JP2008192102A - Metadata generation device and metadata generation method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technology for effectively generating metadata. <P>SOLUTION: In this metadata generation device, a low-level metadata extraction part 120 extracts low-level metadata from content data, and a medium-level metadata extraction part 140 extracts medium-level metadata from the low-level metadata. A beat feature amount analysis part 122 and a time-interval analysis part 124 extract low-level metadata respectively from the same music data 112a. A beat information extraction part 142 and a medium-level music feature amount extraction part 144 extract medium-level metadata respectively from the low-level metadata extracted by the beat feature amount analysis part 122 and the time-interval analysis part 124. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technique for generating content metadata.

  Metadata is content-related information, and often indicates content-related information such as the date and time of creation, creator, data format, and title. For example, in an MP3 (MPEG1 Aidio Layer-3) file, bibliographic information in which a song title, artist name, etc. are written as text data is embedded according to a standard called ID3 tag. Recently, the entry size by the ID3 tag has been expanded, the number of items that can be entered has increased, and image data can also be included.

  In recent years, attempts have been made to analyze music data and analyze and extract music features. By extracting rhythms or the like that are characteristic features of music as metadata, it is possible to perform processing such as categorizing music. As described above, various types of metadata can be extracted from one content, and by increasing the number of types of metadata, various applications using the metadata can be realized.

  In order to apply metadata to an application, it is preferable to increase the accuracy of metadata extraction. For example, from MP3 files, metadata derived from music features and metadata described in ID3 tags can be extracted, but these metadata exist independently. There have been no attempts to integrate and process them. The present inventor has developed a technique that can effectively generate metadata by processing content data and metadata that exist independently in association with each other.

  In order to solve the above-described problem, a metadata generation apparatus according to an aspect of the present invention includes a first metadata extraction unit that extracts first-level metadata from content data, and a second metadata from the first-level metadata. A second metadata extraction unit for extracting level metadata, and hierarchically generating metadata related to the content. In the metadata generation apparatus, the first metadata extraction unit extracts a plurality of types of first level metadata from the first data included in the content data, and the second metadata extraction unit includes a plurality of types. The second level metadata is extracted from the first level metadata.

  A metadata generation method according to another aspect of the present invention includes a step of extracting first level metadata from content data, and a step of extracting second level metadata from the first level metadata. The present invention relates to a metadata generation method for hierarchically generating metadata related to content. In this method, the step of extracting the first level metadata includes a step of extracting a plurality of types of first level metadata and extracting a second level metadata from one type of data included in the content data. Extracts one type of second level metadata from a plurality of types of first level metadata.

  It should be noted that any combination of the above-described constituent elements and a conversion of the expression of the present invention between a method, an apparatus, a system, a recording medium, a computer program, etc. are also effective as an aspect of the present invention.

  According to the present invention, it is possible to provide a technique for effectively generating metadata.

  FIG. 1 shows a use environment of an information processing system according to an embodiment of the present invention. The information processing system 1 includes an information processing apparatus 10 that executes an application and a display device 12 that outputs a processing result in the information processing apparatus 10. The display device 12 may be a television set that includes a display unit that outputs an image and further includes an audio output unit that outputs audio. The display device 12 may be connected to the information processing apparatus 10 by a wired cable, or may be wirelessly connected by a wireless local area network (LAN) or the like. In the information processing system 1, the information processing apparatus 10 is connected to an external network such as the Internet via a cable 14. Connection to an external network may be performed by wireless communication. The information processing apparatus 10 may be a game apparatus that is loaded with a medium on which game data is recorded, executes the game data, and generates an image signal and an audio signal indicating the processing result of the game application.

  In the present embodiment, the information processing apparatus 10 has a function of receiving and reproducing content data such as a music file and a video file from a content providing server connected to an external network. The content data may be file format data or streaming format data. As content data, for example, a review file of a music album by a critic may be included. The content data may be supplied from a recording medium such as an optical disc, a magneto-optical disc, or a Blu-ray disc instead of via a network. The information processing apparatus 10 functions as a metadata generation apparatus that hierarchically generates metadata related to content from the acquired content data. The information processing apparatus 10 also has a function of extracting new metadata that has not existed before. When the information processing apparatus 10 temporarily stores content data in a large-capacity auxiliary storage device and generates metadata from the content data, the information processing apparatus 10 stores a metadata file in the auxiliary storage device separately from the content data. The generated metadata is used for searching content data.

