CN104021791B - Detecting method based on digital audio waveform sudden changes - Google Patents

Abstract

The invention discloses a detecting method based on digital audio waveform sudden changes, which is a statistical judging method disclosed according to the characteristic of a ridge of a speech spectrum caused by audio waveform sudden changes, and belongs to the multimedia information safety field. For waveform sudden change detection due to copying-pasting operation of a digital audio, ridge span changes before and after audio joint points in a speech spectrum log domain are analyzed, and a ridge factor is constructed to describe ridge bandwidth of a one-frame audio log speech spectrum to represent short-time energy change characteristics, and a difference operator is utilized for distinguishing sudden noises of audio, and ridge factor identification is caused by audio waveform sudden changes. The method comprises the following steps of carrying out short-time Fourier transformation and logarithmic transformation on an audio signal to obtain an audio spectrum of the log domain; calculating the ridge factor of each frame in the speech spectrum; and detecting and judging the difference transformation result for the ridge factor. The detecting method disclosed by the invention can identify audio waveform sudden changes effectively, and provides an effective method for detecting a digital audio editing operation boundary.

Description

The detection method of DAB waveform mutation

Technical field

The present invention relates to field of multi-media information safety, especially a kind of detection method of DAB waveform mutation.

Background technology

The progress of the digitized of multimedia technology and transmission technology is so that the application of DAB increases rapidly.Numeral Audio frequency is easy to be transmitted propagating by physics or electronics system, but simultaneously exactly these advantages also bring many new Problem：DAB record and copy procedure in may suffer from distorting consciously or unconsciously.Either premeditate artificial destruction Integrity, verity distort it is also possible to transmission communication process in occur mistake, all can be original to the information data of itself Property damage, the information especially acquiring a special sense at some, such as court's proof, department's classified papers, historical document are standby The important contents such as part, once maliciously being distorted, can cause very serious consequence.

Digital speech replicates that to paste/delete and distort be that one section of sound bite in digital speech copies to another voice sheet Duan Zhong, or delete the fragment in one section of voice, it is a kind of tampering methods simply effectively changing voice important information, such as says Shown in bright book accompanying drawing 1,2.Due to there being consistent or similar noise in same section of voice, speaker's vocal print so that human ear very Hardly possible distinguishes to the sound bite after distorting, and detects that the judgement distorted to voice true or false of this form has important reality Meaning.

Content of the invention

The technical problem to be solved is the detection method providing a kind of mutation of DAB waveform, and it can be to giving One section of fixed voice, judges whether it have passed through sound bite and replicate the audio volume control pasted or delete one section of voice and cause It is mutated moreover it is possible to orient the time range that voice is distorted, thus confirming the true or false of voice, to overcome the deficiencies in the prior art.

The present invention is realized in：The detection method of DAB waveform mutation, comprises the following steps：1）Audio frequency is believed Number conversion obtains the audio frequency language spectrum of log-domainY, the audio frequency language spectrum of acquisition is carried out logarithmic transformation, obtains logarithm language spectrumG；2）Enter Row logarithm language spectrum G energy binaryzation calculates；3）Calculate every frame logarithm language spectrumGAudio frequency change coefficientσ _t ；4）To audio frequency mutantion line Numberσ _t Judged, carried out audio volume control abrupt climatic change and Sudden change region positioning.

Step 1）The described audio frequency language spectrum that audio signal conversion is obtained log-domainY, the audio frequency language spectrum of acquisition is carried out Logarithmic transformation, obtains logarithm language spectrumG,Specifically, the digital audio and video signals for h for lengthyCarry out framing, obtaining frame number isN _l , frame length is 2*NMatrix；Add window function and carry out Short Time Fourier Transform, obtaining size isN * N ₁ Audio frequency language SpectrumY；Audio frequency language is composedYCarry out logarithmic transformation, obtain logarithm language spectrumG, its size isN * N ₁ .

Step 2）Described enters logarithm language spectrumGEnergy binaryzation calculates, and specifically, is first calculated logarithm language spectrumGIn MaximumG _max And minimaG _min If the energy value of every frame rate isG _ti (1≤t≤N, 1≤i≤N_l), by such as Lower formula（1）Calculate energy binary valueδ(t,i),

WhereinλFor threshold factor.

