Prediction，线性预测)分析，得到预测系数，然后利用公式$s\&prime;\left(n\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_{i}s(n-i)$进行线性预测，所得的预测值代替数据帧边界附近的跳变值，得到新的音频信号；202: Perform LP (Linear Prediction, linear prediction) analysis on the intercepted signal to obtain the prediction coefficient, and then use the formula$\mathrm{the\; s}\&prime;\left(\mathrm{no}\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_i\mathrm{the\; s}(\mathrm{no}-i)$ Carry out linear prediction, and the obtained predicted value replaces the jump value near the data frame boundary to obtain a new audio signal;

其中，s’(n)代表预测值，p代表预测阶数，a_i代表预测系数；Among them, s'(n) represents the predicted value, p represents the prediction order, and a_i represents the prediction coefficient;

其中，LP分析是最有效的语音分析技术之一，通过一个语音信号的抽样值可以用过去若干个取样值的线性组合来逼近，比较常用杜宾(Durbin)推算法求解线性预测系数。Among them, LP analysis is one of the most effective speech analysis techniques. The sampling value of a speech signal can be approximated by the linear combination of several sampling values in the past. The Durbin method is commonly used to solve the linear prediction coefficient.

203：对新的音频信号的数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；以平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，进行线性内插，用内插的数据取代原来相应位置的数据，进一步获得新的音频信号；203: Take the average value of at least two data points before and after the data frame jump of the new audio signal, and use the data point and the average value to make a linear curve; take the average value as the reference point, or any value before the average value One point is the reference point, or any point after the average value is the reference point, and linear interpolation is performed, and the interpolated data is used to replace the data at the original corresponding position to further obtain a new audio signal;

204：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回202步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。204: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 202 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

实施例二与实施例一相比，在对音频信号进行线性预测，消除畸变的基础上，采用线性内插的方法，用线性内插的值取代原来相应位置的点，使得音频信号具有连续性，进一步保证了音频信号的不连续性。Embodiment 2 Compared withEmbodiment 1, on the basis of performing linear prediction on the audio signal and eliminating distortion, the method of linear interpolation is adopted, and the original corresponding point is replaced by the value of linear interpolation, so that the audio signal has continuity , further ensuring the discontinuity of the audio signal.

实施三：Implementation three:

301：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；301: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

302：对截取的信号进行LP(Linear Prediction，线性预测)分析，得到预测系数，然后利用公式$s\&prime;\left(n\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_{i}s(n-i)$进行线性预测，所得的预测值代替数据帧边界附近的跳变值，得到新的音频信号；302: Perform LP (Linear Prediction, linear prediction) analysis on the intercepted signal to obtain the prediction coefficient, and then use the formula$\mathrm{the\; s}\&prime;\left(\mathrm{no}\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_i\mathrm{the\; s}(\mathrm{no}-i)$ Carry out linear prediction, and the obtained predicted value replaces the jump value near the data frame boundary to obtain a new audio signal;

其中，s’(n)代表预测值，p代表预测阶数，a_i代表预测系数；Among them, s'(n) represents the predicted value, p represents the prediction order, and a_i represents the prediction coefficient;

其中，LP分析是最有效的语音分析技术之一，通过一个语音信号的抽样值可以用过去若干个取样值的线性组合来逼近，比较常用杜宾(Durbin)推算法求解线性预测系数。Among them, LP analysis is one of the most effective speech analysis techniques. The sampling value of a speech signal can be approximated by the linear combination of several sampling values in the past. The Durbin method is commonly used to solve the linear prediction coefficient.

303：对所述新的音频信号进行快速傅立叶变换，将时域变成频域；截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的音频信号，用所述新的信号代替快速傅立叶变换之前的信号；303: Perform fast Fourier transform on the new audio signal to change the time domain into the frequency domain; intercept the high-frequency part of the frequency domain, perform forward or backward misplaced addition and average, and use the The average value is used to replace the high-frequency part; the high-frequency part replaced by the average value is subjected to inverse fast Fourier transform to obtain a new audio signal, and the new signal is used to replace the signal before the fast Fourier transform;

304：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回302步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。304: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 302 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

该实施例与实施例二相比，将实施例二中线性内插的方法替换为频域平滑的方法修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。Compared with the second embodiment, this embodiment replaces the linear interpolation method in the second embodiment with a frequency domain smoothing method to correct the spectrum divergence, smooth the spectrum of the audio signal, and achieve the purpose of eliminating the boundary effect.

实施例四：Embodiment four:

401：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；401: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

402：对截取的信号进行LP(Linear Prediction，线性预测)分析，得到预测系数，然后利用公式$s\&prime;\left(n\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_{i}s(n-i)$进行线性预测，所得的预测值代替数据帧边界附近的跳变值，得到新的音频信号；402: Perform LP (Linear Prediction, linear prediction) analysis on the intercepted signal to obtain the prediction coefficient, and then use the formula$\mathrm{the\; s}\&prime;\left(\mathrm{no}\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_i\mathrm{the\; s}(\mathrm{no}-i)$ Carry out linear prediction, and the obtained predicted value replaces the jump value near the data frame boundary to obtain a new audio signal;

其中，s’(n)代表预测值，p代表预测阶数，a_i代表预测系数；Among them, s'(n) represents the predicted value, p represents the prediction order, and a_i represents the prediction coefficient;

其中，LP分析是最有效的语音分析技术之一，通过一个语音信号的抽样值可以用过去若干个取样值的线性组合来逼近，比较常用杜宾(Durbin)推算法求解线性预测系数。Among them, LP analysis is one of the most effective speech analysis techniques. The sampling value of a speech signal can be approximated by the linear combination of several sampling values in the past. The Durbin method is commonly used to solve the linear prediction coefficient.