  FIG. 2 shows a content distribution system according to an embodiment of the present invention. In the content distribution system 2, the information processing apparatus 10 that is a user terminal and the content providing servers 18 a, 18 b, and 18 c that are terminals owned by the content provider (hereinafter collectively “content”) via the network 16 such as the Internet. And a providing server 18 ”) that are communicably connected. The content provider stores the content data in the content providing server 18 so that the information processing apparatus 10 can download the content data from the content providing server 18.

  When operated by the user, the information processing apparatus 10 accesses the content providing server 18 that holds desired content data, and downloads the content data. The content providing server 18 holds, for example, digital contents of a music album containing a plurality of songs, and the user can acquire a desired album file by paying a fee. The information processing apparatus 10 stores the content data downloaded from the content providing server 18 in a predetermined area of the auxiliary storage device.

  FIG. 3 shows a functional block diagram of the information processing apparatus 10. The information processing apparatus 10 includes a power button 20, an LED 22, a system controller 24, a device controller 30, a media drive 32, a hard disk drive 34, a switch 36, a wireless interface 38, a main controller 100, a main memory 102, and an output processing unit 200. Configured.

  The power button 20 is an input unit where an operation input from a user is performed, and is operated to turn on or off the power supply to the information processing apparatus 10. The power button 20 may be a push button, and the power on or off may be controlled by being pressed. The power button 20 may have another structure such as a touch sensor that allows the user to turn on / off the power. The LED 22 illuminates and displays the power on / off state. The system controller 24 detects whether the power button 20 is pressed or not, and when detecting a state transition from the power-off state to the pressed state, activates the main controller 100 and controls the lighting of the LED 22. When the power cable is inserted into the information processing apparatus 10, the system controller 24 maintains the standby mode even when the power is off, and monitors the pressing of the power button 20.

  The device controller 30 is configured as an LSI (Large-Scale Integrated Circuit) that exchanges information between devices like a south bridge. As illustrated, devices such as a system controller 24, a media drive 32, a hard disk drive 34, a switch 36, and the main controller 100 are connected to the device controller 30. The device controller 30 absorbs the difference in electrical characteristics of each device and the difference in data transfer speed, and controls the timing of data transfer.

  The media drive 32 is a drive device that loads and drives a medium 50 that records application data, and reads application data from the medium 50. The medium 50 may be a read-only recording medium such as an optical disc, a magneto-optical disc, or a Blu-ray disc.

  The hard disk drive 34 is an auxiliary storage device that drives a built-in hard disk and writes / reads data using a magnetic head. Content data downloaded from the content providing server 18 and content data supplied from the medium 50 are stored in the hard disk drive 34. The switch 36 is an Ethernet switch (Ethernet is a registered trademark), and is a device that transmits and receives information by connecting to an external device in a wired or wireless manner. In the present embodiment, the cable 14 is inserted into the switch 36 and is communicably connected to the network 16. Further, the switch 36 is connected to a wireless interface 38, and the wireless interface 38 is connected to a wireless controller 40 having a wireless communication function using a communication protocol such as Bluetooth (registered trademark) protocol or IEEE802.11 protocol. The wireless controller 40 functions as an input unit where operation input from the user is performed.

  The main controller 100 includes a multi-core CPU, and includes one general-purpose processor core and a plurality of simple processor cores in one CPU. The general-purpose processor core is called a PPU (Power Processing Unit), and the remaining processor cores are called a SPU (Synergistic-Processing Unit).

  The main controller 100 includes a memory controller connected to a main memory 102 that is a main storage device. The PPU has a register, has a main processor as an operation execution subject, and efficiently assigns a task as a basic processing unit in each application to each SPU. Note that the PPU itself may execute the task. The SPU has a register, and includes a sub-processor as an operation execution subject and a local memory (dedicated RAM) as a local storage area. The SPU has a dedicated DMA (Direct Memory Access) controller as a control unit, and can transfer data at high speed by transferring data between the main memory 102 and the local memory, and is built in the output processing unit 200. High-speed data transfer can be realized between the frame memory and the local memory.

  The output processing unit 200 is connected to the display device 12 and outputs a video signal and an audio signal that are processing results of the application. The output processing unit 200 includes a GPU (Graphics Processing Unit) that realizes an image processing function. The GPU employs HDMI (High Definition Multimedia Interface) and can digitally output a video signal without using analog.

