JP2006330256A - Electronic watermark embedding method and extraction method for audio signal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an electronic watermark embedding and detection method for an audio signal, with which the influence on sound quality can be suppressed to the utmost, even if the degree of operating the coefficient after frequency conversion increases to some extent. <P>SOLUTION: The two coefficients C [x, i, N] and C [y, i, N] in the MDCT coefficient of a certain frame N are extracted and the information to be embedded E [i, N] is embedded. If it is assumed, here, E [i, N]=0 is embedded, if C [x, i, N]≥C [y, i, N], and E [i, N]=i is embedded, if C [x, i, N]<C [y, i, N], C [y, i, N] is so operated that C [x, i, N] ≥ C [y, i, N] holds if C [x, i, N]<C [y, i, N], when, for example, "0" is desired to be embedded. At this time, a reverse operation is performed for the corresponding coefficients of the adjacent frame, to negate the influence on the sound quality due to the operation of the frame N. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

   The present invention relates to a digital watermark embedding method and detection method for an audio signal, and more particularly to a digital watermark embedding method and detection method for an audio signal that can embed and detect a digital watermark in an audio signal with almost no deterioration in sound quality.

  Conventionally, several techniques for embedding a digital watermark in an audio signal have been studied, but most of them manipulate spectral coefficients (frequency components). In particular, there are many methods for manipulating MDCT (Modified Discrete Cosine Transform) coefficients used in audio encoding such as MP3 and AAC. For example, it is a general method to embed digital watermark data by performing an operation such as adding s (any positive number) to a predetermined spectral coefficient (MDCT coefficient).

  Also, an audio signal composed of a pseudo random number (PN) sequence composed of s and -s is created, and this is divided into areas for several samples. When the bit to be embedded in each area is "1", it remains as it is. In the case of “0”, there is a method of embedding digital watermark data by inverting the sign and performing frequency conversion, and adding it to the frequency conversion of the original audio signal. Here, s represents the strength of the watermark.

For example, in Patent Document 1 below, paying attention to the magnitude of two coefficients after frequency conversion, information of “0” and “1” is assigned to the magnitude relation, and s is assigned to one coefficient as necessary. There has been proposed a method of embedding digital watermark data by assigning information of “0” and “1” by performing an operation such as addition.
JP 2003-143390 A

  However, in the method described in Patent Document 1, the degree of manipulation of the coefficient after frequency conversion, that is, when s representing the strength of the watermark is increased, the greater the degree, the higher the detection probability of the watermark. There was a problem that the sound quality would be greatly affected.

  The present invention has been made in view of the above-described prior art, and an object of the present invention is to provide an audio that can suppress the influence on the sound quality as much as possible even when the degree of operation of the coefficient after frequency conversion is increased to some extent. An object of the present invention is to provide a digital watermark embedding method and a detection method for a signal.

  In order to achieve the above-described object, the present invention is a digital watermark embedding method for an audio signal, wherein a digital audio signal (hereinafter referred to as a frame) divided every unit time in the time domain is a signal in the frequency domain. When the digital watermark information is embedded according to the size of a predetermined set of signal values in the signal in the frequency domain, and the signal value is manipulated to embed the digital watermark information, The first feature is that the reverse operation is added.

  Further, the present invention is a method for detecting digital watermark information embedded by the digital watermark embedding method, wherein a predetermined set of signal values in a frame in which the digital watermark information is embedded is extracted, and the magnitude of the signal value is detected. Thus, the second feature is that the digital watermark information is detected.

  According to the present invention, watermark information can be embedded without substantially degrading the sound quality of an audio signal.

  In addition, it becomes possible to easily detect watermark information, confirm the validity of the digital audio file, and confirm the copyright holder and the like.

  In addition, embedding and detection in a compressed file such as MP3 and AAC do not require signal conversion and frequency conversion such as MDCT conversion, and thus can be realized at high speed.

  Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an embodiment of the present invention.