403：对新的音频信号的数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；以平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点进行线性内插，用内插的数据取代原来相应位置的数据，进一步获得新的音频信号；403: Take the average value of at least two data points before and after the data frame jump of the new audio signal, and use the data point and the average value to make a linear curve; take the average value as the reference point, or any value before the average value One point is the reference point, or any point after the average value is used as the reference point for linear interpolation, and the interpolated data is used to replace the data at the original corresponding position to further obtain a new audio signal;

404：对经过线性内插后获得的音频信号进行快速傅立叶变换，将时域变成频域；截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的音频信号，用所述新的信号代替快速傅立叶变换之前的信号；404: performing fast Fourier transform on the audio signal obtained after linear interpolation, changing the time domain into the frequency domain; intercepting the high-frequency part of the frequency domain, and performing forward or backward misplaced addition to calculate the average value, Replacing the high-frequency part with the average value; performing an inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new audio signal, and using the new signal to replace the signal before the fast Fourier transform;

405：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回402步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。405: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 402 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

该实施例将实施例二和实施例三项结合，在对音频信号进行线性内插的基础上，进一步采用频域平滑的方法修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。In this embodiment, the second embodiment and the third embodiment are combined. On the basis of linear interpolation of the audio signal, the frequency domain smoothing method is further used to correct the spectrum divergence, so that the frequency spectrum of the audio signal is smoothed, and the purpose of eliminating the boundary effect is achieved. .

实施五：Implementation five:

501：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；501: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

502：在截取的信号中从前向后截取一段信号设置为奇数标志，从后向前截取一段信号设置为偶数标志，对奇数标志的信号和偶数标志的信号进行线性预测获取奇数标志的信号的预测值和偶数标志的信号的预测值，并且对偶数标志的信号的预测值再进行前后倒置，对奇数标志的信号的预测值，和经过前后倒置的偶数标志的信号的预测值取平均值，用所述平均值代替数据帧边界附近的跳变值；502: In the intercepted signal, intercept a section of signal from front to back and set it as an odd sign, intercept a section of signal from back to front and set it as an even sign, perform linear prediction on the signal of odd sign and the signal of even sign to obtain the prediction of the signal of odd sign value and the predicted value of the even-numbered signal, and then invert the predicted value of the even-numbered signal, take the average value of the predicted value of the odd-numbered signal and the predicted value of the even-numbered signal after the inversion, and use The average value replaces the jump value near the data frame boundary;

503：对截取的信号进行LP(Linear Prediction，线性预测)分析，得到预测系数，然后利用公式$s\&prime;\left(n\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_{i}s(n-i)$进行线性预测，所得的预测值代替数据帧边界附近的跳变值，得到新的音频信号；503: Perform LP (Linear Prediction, linear prediction) analysis on the intercepted signal to obtain the prediction coefficient, and then use the formula$\mathrm{the\; s}\&prime;\left(\mathrm{no}\right)=\underset{i=0}{\overset{p}{\mathrm{\&Sigma;}}}{a}_i\mathrm{the\; s}(\mathrm{no}-i)$ Carry out linear prediction, and the obtained predicted value replaces the jump value near the data frame boundary to obtain a new audio signal;

其中，s’(n)代表预测值，p代表预测阶数，a_i代表预测系数；Among them, s'(n) represents the predicted value, p represents the prediction order, and a_i represents the prediction coefficient;

其中，LP分析是最有效的语音分析技术之一，通过一个语音信号的抽样值可以用过去若干个取样值的线性组合来逼近，比较常用杜宾(Durbin)推算法求解线性预测系数。Among them, LP analysis is one of the most effective speech analysis techniques. The sampling value of a speech signal can be approximated by the linear combination of several sampling values in the past. The Durbin method is commonly used to solve the linear prediction coefficient.

504：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回502步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。504: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectrum distortion, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 502 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

该实施例与实施例一相比，在利用对数据帧边界附近的音频信号进行线性预测中增加了奇偶标志的方法，保证线性预测的准确性，进而更好的修正音频信号，使音频信号具有连续性。Compared withEmbodiment 1, this embodiment uses the method of adding parity marks in the linear prediction of the audio signal near the border of the data frame to ensure the accuracy of the linear prediction, and then better correct the audio signal, so that the audio signal has continuity.

其中，该实施例中的步骤502可分别位于实施例二中的202之前构成一个实施例；位于实施例三中的302之前构成一个实施例；位于实施例四中的402之前构成一个实施例，处理过程同实施例五。Wherein, thestep 502 in this embodiment can be respectively located before 202 in the second embodiment to constitute an embodiment; before 302 in the third embodiment constitutes an embodiment; before 402 in the fourth embodiment constitutes an embodiment, The processing procedure is the same as in Embodiment five.

实施例六：Embodiment six:

601：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；601: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

602：对音频信号的数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；以平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据，获得新的音频信号；602: Take the average value of at least two data points before and after the data frame jump of the audio signal, and use the data points and the average value to make a linear curve; take the average value as the reference point, or any point before the average value as a reference point, or any point after the average value as a reference point, interpolate data on the linear curve, replace the data at the original corresponding position with the interpolated data, and obtain a new audio signal;

603：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回602步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。603: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 602 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

该实施例与实施例一相比，将实施例一中的线性预测的方法，替换为线性内插的方法，用线性内插的值取代原来相应位置的点，使得音频信号具有连续性，并且在频域上修正了由于数据不连续造成的锯齿波。Compared withEmbodiment 1, this embodiment replaces the linear prediction method inEmbodiment 1 with a linear interpolation method, and replaces the original corresponding position points with linear interpolation values, so that the audio signal has continuity, and The sawtooth wave caused by data discontinuity is corrected in the frequency domain.