Step 3）Described calculating every frame logarithm language is composedG _t Audio frequency change coefficient, specifically, by formula（2）Calculate sound Frequency change coefficientσ _t （1≤t≤N ₁ ）；

Step 4）Described to audio frequency change coefficientσ _t Judged, carry out audio volume control abrupt climatic change and Sudden change region is fixed Position, specifically it is assumed that in step 3) in obtained logarithm language spectrum G i-th frameG _i And three audio frequency change coefficients of consecutive frameσ _i-1 、 σ _i 、σ _i+1 If meeting：

Then determine in audio frequency there is audio frequency mutation, the wherein i-th frameG _i It is the audio volume control Sudden change region detecting.

Compared with prior art, the present invention is using distorting in splicing voice process signal waveform mutation so that this time The property that the frequency short-time energy of fragment is uprushed, distinguishes, using difference algorithm, audio volume control mutation and the strong signal that editor causes Audio frequency is mutated, and after calculating audio frequency change coefficient, detects that this section audio has duplication to paste/delete and distorts vestige, and accurately The positioning tampering time.The inventive method has preferable detection and positioning function, cannot observe in human ear None- identified, sound spectrograph The waveform mutation of audio frequency can be identified in the case of coming well, be that tampering detection is pasted/deleted in a kind of effective audio dubbing Technology.

Brief description

Fig. 1 is audio data insertion audio fragment schematic diagram；

Fig. 2 is that audio data deletes audio fragment schematic diagram；

Fig. 3 is that waveform mutation schematic diagram at splice point distorted in voice；

Fig. 4 is detection algorithm flow chart；

Fig. 5 is the oscillogram of detection voice；

Fig. 6 is the logarithm language spectrum of detection voice；

Fig. 7 is the audio frequency change coefficient of detection voice.

Specific embodiment

Embodiments of the invention 1：The detection method of DAB waveform mutation,

1）By audio signalyConversion obtains the audio frequency language spectrum of log-domainY：For the digital audio and video signals for h for the lengthy , to audio signalyCarry out framing, every frame length is2 * N（Set N=128）；Duplication isl（Setl= 0.5）, then Frame numberN ₁For：

If framing signal isy _i , i=1...N_l, using formula（1）Carry out Discrete Short Time Fourier Transform, obtainY _ti, its Middle w (N-m) is window function（It is set as Hamming window function）.

Using formula（2）Calculate spectrogram, carry out logarithmic transformation and obtain logarithm language spectrumG _i (i=1...N_l), that is,

All log-magnitude values are formed the logarithm language spectrum that matrix is voice signalG, its size isN * N ₁ ；

2）Carry out logarithm language spectrumG _i Energy binaryzation calculates：Calculate logarithm language spectrum firstG _i In maximumG _max And minimum ValueG _min , for the energy value of every frame rateG _i （k）,k = 1 ... N, by formula（3）Calculate the energy two-value of frequency Change valueδ _(t,i) ：

WhereinλFor threshold factor（Setλ =0.65）, willδThe matrix Δ of composition is defined as short-time energy two-value spectrum,δ _(i,k) Value represents for 0iFramekFrequency content energy is low, and 1 is that energy is high；

3）Calculate every frame logarithm language spectrumG _i Audio frequency change coefficientσ _t ：Because imagineering pastes sound bite, in editor Place's audio signal occurs that waveform is mutated, and this operation introduces new radio-frequency component so that log-magnitude spectrum comprises audio frequency mutation The all frequency energy of frame increase suddenly with respect to consecutive frame, therefore, the nonzero value number of this frame short-time energy two-value spectrum should substantially More than consecutive frame；Statistic procedure 2）Middle short-time energy two-value composes the average of every frameσ _i , according to formula（4）It is calculated audio frequency mutation Coefficientσ _i （1≤i≤N ₁ ）；