  The digital audio signal which is the original signal a is divided by the dividing unit 1 every unit time having a time domain. This divided digital audio signal is called an original signal frame. The MDCT unit 2 shifts two frames (first frame and second frame) one frame at a time and performs MDCT conversion, that is, conversion to a frequency domain signal. The MDCT unit 2 outputs an MDCT coefficient sequence b.

  The embedding processing unit 3 extracts an arbitrary set of coefficient pairs (pairs) from each frame subjected to the MDCT conversion, and the digital watermark information “1” or “0” depending on the magnitude of one coefficient and the other coefficient of the pair. Is allocated. On the other hand, the digital watermark information is not allocated to the next frame, and when the previous frame is operated for embedding the digital watermark information, an operation to cancel or cancel this operation is performed. Details of the embedding processing unit 3 will be described later with reference to FIGS. The IMDCT unit 4 performs inverse MDCT conversion on the digital audio signal in which the digital watermark information is embedded and outputs it.

  Next, the function or operation of the embedding processing unit 3 will be described with reference to FIGS.

  As shown in FIG. 3, the original signal a is subjected to MDCT conversion by shifting one frame at a time with respect to two frames, that is, N and N + 1, N + 1 and N + 2,. The reason why the two frames are set as one set and shifted one frame at a time to perform MDCT conversion is to prevent discontinuity due to MDCT conversion at the boundary portion of each frame.

  The embedding processing unit 3 extracts a pair (one pair) of coefficients C [x, i, N] and C [y, i, N] from each frame of the MDCT coefficients. Here, x and y are the positions of the coefficients in the frame, i is the pair number, and N is the frame number. In the following, in order to simplify the description, one pair (i = 1) is used, but the present invention is not limited to this, and a plurality of pairs may be used.

In the present invention, when the information to be embedded is E [i, N], the relationship between the E [i, N] and the coefficients C [x, i, N] and C [y, i, N] is as follows. Promise as in Equation 1.

E [i, N] = 0 (when C [x, i, N] ≧ C [y, i, N])
= 1 (when C [x, i, N] <C [y, i, N])
... (Formula 1)

  In FIG. 2, in step S <b> 1, a number (N is assumed) representing a frame number is set to 1. In step S2, MDCT coefficients C [x, i, N] and C [y, i, N] are extracted from the frame N. In step S3, it is determined whether the information to be embedded is “0” or “1”. If “0”, the process proceeds to step S4, and if “1”, the process proceeds to step S5.

  In step S4, it is determined whether C [x, i, N] ≧ C [y, i, N] is satisfied. If this determination is affirmative, there is no need to perform an operation on the coefficient of the frame N, so the process proceeds to step S6 and N is simply incremented by one. On the other hand, when the determination in step S4 is negative, the process proceeds to step S7, where C [y, i, N] ← C [y, i, N] -s is operated, and C [x, i, N ] ≧ C [y, i, N]. In step S8, N is incremented by one.

  In step S5, it is determined whether C [x, i, N] <C [y, i, N] is satisfied. If this determination is affirmative, the operation proceeds to step S6 without performing any operation on the coefficient of the frame N. On the other hand, when the determination in step S5 is negative, the operation C [x, i, N] <C [x, i, N] -s is performed, and C [x, i, N] <C [y, i , N].

  As a result, information “0” or “1” is embedded in the frame N.

  Next, when the operation in step S7 or S9 is performed in frame N, processing for canceling or canceling the operation is performed in the next frame N + 1. That is, in step S11, C [y, i, N + 1] is extracted, and the operation of C [y, i, N + 1] ← C [y, i, N + 1] + s is performed. In step S13, an operation of C [x, i, N + 1] ← C [x, i, N + 1] + s is performed. By this operation, the operation performed in the frame N is almost canceled by the reverse operation of the frame N + 1, and deterioration of sound quality can be prevented.

  In step S15, N is further incremented, and the process returns to step S2. Thereafter, the operations in steps S2 to S15 are repeated.

  FIG. 3 expresses the processing of FIG. 2 in an easy-to-understand manner, and thus description thereof is omitted.

  In the above-described embodiment, in the frames N and N + 1, the operation at the frame N is canceled by using the pair i of the same positions x and y. However, the present invention is not limited to this, and the frame N + 1 A pair of positions in the vicinity of the positions x and y may be used.