实施例七：Embodiment seven:

701：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；701: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

702：对音频信号的数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；以平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据，获得经过线性内插的音频信号；702: Take the average value of at least two data points before and after the data frame jump of the audio signal, and use the data points and the average value to make a linear curve; take the average value as the reference point, or any point before the average value be a reference point, or any point after the average value is a reference point, interpolate data on the linear curve, replace the data at the original corresponding position with the interpolated data, and obtain a linearly interpolated audio signal;

703：对经过线性内插的音频信号进行快速傅立叶变换，将时域变成频域；截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的音频信号，用所述新的信号代替快速傅立叶变换之前的信号；703: Perform fast Fourier transform on the linearly interpolated audio signal to change the time domain into the frequency domain; intercept the high-frequency part of the frequency domain, perform forward or backward dislocation addition and average, and use the The average value is used to replace the high-frequency part; the high-frequency part replaced by the average value is subjected to inverse fast Fourier transform to obtain a new audio signal, and the new signal is used to replace the signal before the fast Fourier transform;

704：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回702步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。704: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 702 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

该实施例与实施例六相比，在对音频信号进行线性内插的基础上，进一步采用频域平滑的方法修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。Compared with the sixth embodiment, in this embodiment, on the basis of linear interpolation of the audio signal, the frequency domain smoothing method is further used to correct the spectrum divergence, so as to smooth the spectrum of the audio signal and achieve the purpose of eliminating the boundary effect.

实施例八：Embodiment eight:

801：在音频信号数据帧边界附近截取一段信号；例如，X点为信号1的边界点，由于一般语谱图的帧长为256点，则从X点开始分别向前截取128点，向后截取128点，截取的256点组成一帧信号；801: Intercept a section of signal near the boundary of the audio signal data frame; for example, point X is the boundary point ofsignal 1, since the frame length of the general spectrogram is 256 points, start from point X to intercept 128 points forward and backward Intercept 128 points, and the intercepted 256 points form a frame signal;

802：对接收的音频信号进行快速傅立叶变换，将时域变成频域；截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的音频信号，用所述新的信号代替快速傅立叶变换之前的信号；802: Perform fast Fourier transform on the received audio signal to change the time domain into the frequency domain; intercept the high-frequency part of the frequency domain, perform forward or backward dislocation addition to obtain an average value, and use the average value Replacing the high-frequency part; performing an inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new audio signal, and using the new signal to replace the signal before the fast Fourier transform;

803：计算新的音频信号的测评指标，如，信噪比、频谱失真度、平均意见分，当所述测评指标小于预置的测评指标时，返回802步骤，继续进行线性预测，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。803: Calculate the evaluation index of the new audio signal, such as signal-to-noise ratio, spectral distortion degree, and average opinion score. When the evaluation index is less than the preset evaluation index, return to step 802 and continue to perform linear prediction until linear processing The evaluation index of the final signal is greater than or equal to the preset evaluation index.

该实施例与实施例一相比，将通过线性预测消除畸变的方法替换为频域平滑的方法修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。Compared withEmbodiment 1, this embodiment replaces the method of eliminating distortion by linear prediction with the method of frequency domain smoothing to correct spectrum divergence, so as to smooth the frequency spectrum of the audio signal and achieve the purpose of eliminating boundary effects.

其中，上述各实施例中的步骤之间并无严格的时序关序，各标号只是代表实现本发明实施例的过程。Wherein, there is no strict time sequence between the steps in the above-mentioned embodiments, and each label only represents the process of realizing the embodiment of the present invention.

下面参照附图，对本发明实施例提供的装置进行详细说明：The device provided by the embodiment of the present invention is described in detail below with reference to the accompanying drawings:

参见图2，为本法实施例的装置示意图，包括：Referring to Fig. 2, it is the device schematic diagram of this method embodiment, comprises:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

线性处理单元202，用于对接收的信号进行线性处理，得到新的信号；alinear processing unit 202, configured to perform linear processing on the received signal to obtain a new signal;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性处理单元202，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the linear system when the evaluation index is smaller than the preset evaluation index. Theprocessing unit 202, until the received evaluation index is greater than or equal to the preset evaluation index.

上述是对本发明实施例提供的装置示意图的总体描述，下面分别列举实施例进行详细描述：The above is a general description of the schematic diagram of the device provided by the embodiment of the present invention, and the following examples are listed for detailed description:

参见图3，为本发明实施例一提供的装置示意图，包括：Referring to Figure 3, it is a schematic diagram of the device provided byEmbodiment 1 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

线性预测单元301，用于对接收的信号进行线性预测，获取预测值；Alinear prediction unit 301, configured to linearly predict the received signal to obtain a predicted value;

替换单元302，用于将接收的预测值代替数据帧边界附近的跳变值，得到新的信号；Thereplacement unit 302 is used to replace the jump value near the boundary of the data frame with the received predicted value to obtain a new signal;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性预测单元301，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the linear system when the evaluation index is smaller than the preset evaluation index. Theprediction unit 301, until the received evaluation index is greater than or equal to the preset evaluation index.

其中，线性预测单元301，替换单元302，置于线性处理单元202中。Wherein, thelinear prediction unit 301 and thereplacement unit 302 are placed in thelinear processing unit 202 .

参见图4，为本发明实施例二提供的装置示意图，包括：Referring to Figure 4, it is a schematic diagram of the device provided by Embodiment 2 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

线性预测单元301，用于对接收的信号进行线性预测，获取预测值；Alinear prediction unit 301, configured to linearly predict the received signal to obtain a predicted value;

替换单元302，用于将接收的预测值代替数据帧边界附近的跳变值，得到新的信号。Thereplacement unit 302 is configured to replace the jump value near the boundary of the data frame with the received prediction value to obtain a new signal.