  FIG. 4 shows an internal configuration of the main controller 100 that executes the metadata generation process. The main controller 100 includes a low level metadata extraction unit 120, a medium level metadata extraction unit 140, and a high level metadata extraction unit 160. In FIG. 4, each element described as a functional block for performing various processes can be configured by a CPU (Central Processing Unit), a memory, and other LSIs in terms of hardware. This is realized by a program loaded on the computer. As described above, the main controller 100 is provided with one PPU and a plurality of SPUs, and each of the PPUs and SPUs can be configured individually or in cooperation with each other. Therefore, it is understood by those skilled in the art that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof, and is not limited to any one.

  The low-level metadata extraction unit 120, the medium-level metadata extraction unit 140, and the high-level metadata extraction unit 160 are hierarchical metadata from the content data 110a and 110b (hereinafter collectively referred to as “content data 110”). Extract data. The low-level metadata extraction unit 120 extracts low-level (primary level) metadata by directly using data included in the content data 110. The intermediate level metadata extraction unit 140 extracts intermediate level (secondary level) metadata using the low level metadata extracted by the low level metadata extraction unit 120. The high level metadata extraction unit 160 uses the medium level metadata extracted by the medium level metadata extraction unit 140 to extract high level (third level) metadata. As described above, the main controller 100 generates metadata step by step (hierarchical), and generates metadata useful for searching content data with high accuracy. From the viewpoint of signal processing, the hierarchization of metadata in this embodiment is executed in the order of analysis processing, recognition processing, and understanding processing so that the level of metadata in each layer is uniform. Note that the process from the analysis process to the recognition process may be set as one level, that is, the primary level. In this case, metadata is extracted in two levels.

  First, the low-level metadata extraction unit 120 executes an analysis process, and extracts low-level metadata obtained by analyzing content data as a physical quantity. The low level metadata extraction unit 120 may extract low level metadata from the plurality of content data 110a and 110b by performing signal processing. Here, among the plurality of content data 110, one content data 110a exists as a creation target of metadata, and other content data 110b and the like are auxiliary for creating metadata for the content data 110a. It may be used as data. For example, when the content data 110a is an MP3 file obtained by compressing music data, and the content data 110b is a review text file for the music, the content data 110b includes metadata about the music data of the content data 110a. It may be used to create. In this case, the content data 110a can be used to create metadata of the content data 110b.

  In addition to compressed music data, the MP3 file includes text data such as a song name and an artist name according to a standard called an ID3 tag, and may further include image data of a jacket photo of a music album. In the information processing apparatus 10 according to the present embodiment, the low-level metadata extraction unit 120 acquires three types of data, that is, music data, text data, and image data, from the MP3 file, and obtains low-level metadata from each data. Extract. At this time, the low level metadata extraction unit 120 may extract a plurality of types of low level metadata from one type of data included in the MP3 file. For example, the low-level metadata extraction unit 120 may extract a plurality of types of low-level metadata from compressed music data, or may extract a plurality of types of low-level metadata from text data.

  The low-level metadata extraction unit 120 can perform time-pitch analysis by performing signal processing such as fast Fourier transform on the music data, and the pitch analysis result as a physical quantity is converted into low-level metadata. Can be extracted. Similarly, it is possible to analyze the beat feature amount of the music data, thereby analyzing the change position of the sound in the music. This beat feature amount is also extracted as low-level metadata. The low-level metadata extraction unit 120 can extract low-level metadata by performing morphological analysis on text data, and can extract low-level metadata by performing image analysis on the image data.

  Next, the intermediate level metadata extraction unit 140 executes a recognition process, and extracts the intermediate level metadata by uniquely converting the low level metadata. For example, by dividing the pitch analysis result into segments of a certain unit and performing recognition processing, music features such as the height and size of the main musical instrument and voice are extracted as medium level metadata. Further, more accurate beat information is extracted as intermediate level metadata from the beat feature amount. In the analysis process, when the pitch analysis result and the beat feature value are extracted as low-level metadata from one music data, they are mutually used to obtain a more accurate medium-level music feature value and / or Alternatively, beat information can be extracted. Further, in the recognition process, keywords and syntax in the text can be extracted as medium level metadata from the result of morphological analysis in the low level metadata extraction unit 120, and an image in the medium level metadata extraction unit 140 can be extracted. From the result of the analysis, the number of people included in the image and image feature quantities such as brightness (atmosphere) can be extracted as medium level metadata.