4）To audio frequency change coefficientσ _t Judged, carried out audio volume control abrupt climatic change and Sudden change region positioning：Due to language Sound feature, strong voice signal can make the audio frequency change coefficient that detection obtains become big, but this signal has time duration, manually Duplication is pasted the audio frequency change coefficient causing and only can be present in a frame, therefore, distinguishes, using difference algorithm, the sound that editor causes The audio frequency mutation of the mutation of frequency waveform and strong signal, it comprises the concrete steps that, by formula（5）Calculate every frame logarithm language spectrumG _i Audio frequency Change coefficientσ _t （1≤t≤N1）；

Step 4）Described to audio frequency change coefficientσ _t Judged, carry out audio volume control abrupt climatic change and Sudden change region is fixed Position, specifically it is assumed that in step 3) in obtained logarithm language and composed the i-th frameG _i And three audio frequency change coefficients of consecutive frameσ _i-1 、 σ _i 、σ _i+1 If meeting：

Then determine in audio frequency there is audio frequency mutation, the wherein i-th frameG _i It is the audio volume control Sudden change region detecting.

As shown in figure 5, Fig. 5 is to distort speech waveform figure after deleting part of speech, the wherein dotted line mark moment is sound Frequency clipped position, and this editing trace cannot be gone out by auditory discrimination..As shown in Figure 6 although in the logarithm language spectrum of this section audio No significantly short-time energy mutation, as shown in fig. 7, detecting that this section audio has duplication viscous after calculating audio frequency change coefficient Patch/delete and distort vestige, and it is accurately positioned the time of distorting.

From above-described embodiment as can be seen that the inventive method has preferable detection and positioning function.Human ear None- identified, Sound spectrograph can identify the waveform mutation of audio frequency well in the case of cannot observing out, is that a kind of effective audio dubbing is glued Patch/delete tampering detection technology.

Claims (1)

1. a kind of mutation of DAB waveform detection method it is characterised in that：Comprise the following steps：1) audio signal is converted Obtain the audio frequency language spectrum Y of log-domain, the audio frequency language spectrum of acquisition is carried out logarithmic transformation, obtain logarithm language spectrum G；2) carry out logarithm language Spectrum G energy binaryzation calculates；3) calculate every frame logarithm language and compose G_tAudio frequency change coefficient σ_t；4) to audio frequency change coefficient σ_tCarry out Judge, carry out audio volume control abrupt climatic change and Sudden change region positioning；

Step 1) described in the audio frequency language spectrum Y that audio signal conversion is obtained log-domain, the audio frequency language of acquisition spectrum is carried out logarithm Conversion, obtains logarithm language spectrum G, specifically, carries out framing for the digital audio and video signals y for h for the length, and obtaining frame number is N_l, frame The matrix of a length of 2*N；Add window function and carry out Short Time Fourier Transform, obtaining size is N*N₁Audio frequency language spectrum Y；To audio frequency Language spectrum Y carries out logarithmic transformation, obtains logarithm language spectrum G, and its size is N*N₁；

Step 2) described in carry out logarithm language spectrum G energy binaryzation calculate, specifically, be first calculated logarithm language compose G in maximum Value G_maxWith minima G_minIf the energy value of every frame rate is G_ti(1≤t≤N,1≤i≤N_l), energy is calculated by equation below (1) Amount binary value δ (t, i),

Wherein λ is threshold factor；

Step 3) described in calculating every frame logarithm language spectrum G_tAudio frequency change coefficient, specifically, by formula (2) calculate audio frequency dash forward Variable coefficient σ_t(1≤t≤N₁)；

${\mathrm{\σ}}_t=\frac{\underset{i=1}{\overset{n}{\Σ}}\mathrm{\δ}(t,i)}{n}---\left(2\right);$

Step 4) described in audio frequency change coefficient σ_tJudged, carried out audio volume control abrupt climatic change and Sudden change region positioning, tool Body is it is assumed that in step 3) in obtained logarithm language spectrum G the i-th frame G_iAnd three audio frequency change coefficient σ of consecutive frame_i-1、σ_i、σ_i+1, If meeting：

σ_i>0.85 and | σ_i-σ_i-1|*|σ_i-σ_i+1|>σ_i/16

Then determine in audio frequency there is audio frequency mutation, the wherein i-th frame G_iIt is the audio volume control Sudden change region detecting.