绘制单元401，用于对数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；Adrawing unit 401, configured to average at least two data points before and after the data frame jump, and use the data points and the average value to make a linear curve;

线性内插单元402，用于以所述平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据，获得新的信号；Alinear interpolation unit 402, configured to use the average value as a reference point, or any point before the average value as a reference point, or any point after the average value as a reference point, and interpolate data on the linear curve , replace the data at the original corresponding position with the interpolated data to obtain a new signal;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性线性预测单元301，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the linear system when the evaluation index is smaller than the preset evaluation index. Thelinear prediction unit 301, until the received evaluation index is greater than or equal to the preset evaluation index.

其中，线性预测单元301、替换单元302、绘制单元401、线性内插单元402置于线性处理单元202中。Wherein, thelinear prediction unit 301 , thereplacement unit 302 , therendering unit 401 , and thelinear interpolation unit 402 are placed in thelinear processing unit 202 .

实施例二提供的示意图与实施例一提供的示意图相比，增加了绘制单元401和线性内插单元402，用于对音频信号进行线性预测消除畸变的基础上，进一步修正音频信号的不连续性。Compared with the schematic diagram provided inEmbodiment 1, the schematic diagram provided in Embodiment 2 adds adrawing unit 401 and alinear interpolation unit 402, which are used to further correct the discontinuity of the audio signal on the basis of performing linear prediction on the audio signal and eliminating distortion. .

参见图5，为本发明实施例三提供的装置示意图，包括：Referring to Figure 5, it is a schematic diagram of the device provided by Embodiment 3 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

线性预测单元301，用于对接收的信号进行线性预测，获取预测值；Alinear prediction unit 301, configured to linearly predict the received signal to obtain a predicted value;

替换单元302，用于将接收的预测值代替数据帧边界附近的跳变值，得到新的信号；Thereplacement unit 302 is used to replace the jump value near the boundary of the data frame with the received predicted value to obtain a new signal;

傅立叶变换单元501，用于对所述新的信号进行快速傅立叶变换，将时域变成频域；AFourier transform unit 501, configured to perform fast Fourier transform on the new signal to change the time domain into the frequency domain;

频域平滑单元502，用于截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；The frequencydomain smoothing unit 502 is configured to intercept the high frequency part of the frequency domain, perform forward or backward misplaced addition and average, and replace the high frequency part with the average value;

傅立叶逆变换单元503，用于对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用傅立叶变换后的新的信号代替快速傅立叶变换之前的信号；The inverseFourier transform unit 503 is used to perform inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new signal, and replace the signal before the fast Fourier transform with the new signal after the Fourier transform;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性预测单元301，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the linear system when the evaluation index is smaller than the preset evaluation index. Theprediction unit 301, until the received evaluation index is greater than or equal to the preset evaluation index.

其中，线性预测单元301、替换单元302、傅立叶变换单元501、频域平滑单元502、傅立叶逆变换单元503置于线性处理单元202中。Wherein, thelinear prediction unit 301 , thereplacement unit 302 , theFourier transform unit 501 , the frequencydomain smoothing unit 502 , and the inverseFourier transform unit 503 are placed in thelinear processing unit 202 .

该实施例提供的装置示意图与实施例二提供的装置示意图相比，将实施例二中的绘制单元401和线性内插单元402替换为傅立叶变换单元501、频域平滑单元502、傅立叶逆变换单元503，用于修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。Compared with the schematic diagram of the device provided in the second embodiment, the schematic diagram of the device provided in this embodiment is replaced by thedrawing unit 401 and thelinear interpolation unit 402 in the second embodiment with aFourier transform unit 501, a frequencydomain smoothing unit 502, and an inverseFourier transform unit 503, for correcting spectrum divergence, smoothing the spectrum of the audio signal, and achieving the purpose of eliminating boundary effects.

参见图6，为本发明实施例四提供的装置示意图，包括：Referring to Figure 6, it is a schematic diagram of the device provided by Embodiment 4 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

线性预测单元301，用于对接收的信号进行线性预测，获取预测值；Alinear prediction unit 301, configured to linearly predict the received signal to obtain a predicted value;

替换单元302，用于将接收的预测值代替数据帧边界附近的跳变值，得到新的信号；Thereplacement unit 302 is used to replace the jump value near the boundary of the data frame with the received predicted value to obtain a new signal;

绘制单元401，用于对数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；Adrawing unit 401, configured to average at least two data points before and after the data frame jump, and use the data points and the average value to make a linear curve;

线性内插单元402，用于以所述平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据，获得新的信号；Alinear interpolation unit 402, configured to use the average value as a reference point, or any point before the average value as a reference point, or any point after the average value as a reference point, and interpolate data on the linear curve , replace the data at the original corresponding position with the interpolated data to obtain a new signal;

傅立叶变换单元501，用于对所述新的信号进行快速傅立叶变换，将时域变成频域；AFourier transform unit 501, configured to perform fast Fourier transform on the new signal to change the time domain into the frequency domain;

频域平滑单元502，用于截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；The frequencydomain smoothing unit 502 is configured to intercept the high frequency part of the frequency domain, perform forward or backward misplaced addition and average, and replace the high frequency part with the average value;

傅立叶逆变换单元503，用于对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用傅立叶变换后的新的信号代替快速傅立叶变换之前的信号；The inverseFourier transform unit 503 is used to perform inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new signal, and replace the signal before the fast Fourier transform with the new signal after the Fourier transform;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性预测单元301，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the linear system when the evaluation index is smaller than the preset evaluation index. Theprediction unit 301, until the received evaluation index is greater than or equal to the preset evaluation index.

其中，线性预测单元301、替换单元302、绘制单元401、线性内插单元402、傅立叶变换单元501、频域平滑单元502、傅立叶逆变换单元503置于线性处理单元202中。Among them, thelinear prediction unit 301 , thereplacement unit 302 , therendering unit 401 , thelinear interpolation unit 402 , theFourier transform unit 501 , the frequencydomain smoothing unit 502 , and the inverseFourier transform unit 503 are placed in thelinear processing unit 202 .

