JPH11265181A - Watermark information embedding processor, watermark information embedding method, watermark information reading processor and watermark information reading method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make the removal of MIDI data without destruction difficult and to embed watermark information without damaging quality as MIDI data. SOLUTION: This embedding processor is composed of a sound volume density calculating means 5 for calculating the sound volume density of supplied MIDI data 1 per unit time for each musical scale, sync pattern selecting means 6 for selecting the pattern of small correlation 8 (sync pattern) in the sound volume density calculated by this sound volume density calculating means 5, watermark embedding position selecting means 10 for selecting a watermark information embedding position corresponding to the sound volume density as a relative position with the pattern selected by this sync pattern selecting means 6 as a reference, change event detecting means 7 for detecting an event corresponding to the watermark information embedding position selected by this watermark embedding position selecting means 10, and watermark information embedding means 9 for embedding watermark information 2 in the event detected by the change event detecting means 7.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION For MIDI data used as performance information of a communication karaoke or the like, in order to prevent unauthorized copying or unauthorized use of MIDI data, information about a creator such as a copyright is watermarked (digital). wat
The present invention relates to a watermark information embedding apparatus and a watermark information embedding method for recording as a watermark information, and a watermark information reading processing apparatus and a watermark information reading method for reading the watermark information.

[0002]

2. Description of the Related Art In recent years, content data such as image data and audio data has been digitized, recorded on a storage medium such as a CD-ROM and distributed, or transmitted / received via a network such as the Internet. Unauthorized use, such as unauthorized copying of data or sale of the copied data to the copyright holder without permission, has become a problem. As a means for preventing such unauthorized use, a digital watermark embedding method of recording information about a creator such as copyright in content data as watermark information (digital watermark) has been researched and developed.

The conditions required for this digital watermark are as follows:
Embedding the watermark information so that the content data will be broken if you try to forcibly remove it (the watermark information indicating the copyright remains in the content data), it is difficult to understand where the watermark information is embedded in the content data, Even when the watermark information is embedded, the original state of the content data may be kept. Digital watermarking technology that meets these conditions includes:
A method has been considered in which watermark information is embedded as noise in a portion (redundant portion) that is not important for human perception in image data and audio data, and the watermark information is embedded without changing the entire data amount. The methods are roughly classified into a method of embedding watermark information in sample values of image data and audio data, and a method of embedding watermark information in frequency components.

[0004]

The MIDI data has recently been used as performance information for communication karaoke, and has itself been used as music information. Therefore, a copyright is generated for each MIDI data, and it is necessary to protect the copyright. Recently, as communication karaoke not only for business use but also for home use, music data (MIDI data) is transmitted over the Internet or the like and played by a software synthesizer or the like that operates on a personal computer at home. .

[0005] In this home-use communication karaoke,
Once the song data (MIDI data) has been received, it can be used over and over again,
It can be stored on a hard disk or the like. For this reason, music data (MIDI data) can be illegally copied, and the copied data can be used commercially without the permission of the right holder.

Therefore, it is recognized that it is necessary to protect the copyright by embedding watermark information in MIDI data. However, due to the nature of MIDI data, frequency components such as image data and audio data as described above are required. Cannot be embedded, and sufficient strength cannot be obtained even if watermark information is individually embedded in sample values. Therefore, the present invention
An object of the present invention is to embed watermark information in MIDI data so that it is difficult to remove the MIDI data without destroying the MIDI data, and that the quality of the MIDI data is not impaired.

[0007]

The standard MIDI data includes a note-on event indicating a timing at which a sound starts to be emitted and a note-off event indicating a timing at which the sound is stopped to be generated. Recorded with the velocity representing The volume can be controlled by volume control events such as a volume control event and an expression control event for setting the volume of the sound source for each channel. The sound volume of each note when actually sounding a sound source such as a synthesizer is determined by the parameters of these velocity and volume control events. When a third party edits the MIDI data in which the watermark information is discretely embedded, such as by cutting out or transposing a part of the data,
It becomes difficult for the right holder to specify the embedding position of the watermark information, and the right holder cannot read the watermark information.

Therefore, in the watermark information embedding method of the present invention, the volume density per unit time is obtained for each interval, and a characteristic (small correlation) constant-size pattern is selected therefrom, which is used as a sync pattern. . Then, when embedding the watermark information into the volume density, the watermark information embedding position is determined by a vector including a time component and a pitch component with the head of the sync pattern as a reference point, thereby performing processing such as data cutting and modulation. (As long as the sync pattern is not destroyed), the embedding position of the watermark data can be found. Here, since the volume density is information that is difficult to falsify among the information included in the MIDI data, it can be considered that the sync pattern is hardly destroyed.