  In the embodiment, one watermark information is embedded in one frame. However, if the number of pairs per frame is plural, for example, 2 (i = 2), a plurality of pairs per frame, for example, Two pieces of watermark information can be embedded. In this case, when an operation is performed on a pair, an inverse operation is performed on the coefficient of each corresponding pair of adjacent frames so as to cancel the influence of the operation.

  Next, when digital watermark embedding is performed on ISO MPEG-1 / 2 Layer3 (so-called MP3), which is encoding using MDCT, or AAC, it is not necessary to completely decode the audio signal. A simple MDCT coefficient may be decoded and operated by the above method. In other words, when the audio signal is encoded by MP3 or AAC or the like, as described above, the processing is divided for each unit time and the MDCT conversion is omitted, and the digital watermark information is directly added to the quantized MDCT coefficient. It can be embedded as described above.

  Next, processing for detecting watermark information embedded as described above will be described with reference to FIG.

  As in the case of FIG. 3, the original signal p in which the watermark information is embedded is shifted by one frame with respect to two frames, that is, MDCT conversion with N and N + 1, N + 1 and N + 2,. I do. Next, the MDCT coefficients C [x, i, N] and C [y, i, N] of the frame in which the watermark information is embedded are extracted, the magnitude relationship between them is examined, The watermark data E [i, N] is extracted. That is, if C [x, i, N] ≧ C [y, i, N], the watermark information is “0”, and C [x, i, N] <C [y, i, N]. , “1”. This process is performed for all pairs (C [x, i, N], C [y, i, N]) of the frame in which the watermark information is embedded. In the embedding process, the variable of the adjacent frame is manipulated. However, in the detection process, it is clear that such an operation is not necessary at all.

  Therefore, it becomes possible to easily detect watermark information, confirm the validity of the digital audio file, and confirm the copyright holder.

  Next, when detecting digital watermark information for ISO MPEG-1 / 2 Layer 3 (so-called MP3), which is encoding using MDCT, or AAC, it is not necessary to completely decode, and the necessary MDCT What is necessary is just to decode a coefficient and to detect watermark information by said method.

  In this manner, embedding and detection in a compressed file such as MP3 and AAC do not require signal conversion and frequency conversion such as MDCT conversion, and can be realized at high speed.

It is a block diagram which shows the outline | summary of the process of this invention. It is a flowchart explaining operation | movement of one Embodiment of watermark information embedding processing. It is explanatory drawing of a watermark information embedding process. It is explanatory drawing of a watermark information detection process.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Dividing part, 2 ... MDCT part, 3 ... Embedding process part, 4 ... IMDCT part.

Claims (5)

An electronic watermark embedding method for an audio signal, comprising:
A digital audio signal (hereinafter referred to as a “frame”) divided into unit times in the time domain is converted into a frequency domain signal,
Embedding digital watermark information by the size of a predetermined set of signal values in the frequency domain signal,
An electronic watermark embedding method, wherein when the signal value is manipulated to embed the electronic watermark information, an inverse operation is applied to an adjacent frame. The digital watermark embedding method according to claim 1,
An electronic watermark embedding method, wherein MDCT is applied as the frequency domain signal conversion. An electronic watermark embedding method for an audio signal, comprising:
For the MP3 or AAC encoding using the MDCT, the watermark information is embedded by the size of a predetermined set of signal values in the quantized MDCT coefficient of the frame,
An electronic watermark embedding method, wherein when the signal value is manipulated to embed the electronic watermark information, an inverse operation is applied to an adjacent frame. The digital watermark embedding method according to any one of claims 1 to 3,
An electronic watermark embedding method characterized by embedding one or more pieces of electronic watermark information per frame. A method for detecting digital watermark information embedded by the digital watermark embedding method according to claim 1, comprising:
Extracting a predetermined set of signal values in a frame embedded with digital watermark information;
A method for detecting digital watermark information, comprising: detecting digital watermark information based on the magnitude of the signal value.

Images