该实施例提供的装置示意图相比实施例一提供的装置示意图，增加了绘制单元401、线性内插单元402、傅立叶变换单元501、频域平滑单元502和傅立叶逆变换单元503，用于在对音频信号进行线性内插的基础上，进一步采用频域平滑的方法修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。Compared with the schematic diagram of the device provided inEmbodiment 1, the schematic diagram of the device provided in this embodiment adds adrawing unit 401, alinear interpolation unit 402, aFourier transform unit 501, a frequencydomain smoothing unit 502, and an inverseFourier transform unit 503 for On the basis of linear interpolation of the audio signal, the frequency domain smoothing method is further used to correct the spectrum divergence, so as to smooth the spectrum of the audio signal and achieve the purpose of eliminating the boundary effect.

参见图7，为本发明实施例五提供的装置示意图，包括：Referring to Figure 7, it is a schematic diagram of the device provided by Embodiment 5 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

线性预测单元301，用于对接收的信号进行线性预测，获取预测值；Alinear prediction unit 301, configured to linearly predict the received signal to obtain a predicted value;

奇偶标志单元701，用于从截取信号单元201接收信号，对所述信号从前向后截取一段信号设置为奇数标志，并发送到所述线性预测单元301，获取奇数标志的信号的预测值；从后向前截取一段信号设置为偶数标志；Theparity flag unit 701 is used to receive the signal from the interceptingsignal unit 201, intercept a segment of the signal from front to back and set it as an odd number flag, and send it to thelinear prediction unit 301 to obtain the predicted value of the signal of the odd number flag; Intercept a section of signal backward and forward and set it as an even sign;

前后倒置单元702，用于从所述线性预测单元301接收到偶数标志的信号的预测值时，进行前向倒置；A front-back inversion unit 702, configured to perform forward inversion when receiving the predicted value of the even-numbered signal signal from thelinear prediction unit 301;

平均值单元703，用于接收所述奇数标志的信号的预测值，及经过前后倒置的偶数标志的信号的预测值，并对所述奇数标志的信号的预测值和经过前后倒置的偶数标志的信号的预测值取平均值，获得截取信号单元截取信号的预测值。Theaverage value unit 703 is configured to receive the predicted value of the signal of the odd number sign and the predicted value of the signal of the even number sign that have undergone front-to-back inversion, and calculate the predicted value of the signal of the odd number sign and the signal of the even number sign after the front-back inversion The predicted value of the signal is averaged to obtain the predicted value of the signal intercepted by the intercepting signal unit.

替换单元302，用于将接收的预测值代替数据帧边界附近的跳变值，得到新的信号；Thereplacement unit 302 is used to replace the jump value near the boundary of the data frame with the received predicted value to obtain a new signal;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性预测单元301，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from the calculation unit, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the linear predictor when the evaluation index is smaller than the presetevaluation index Unit 301, until the received evaluation index is greater than or equal to the preset evaluation index.

其中，线性预测单元301、替换单元302、绘制单元401、线性内插单元402、奇偶标志单元701、前后倒置单元702、平均值单元703置于线性处理单元202中。Among them, thelinear prediction unit 301 , thereplacement unit 302 , therendering unit 401 , thelinear interpolation unit 402 , theparity flag unit 701 , the front-back inversion unit 702 , and theaverage value unit 703 are placed in thelinear processing unit 202 .

其中，该实施例中提供的装置示意图中的奇偶标志单元701、前后倒置单元702、平均值单元703可分别与实施例二提供的装置、施例三提供的装置、实施例四提供的装置，进一步构成新的装置。Wherein, theparity flag unit 701, front-back inversion unit 702, andaverage value unit 703 in the schematic diagram of the device provided in this embodiment can be respectively compared with the device provided in Embodiment 2, the device provided in Embodiment 3, and the device provided in Embodiment 4, Further constitute a new device.

该实施例提供的装置示意图与实施例一提供的装置示意图相比，增加了奇偶标志单元701、前后倒置单元702、平均值单元703，用于保证线性预测的准确性，进而更好的修正音频信号，使音频信号具有连续性。Compared with the schematic diagram of the device provided inEmbodiment 1, the schematic diagram of the device provided in this embodiment adds aparity flag unit 701, a front-to-back inversion unit 702, and anaverage value unit 703 to ensure the accuracy of linear prediction and better audio correction. signal, making the audio signal continuous.

参见图8，为本发明实施例六提供的装置示意图，包括：Referring to Figure 8, it is a schematic diagram of the device provided by Embodiment 6 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

绘制单元401，用于对数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；Adrawing unit 401, configured to average at least two data points before and after the data frame jump, and use the data points and the average value to make a linear curve;

线性内插单元402，用于以所述平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据，获得新的信号；Alinear interpolation unit 402, configured to use the average value as a reference point, or any point before the average value as a reference point, or any point after the average value as a reference point, and interpolate data on the linear curve , replace the data at the original corresponding position with the interpolated data to obtain a new signal;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到绘制单元401，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the drawing when the evaluation index is smaller than the preset evaluation index.Unit 401, until the received evaluation index is greater than or equal to the preset evaluation index.

其中，绘制单元401、线性内插单元402置于线性处理单元202中。Wherein, therendering unit 401 and thelinear interpolation unit 402 are placed in thelinear processing unit 202 .

该实施例提供的示意图与实施例一提供的示意图相比，将线性预测单元301，替换单元302，替换为绘制单元401和线性内插单元402，用于修正音频信号的不连续性。Compared with the schematic diagram provided inEmbodiment 1, the schematic diagram provided in this embodiment replaces thelinear prediction unit 301 and thereplacement unit 302 with arendering unit 401 and alinear interpolation unit 402 for correcting the discontinuity of the audio signal.