The present invention aims to provide the following apparatus and method as means for solving the above problems. 1. Volume density calculating means for calculating the volume density per unit time for each interval based on the velocity information of all note events and the volume-related control event information contained in the supplied MIDI data; A sync pattern selecting unit for selecting a pattern having a small correlation in the volume density calculated by the volume density calculating unit; and a watermark for the volume density as a relative position based on the pattern selected by the sync pattern selecting unit. A watermark embedding position selecting means for selecting an information embedding position; a change event detecting means for detecting an event corresponding to the watermark information embedding position selected by the watermark embedding position selecting means; and a watermark information detected by the change event detecting means. Watermark information embedding means for embedding in a specified event. And a watermark information embedding processing device.

2. The volume density per unit time is calculated for each interval based on the velocity information of all note events and the volume-related control event information included in the MIDI data, and the correlation among the volume densities is calculated. A watermark information embedding method characterized by selecting a pattern having a small size, and embedding watermark information in MIDI data based on the selected pattern.

3. Volume for calculating volume density per unit time for each interval based on velocity information of all note events and volume-related control event information included in MIDI data in which watermark information is embedded. The density calculator calculates a correlation between the volume density supplied from the volume density calculator and a sync pattern obtained from MIDI data before embedding watermark information, and determines a portion having the largest correlation with a sync pattern of the volume density. The embedded watermark information is output by detecting and reading the position of the watermark information based on the sync pattern detecting means for detecting the watermark information and the watermark relative embedding position information indicating the embedding position of the watermark information with the head of the sync pattern as the reference point. Watermark information reading means having watermark information reading means. Equipment.

4. Based on the velocity information of all note events included in the MIDI data in which the watermark information is embedded and the control event information of the volume system, the volume density per unit time is obtained for each interval, and before embedding the watermark information A correlation with a sync pattern obtained from MIDI data is obtained, a portion having the largest correlation is set as a sync pattern of this volume density, and watermark relative embedding position information indicating an embedding position of watermark information with the head of the sync pattern as a reference point. A watermark information reading method, wherein the watermark information is output by detecting and reading the position of the watermark information.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a watermark information embedding processing apparatus according to the present invention. The watermark information embedding processing apparatus shown in FIG.
Volume density calculating means 5 for calculating the volume density per unit time of each interval of the IDI data 1 and a pattern (sink pattern) 8 having a small correlation in the volume density calculated by the volume density calculating means 5 are selected. Sync pattern selecting means 6, a watermark embedding position selecting means 10 for selecting a watermark information embedding position with respect to the volume density as a relative position based on the pattern selected by the sync pattern selecting means 6, and a watermark embedding position The change event detection means 7 detects an event corresponding to the watermark information embedding position selected by the selection means 10, and the watermark information embedding means 9 embeds the watermark information 2 in the event detected by the change event detection means 7. ing.

[0014] The performance information (song data) MI
The DI data 1 is first supplied to the volume density calculating means 5 to calculate the volume density per unit time for each interval.
The calculation of the volume density is based on the assumption that there is a signal of a certain strength between the note-on event and the corresponding note-off event, and the velocity of this note event is calculated using a volume control event such as a volume control event or an expression control event. This is performed by integrating the corrected value in a section from note-off to note-on.

That is, note events having the same pitch (pitch) among all note events in which timbres having pitches of all MIDI channels (excluding timbres having no pitch such as rhythm timbres) are designated. Are all integrated while multiplying by a Hamming window function having a width slightly wider than the unit time to obtain a volume density at each unit time.

FIG. 3 is a flowchart showing the processing procedure for calculating the volume density in the volume density calculating means 5 and will be described below. Note that frameLength in the figure is the length of a unit time window (constant value), and the unit of time is ticks (ticks). In addition, powerTable is a value of a volume table obtained from a velocity determined for each tone and a value of a volume control event, humming_window is a humming window function, powerDensity is a volume density, and size_of_song is a length of a song. .

In the process shown in FIG. 1, first, initialization is performed. For the pitch (pitch), the initial value is set to 0, and 127 is added one by one. The frame number (unit time frame number) is set to an initial value of 0, and is added one by one up to (length of music / length of unit time). Then, the initial values of the sound volume densities determined for each pitch and frame number are all set to 0 (step 21).

Next, referring to the first one of all the note events included in the designated pitch and frame number (step 22), the note-on timing (the time from the start of the music is ticks. ), The pitch of the note, and the velocity of the note are substituted for the respective variables (step 23). Then, the frame number is calculated by dividing the note-on timing by the length of the unit time window (step 24), and the value of the volume table determined by the velocity of the note and the value of the volume control event is substituted for the variable power. (Step 25). Further, the variable power is multiplied by a hamming window and integrated into the volume density (step 26). The value of the humming window at this time is prepared in a table as a value corresponding to a value obtained by multiplying the ratio of the time given by “time” to the whole music by the length of the unit time window.