参见图9，为本发明实施例七提供的装置示意图，包括：Referring to Figure 9, it is a schematic diagram of the device provided by Embodiment 7 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

绘制单元401，用于对数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；Adrawing unit 401, configured to average at least two data points before and after the data frame jump, and use the data points and the average value to make a linear curve;

线性内插单元402，用于以所述平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据，获得新的信号；Alinear interpolation unit 402, configured to use the average value as a reference point, or any point before the average value as a reference point, or any point after the average value as a reference point, and interpolate data on the linear curve , replace the data at the original corresponding position with the interpolated data to obtain a new signal;

傅立叶变换单元501，用于对所述新的信号进行快速傅立叶变换，将时域变成频域；AFourier transform unit 501, configured to perform fast Fourier transform on the new signal to change the time domain into the frequency domain;

频域平滑单元502，用于截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；A frequencydomain smoothing unit 502, configured to intercept the high frequency part of the frequency domain, perform forward or backward misplaced addition and average, and replace the high frequency part with the average value;

傅立叶逆变换单元503，用于对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用傅立叶变换后的新的信号代替快速傅立叶变换之前的信号；The inverseFourier transform unit 503 is used to perform inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new signal, and replace the signal before the fast Fourier transform with the new signal after the Fourier transform;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到绘制单元401，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to the drawing when the evaluation index is smaller than the preset evaluation index.Unit 401, until the received evaluation index is greater than or equal to the preset evaluation index.

其中，绘制单元401、线性内插单元402、傅立叶变换单元501、频域平滑单元502和傅立叶逆变换单元503置于线性处理单元202中。Wherein, therendering unit 401 , thelinear interpolation unit 402 , theFourier transform unit 501 , the frequencydomain smoothing unit 502 and the inverseFourier transform unit 503 are placed in thelinear processing unit 202 .

该实施例提供的装置示意图相比实施例一提供的装置示意图，将线性预测单元301，替换单元302，替换为绘制单元401、线性内插单元402、傅立叶变换单元501、频域平滑单元502和傅立叶逆变换单元503，用于在对音频信号进行线性内插的基础上，进一步采用频域平滑的方法修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。Compared with the schematic diagram of the device provided inEmbodiment 1, the schematic diagram of the device provided in this embodiment replaces thelinear prediction unit 301 and thereplacement unit 302 with arendering unit 401, alinear interpolation unit 402, aFourier transform unit 501, a frequencydomain smoothing unit 502 and The inverseFourier transform unit 503 is used to correct the spectrum divergence by further adopting a frequency domain smoothing method on the basis of linear interpolation of the audio signal, so as to smooth the spectrum of the audio signal and achieve the purpose of eliminating boundary effects.

参见图10，为本发明实施例八提供的装置示意图，包括：Referring to Figure 10, it is a schematic diagram of the device provided by Embodiment 8 of the present invention, including:

截取信号单元201，用于在音频信号数据帧边界附近截取一段信号；An interceptingsignal unit 201, configured to intercept a section of signal near the audio signal data frame boundary;

傅立叶变换单元501，用于对所述新的信号进行快速傅立叶变换，将时域变成频域；AFourier transform unit 501, configured to perform fast Fourier transform on the new signal to change the time domain into the frequency domain;

频域平滑单元502，用于截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；A frequencydomain smoothing unit 502, configured to intercept the high frequency part of the frequency domain, perform forward or backward misplaced addition and average, and replace the high frequency part with the average value;

傅立叶逆变换单元503，用于对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用傅立叶变换后的新的信号代替快速傅立叶变换之前的信号；The inverseFourier transform unit 503 is used to perform inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new signal, and replace the signal before the fast Fourier transform with the new signal after the Fourier transform;

计算单元203，用于计算所述新的信号的测评指标，如，信噪比、频谱失真度、平均意见分；Calculation unit 203, configured to calculate the evaluation index of the new signal, such as signal-to-noise ratio, spectrum distortion degree, and average opinion score;

比较单元204，用于从计算单元203接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到傅立叶变换单元501，直到接收的测评指标大于或等于所述预置的测评指标。Thecomparison unit 204 is configured to receive the evaluation index from thecalculation unit 203, compare the size of the evaluation index with the preset evaluation index, and send the new signal to Fourier when the evaluation index is smaller than the preset evaluation index.Transformation unit 501 until the received evaluation index is greater than or equal to the preset evaluation index.

其中，傅立叶变换单元501、频域平滑单元502和傅立叶逆变换单元503置于线性处理单元202中。Wherein, theFourier transform unit 501 , the frequencydomain smoothing unit 502 and the inverseFourier transform unit 503 are placed in thelinear processing unit 202 .

该实施例提供的装置示意图相比实施例一提供的装置示意图，将线性预测单元301，替换单元302，替换为傅立叶变换单元501、频域平滑单元502和傅立叶逆变换单元503，用于修正频谱发散，使音频信号的频谱平滑，达到消除边界效应的目的。Compared with the schematic diagram of the device provided inEmbodiment 1, the schematic diagram of the device provided in this embodiment replaces thelinear prediction unit 301 and thereplacement unit 302 with aFourier transform unit 501, a frequencydomain smoothing unit 502 and an inverseFourier transform unit 503 for modifying the frequency spectrum Divergence, to smooth the frequency spectrum of the audio signal, to achieve the purpose of eliminating the boundary effect.

以上实施例可以看出，由于音频信号的不连续发生在相邻数据帧的边界附近，因此在音频信号数据帧边界附近截取一段信号，对截取的所述信号进行线性处理，用线性处理后的信号替换原来的信号，并计算经过线性处理后的新的信号的测评指标，当经过线性处理后的新的信号的测评指标小于预置的测评指标时，继续进行线性处理，进一步修正边界附近的音频信号，直到线性处理后的信号的测评指标大于或等于预置的测评指标，使得相邻数据帧边界附近的信号具有连续性，实现了消除边界效应的目的。As can be seen from the above embodiments, since the discontinuity of the audio signal occurs near the boundary of adjacent data frames, a section of signal is intercepted near the boundary of the audio signal data frame, and the intercepted signal is linearly processed, and the linearly processed The signal replaces the original signal, and the evaluation index of the new signal after linear processing is calculated. When the evaluation index of the new signal after linear processing is smaller than the preset evaluation index, linear processing is continued to further correct the audio signal, until the evaluation index of the linearly processed signal is greater than or equal to the preset evaluation index, so that the signal near the border of adjacent data frames has continuity, and the purpose of eliminating the border effect is achieved.

本领域普通技术人员可以理解实现上述实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件完成，所述的程序可以存储于一种计算机可读存储介质中。Those skilled in the art can understand that all or part of the steps in the methods of the above embodiments can be implemented by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium.

上述提到的存储介质可以是只读存储器，磁盘或光盘等。The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, and the like.

以上对本发明所提供的一种修正音频信号的方法及装置进行了详细介绍，对于本领域的一般技术人员，依据本发明实施例的思想，在具体实施方式及应用范围上均会有改变之处，综上所述，本说明书内容不应理解为对本发明的限制。The method and device for correcting an audio signal provided by the present invention have been introduced in detail above. For those skilled in the art, according to the idea of the embodiment of the present invention, there will be changes in the specific implementation and application scope. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims (14)

Translated fromChinese

1.一种修正音频信号的方法，其特征在于，包括：1. A method for revising an audio signal, comprising:在音频信号数据帧边界附近截取一段信号；Intercept a section of signal near the boundary of audio signal data frame;对截取的所述信号进行线性处理，得到新的信号；performing linear processing on the intercepted signal to obtain a new signal;计算所述新的信号的测评指标，当所述测评指标小于预置的测评指标时，继续进行线性处理，直到线性处理后的信号的测评指标大于或等于所述预置的测评指标。Calculate the evaluation index of the new signal, and when the evaluation index is less than the preset evaluation index, continue the linear processing until the evaluation index of the linearly processed signal is greater than or equal to the preset evaluation index.2.根据权利要求1所述的方法，其特征在于，所述对截取的所述信号进行线性处理，包括：2. The method according to claim 1, wherein said performing linear processing on the intercepted signal comprises:对截取的信号进行线性预测，所得的预测值代替数据帧边界附近的跳变值。Linear prediction is performed on the intercepted signal, and the obtained prediction value replaces the jump value near the boundary of the data frame.3.根据权利要求1所述的方法，其特征在于，所述对截取的所述信号进行线性处理，包括：3. The method according to claim 1, wherein said performing linear processing on the intercepted signal comprises:对数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；Taking the average value of at least two data points before and after the data frame jump, and using the data points and the average value to make a linear curve;以平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据。Take the average value as the reference point, or any point in front of the average value as the reference point, or any point after the average value as the reference point, interpolate data on the linear curve, and replace the original corresponding position with the interpolated data The data.4.根据权利要求1所述的方法，其特征在于，所述对截取的所述信号进行线性处理，包括：4. The method according to claim 1, wherein said performing linear processing on the intercepted signal comprises:对所述信号进行快速傅立叶变换，将时域变成频域；performing a fast Fourier transform on the signal to change the time domain into the frequency domain;截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；Intercepting the high-frequency part of the frequency domain, performing forward or backward dislocation addition and averaging, and using the average value to replace the high-frequency part;对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用所述新的信号代替快速傅立叶变换之前的信号。Inverse Fast Fourier Transform is performed on the high frequency part replaced by the average value to obtain a new signal, and the new signal is used to replace the signal before Fast Fourier Transform.5.根据权利要求2所述的方法，其特征在于，所述对截取的所述信号进行线性处理，包括：5. The method according to claim 2, wherein said performing linear processing on the intercepted signal comprises:在所述信号中从前向后截取一段信号设置为奇数标志，从后向前截取一段信号设置为偶数标志；Intercepting a section of signal from front to back in the signal is set as an odd sign, and intercepting a section of signal from back to front is set as an even sign;对奇数标志的信号和偶数标志的信号进行线性预测，获取奇数标志的信号的预测值和偶数标志的信号的预测值，并对偶数标志的信号的预测值再进行前后倒置；Carry out linear prediction on the signal of the odd number sign and the signal of the even number sign, obtain the predicted value of the signal of the odd number sign and the predicted value of the signal of the even number sign, and invert the predicted value of the signal of the even number sign;对所述奇数标志的信号的预测值，和经过前后倒置的偶数标志的信号的预测值取平均值，用所述平均值代替数据帧边界附近的跳变值。Taking the average value of the predicted value of the odd-numbered signal signal and the predicted value of the even-numbered signal signal after inversion, and using the average value to replace the jump value near the boundary of the data frame.6.根据权利要求1所述的方法，其特征在于，所述测评指标，包括：6. The method according to claim 1, wherein the evaluation index comprises:音频信号的信噪比、频谱失真度、平均意见分。Signal-to-noise ratio, spectral distortion, average opinion score of audio signal.7.一种修正音频信号的装置，其特征在于，包括：7. A device for correcting an audio signal, comprising:截取信号单元，用于在音频信号数据帧边界附近截取一段信号；The intercept signal unit is used to intercept a section of signal near the boundary of the audio signal data frame;线性处理单元，用于对接收的信号进行线性处理，得到新的信号；a linear processing unit, configured to linearly process the received signal to obtain a new signal;计算单元，用于计算所述新的信号的测评指标；a calculation unit, configured to calculate the evaluation index of the new signal;比较单元，用于从计算单元接收测评指标，比较所述测评指标与预置的测评指标的大小，当所述测评指标小于预置的测评指标时，将所述新的信号发送到线性处理单元，直到接收的测评指标大于或等于所述预置的测评指标。A comparison unit, configured to receive the evaluation index from the calculation unit, compare the size of the evaluation index with a preset evaluation index, and send the new signal to the linear processing unit when the evaluation index is smaller than the preset evaluation index until the received evaluation index is greater than or equal to the preset evaluation index.8.