Then, the time given by time is set to 1 ticks
The addition is performed (Step 27), and Step 24 and the subsequent steps are repeated until the time given by time becomes the note-off timing (Step 28). Further, when the time given by "time" becomes the note-off timing, the next note event is referred to, and step 22 and thereafter are repeated (step 29). In this way, the pitch and the volume density per unit time can be obtained. In step 27, the time width to be added is not limited to 1, and may be any number as long as it is equal to or less than the length of the unit time window.

Further, by applying a Hamming window function for reducing the weight of the boundary between the sections, even if an error occurs in the section detection position in a unit time when reading the watermark information, the calculated volume density has an error. The influence of is hardly caused. The volume density calculated in this manner is a parameter that does not greatly change even when the structure of the MIDI data is changed, and thus is considered suitable for embedding watermark information.

The volume density calculated by the volume density calculating means 5 is supplied to the sync pattern selecting means 6 shown in FIG. Here, the correlation function is multiplied by the volume density to find a portion where the correlation is small (emulation is unlikely to occur), and a part of the volume density that is continuous in the time direction and the pitch direction of a fixed length is extracted. Is a sync pattern 8. It is desirable to extract one or more sync patterns 8 per phrase so that the sync patterns 8 are included even when the music is cut.

The sync pattern 8 is held by the right holder for use at the time of reading, and is output to the watermark embedding position selecting means 10. The watermark embedding position selecting means 10 selects a position for embedding watermark information with respect to the volume density supplied from the volume density calculating means 5, obtains this position as a relative position from the sync pattern 8, and calculates a time component and a pitch component. The information is output as watermark relative embedding position information 4 (owned by the right holder) and output to the change event detecting means. The change event detecting means 7 detects a velocity or volume control event corresponding to the embedding position indicated by the watermark relative embedding position information 4.

Further, the watermark information embedding means 9 embeds watermark information in the velocity or volume control event designated by the change event detecting means 7 for the input MIDI data 1.
The simplest example of embedding the watermark information is a method of directly adding the watermark information to the volume density. At the time of embedding the watermark information, the relative distance (vector) in the time direction and the pitch direction from the head of the sync pattern as a reference point to the volume density at which the watermark data is to be embedded is set as watermark relative embedding position information 4 as the right holder's side. Keep in.

The embedded MIDI data 3 in which the watermark information is embedded as described above is generally distributed and used. Since the watermark relative embedding position information 4 and the sync pattern 8 are owned by the right holder, a third party can extract only the watermark information from the embedded MIDI data 3 without destroying the contents of the MIDI data. Is very difficult.

Next, a description will be given of a watermark information reading processing apparatus for reading watermark information from the embedded MIDI data 3 created as described above, with reference to FIG.
The information reading processing device shown in FIG. 1 is used by a right holder to read watermark information from MIDI data that may be illegally used.

First, the MIDI data 3 after embedding is
It is supplied to the volume density calculating means 11 and calculates the volume density (powerDensity) in exactly the same way as the volume density calculating means 5 of the watermark information embedding processing device described above. Then, the sync pattern detecting means 12 correlates the volume density with the sync pattern 8 held by the right holder, and determines a portion having the largest correlation or a pattern coincident with the sync pattern of this volume density, and detects the detected sync pattern. The pattern is output to the watermark information reading means 13.

The watermark information reading means 13 detects and reads the position of the watermark information embedded in the volume density by the watermark relative embedding position information 4 with the head of the supplied sync pattern 8 as a reference point. Embedded watermark information can be obtained.

[0028]

According to the present invention, since the embedding position of the watermark information is determined with respect to the volume density, the watermark information cannot be changed or read out unless the data is changed so as to destroy the volume during reproduction. Can not be destroyed. Further, since a data change that destroys the volume during reproduction destroys the MIDI data itself, it is extremely difficult to change the watermark information.

Also, the sync pattern and the relative embedding position information of the watermark are held only by the right holder, and without this, the embedding position of the watermark information cannot be specified. And it is virtually impossible to remove. Further, there is an effect that the embedding position of the watermark data can be searched out even when processing such as data cutout or modulation is performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a watermark information embedding processing apparatus according to the present invention.

FIG. 2 is a block diagram illustrating an embodiment of a watermark information reading processing device according to the present invention.

FIG. 3 is a flowchart illustrating a processing procedure of a sound volume density calculating unit.

[Explanation of symbols]

 Reference Signs List 1 MIDI data 2 Watermark information 3 MIDI data after embedding 4 Watermark relative embedding position information 5, 11 Volume density calculating means 6 Sync pattern selecting means 7 Change event detecting means 8 Sync pattern 9 Watermark information embedding means 10 Watermark embedding position selecting means 12 sync pattern detecting means 13 watermark information reading means

────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] April 28, 1999

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0016

[Correction method] Change

[Correction contents]

FIG. 3 is a flowchart showing the processing procedure for calculating the volume density in the volume density calculating means 5 and will be described below. Note that frameLength in the figure is the length of a unit time window (constant value), and the unit of time is ticks (ticks). In addition, powerTable is a value of a volume table obtained from a velocity determined for each tone and a value of a volume control event, hamming_window is a hamming window function, powerDensity is a volume density, and size_of_song is a length of a song. .