根据权利要求7所述的装置，其特征在于，所述线性预测单元包括：8. The device according to claim 7, wherein the linear prediction unit comprises:线性预测单元，用于对接收的信号进行线性预测，获取预测值；a linear prediction unit, configured to linearly predict the received signal to obtain a predicted value;替换单元，用于将接收的预测值代替数据帧边界附近的跳变值，得到新的信号。The replacement unit is used to replace the jump value near the boundary of the data frame with the received prediction value to obtain a new signal.9.根据权利要求8所述的装置，其特征在于，所述装置进一步包括：9. The device according to claim 8, wherein the device further comprises:奇偶标志单元，用于从截取信号单元接收信号，对所述信号从前向后截取一段信号设置为奇数标志，并发送到所述线性预测单元，获取奇数标志的信号的预测值；从后向前截取一段信号设置为偶数标志；The parity flag unit is used to receive the signal from the intercepting signal unit, intercept a segment of the signal from front to back and set it as an odd flag, and send it to the linear prediction unit to obtain the predicted value of the signal of the odd flag; from back to front Intercept a section of signal and set it as an even sign;前后倒置单元，用于从所述线性预测单元接收到偶数标志的信号的预测值时，进行前后倒置；A front-to-back inversion unit, configured to perform front-to-back inversion when receiving the predicted value of the even-numbered signal signal from the linear prediction unit;平均值单元，用于接收所述奇数标志的信号的预测值，及偶数标志的信号的预测值，并对所述奇数标志的信号的预测值和经过前后倒置的偶数标志的信号的预测值取平均值，获得截取信号单元截取信号的预测值。The average value unit is used to receive the predicted value of the signal of the odd number sign and the predicted value of the signal of the even number sign, and take the predicted value of the signal of the odd number sign and the predicted value of the signal of the even number sign after the inversion Average, to get the predicted value of the intercepted signal unit intercepted signal.10.根据权利要求8所述的装置，其特征在于，所述装置进一步包括：10. The device according to claim 8, further comprising:绘制单元，用于对数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；A drawing unit, configured to average at least two data points before and after the data frame jump, and use the data points and the average value to make a linear curve;线性内插单元，用于以所述平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据。A linear interpolation unit for interpolating data on the linear curve with the average value as a reference point, or any point before the average value as a reference point, or any point after the average value as a reference point, The interpolated data is used to replace the original corresponding position data.11.根据权利要求8所述的装置，其特征在于，所述装置进一步包括：11. The device according to claim 8, further comprising:傅立叶变换单元，用于对所述信号进行快速傅立叶变换，将时域变成频域；A Fourier transform unit, configured to perform fast Fourier transform on the signal to change the time domain into the frequency domain;频域平滑单元，用于截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；A frequency-domain smoothing unit, configured to intercept the high-frequency part of the frequency domain, perform forward or backward dislocation addition and average, and replace the high-frequency part with the average;傅立叶逆变换单元，用于对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用所述新的信号代替快速傅立叶变换之前的信号。The inverse Fourier transform unit is configured to perform inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new signal, and use the new signal to replace the signal before the fast Fourier transform.12.根据权利要求7所述的装置，其特征在于，所述线性处理单元包括：12. The device according to claim 7, wherein the linear processing unit comprises:绘制单元，用于对数据帧跳变前后的至少两个数据点取平均值，用所述数据点和所述平均值作线性曲线；A drawing unit, configured to average at least two data points before and after the data frame jump, and use the data points and the average value to make a linear curve;线性内插单元，用于以所述平均值为基准点，或者平均值前面的任意一个点为基准点，或者平均值后面的任意一个点为基准点，在所述线性曲线上内插数据，用内插的数据取代原来相应位置的数据。A linear interpolation unit for interpolating data on the linear curve with the average value as a reference point, or any point before the average value as a reference point, or any point after the average value as a reference point, The interpolated data is used to replace the original corresponding position data.13.根据权利要求12所述的装置，其特征在于，所述装置进一步包括：13. The device according to claim 12, further comprising:傅立叶变换单元，用于对所述信号进行快速傅立叶变换，将时域变成频域；A Fourier transform unit, configured to perform fast Fourier transform on the signal to change the time domain into the frequency domain;频域平滑单元，用于截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；A frequency-domain smoothing unit, configured to intercept the high-frequency part of the frequency domain, perform forward or backward dislocation addition and average, and replace the high-frequency part with the average;傅立叶逆变换单元，用于对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用所述新的信号代替快速傅立叶变换之前的信号。The inverse Fourier transform unit is configured to perform inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new signal, and use the new signal to replace the signal before the fast Fourier transform.14.根据权利要求7所述的装置，其特征在于，所述线性处理单元包括：14. The device according to claim 7, wherein the linear processing unit comprises:傅立叶变换单元，用于对所述信号进行快速傅立叶变换，将时域变成频域；A Fourier transform unit, configured to perform fast Fourier transform on the signal to change the time domain into the frequency domain;频域平滑单元，用于截取所述频域的高频部分，进行向前，或者向后的错位相加求平均值，用所述平均值代替所述高频部分；A frequency-domain smoothing unit, configured to intercept the high-frequency part of the frequency domain, perform forward or backward dislocation addition and average, and replace the high-frequency part with the average;傅立叶逆变换单元，用于对用平均值替代后的高频部分进行快速傅立叶逆变换，得到新的信号，用所述新的信号代替快速傅立叶变换之前的信号。The inverse Fourier transform unit is configured to perform inverse fast Fourier transform on the high-frequency part replaced by the average value to obtain a new signal, and use the new signal to replace the signal before the fast Fourier transform.

Images