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0017

[Correction method] Change

[Correction contents]

In the process shown in FIG. 1, first, initialization is performed. The initial values of the pitch (pitch) and the frame number (unit time frame number) are set to 0 (step 21). And
The initial value of each volume density determined for each pitch and frame number is all set to 0 (step 22).

[Procedure amendment 3]

[Document name to be amended] Statement

[Correction target item name] 0018

[Correction method] Change

[Correction contents]

Next, all the note events having the tone of the designated pitch included in the frame number are stored in a buffer (not shown) (step 23), and the first note event is referred to (step 24). , Note-on timing (time from the start of the song is represented by ticks),
The velocity of the note is assigned to each variable (step 25). Then, the value obtained by multiplying the frame number and the frame length (the head position of the frame) is compared with the note-on timing (time) (step 26). → n
o), adjust the time to the beginning of the frame (step 2)
7), the value of the volume table determined by the velocity of the note and the value of the volume control event is set to the variable powe.
Substitute for r (step 28). In addition, this variable power
Is multiplied by the humming window function to the sound volume density (step 29).

[Procedure amendment 4]

[Document name to be amended] Statement

[Correction target item name] 0019

[Correction method] Change

[Correction contents]

Then, a predetermined integration time unit length is added to the time given by the time (step 70), and the time given by the time becomes the head position of the next frame (step 71 → no), until the note-off timing is reached (step 72 → no), step 28
The subsequent steps are repeated (step 72 → yes). Then time
When the time given by becomes the note-off timing,
Step 24 and subsequent steps are repeated with reference to the next note event stored in the buffer (step 73 → yes).
s). Further, when there are no more note events stored in the buffer (step 73 → no), the process shifts to the next frame (step 74), and a note event whose pitch is equal to the value given by pitch is searched until the end of the music. Step 22 and subsequent steps are repeated until the processing is completed (step 75 → yes)
s). When the search for the note of the same pitch is completed until the end of the music (step 75 → no), the pitch condition of the note event to be stored in the buffer is raised by a semitone (step 7).
6) Step 22 and subsequent steps are repeated until the pitch becomes 128 (step 77). By doing this,
The pitch and volume density per unit time can be determined. Then, the volume density calculated in this way is M
Even if the structure of the IDI data is changed, it is a parameter that does not greatly change (a region in which data is hardly changed).
It is considered suitable for embedding watermark information.

[Procedure amendment 5]

[Document name to be amended] Drawing

[Correction target item name] Figure 3

[Correction method] Change

[Correction contents]

FIG. 3

Claims (4)

[Claims] 1. A sound volume density calculating means for calculating a sound volume density per unit time for each interval based on velocity information of all note events contained in supplied MIDI data and control event information of a volume system. And a sync pattern selecting means for selecting a pattern having a small correlation in the volume density calculated by the volume density calculating means; and the volume as a relative position based on the pattern selected by the sync pattern selecting means. Watermark embedding position selecting means for selecting a watermark information embedding position with respect to density; change event detecting means for detecting an event corresponding to the watermark information embedding position selected by the watermark embedding position selecting means; Watermark information embedding means for embedding in the event detected by the means And a watermark information embedding processing device. 2. A volume density per unit time is calculated for each pitch based on velocity information of all note events and volume-related control event information included in MIDI data, and a correlation among the volume densities is calculated. A watermark information embedding method characterized by selecting a pattern having a small size, and embedding watermark information in MIDI data based on the selected pattern. 3. A volume for calculating a volume density per unit time for each interval based on velocity information of all note events and volume-related control event information included in MIDI data in which watermark information is embedded. A density calculator, and a correlation between the volume density supplied from the volume density calculator and a sync pattern obtained from the MIDI data before embedding the watermark information. The embedded watermark information is output by detecting and reading the position of the watermark information based on the watermark relative embedding position information indicating the embedding position of the watermark information with the head of the sync pattern as the reference point. Watermark information reading means having watermark information reading means. Apparatus. 4. A sound density per unit time is determined for each interval based on velocity information of all note events and volume-related control event information included in MIDI data in which watermark information is embedded, and the watermark is obtained. A correlation with a sync pattern obtained from MIDI data before embedding information is obtained, a portion having the highest correlation is set as a sync pattern of this volume density, and a watermark relative indicating a watermark information embedding position with a head of the sync pattern as a reference point. By detecting and reading the position of the watermark information based on the target embedding position information,
A watermark information reading method, which outputs watermark information.