  The high-level metadata extraction unit 160 executes an understanding process, and extracts high-level metadata classified into classes, in other words, high-level metadata symbolized from medium-level metadata. For example, the genre of the music, the presence / absence of an instrument, and the like are extracted as high-level metadata from the music feature amount and beat information. Also, it is possible to extract the relationship between music and keywords as high-level metadata from the extracted keywords and syntax, and to extract human faces, etc., identified from medium-level image feature quantities as high-level metadata. Can do.

  In the main controller 100, a plurality of types of low-level metadata are extracted from a plurality of types of data included in the content data 110a and 110b, and a plurality of types of medium-level metadata are generated therefrom. The high-level metadata extraction unit 160 can also extract, for example, high-level music feature amounts from different types of medium-level metadata. As described above, medium-level metadata generated from different types of data such as music data, text data, and / or image data is integrated and processed to extract one type of high-level metadata. This makes it possible to increase the generation accuracy of high-level metadata.

  FIG. 5 shows details of the main controller 100 shown in FIG. In FIG. 5, the main controller 100 generates metadata hierarchically from content data 110 a that is an MP3 file and content data 110 b that is text data of music reviews. The MP3 file includes compressed music data 112a, JPEG-recorded jacket photo image data 112b, and music bibliographic data 112c described in text data. The jacket photo image data 112b and / or the music bibliographic data 112c may be a file downloaded from the content providing server 18 on the network 16 separately from the MP3 file.

  The low-level metadata extraction unit 120 includes a beat feature amount analysis unit 122, a time-pitch analysis unit 124, an image analysis unit 126, a morpheme analysis unit 128, and a morpheme analysis unit 130. Extract metadata. The beat feature amount analysis unit 122 and the time-pitch analysis unit 124 extract low-level metadata from a music signal obtained by decoding the monaural processed music data 112a. The image analysis unit 126 extracts low-level metadata from the jacket photo image data 112b. Similarly, the morpheme analysis unit 128 and the morpheme analysis unit 130 extract low-level metadata from the music bibliographic data 112c and the music review data 112d, respectively. Extract. In the low level metadata extraction unit 120, the beat feature amount analysis unit 122 and the time-pitch analysis unit 124 extract metadata from the music data 112a, respectively. As a result, a plurality of metadata is extracted from the music data 112a included in the content data 110a, and multifaceted low-level metadata can be obtained. Extraction accuracy can be increased. Further, the low level metadata extraction unit 120 extracts a plurality of low level metadata from the music data 112a, the jacket photo image data 112b, the music bibliographic data 112c, and the music review data 112d, which are different from each other. It is also possible to improve the accuracy of extracting medium and high level metadata.

  The beat feature value analysis unit 122 divides the music signal obtained by decoding the monaural processed music data 112a into a band corresponding to the instrument feature value. Next, by selecting and smoothing the band corresponding to the beat signal, the sound attack time and release time candidate positions are obtained, and the fundamental period is obtained from the autocorrelation function of each band. Thereby, the beat feature amount analysis unit 122 extracts a beat feature amount.

  The time-pitch analysis unit 124 performs frequency analysis on a music signal obtained by decoding and monaurally processing the information-compressed music data 112a at a short time interval to obtain a time-frequency feature quantity. For this frequency analysis, for example, a Fast Fourier Transform (FFT) unit or a filter bank which is an aggregate of bandpass filters may be used. The time-pitch analysis unit 124 divides a digital signal having a sampling frequency of 44.1 kHz by a band corresponding to the pitch, samples the digital signal at intervals of 1 to 20 ms, and extracts a time-frequency (pitch) feature quantity.

  The image analysis unit 126 executes an object segmentation process from the jacket photo image data 112b, and analyzes an object in the jacket photo. The image analysis unit 126 may perform a color analysis process to obtain a hue histogram. Conventionally, there is nothing that uses the image analysis result of the jacket photo image data 112b as metadata. However, in the information processing system 1 of the present embodiment, this is newly handled as low-level metadata, so that the type of metadata And the choices in the metadata extraction processing at the intermediate level and later can be expanded.

  The morphological analysis unit 128 extracts bibliographic data from the music bibliographic data 112c. For example, the bibliographic data may include an album title, artist name, song name, songwriter, composer, label name, genre, and the like. The morpheme analysis unit 130 performs morpheme analysis on the music review data 112d and classifies the morpheme into parts of speech. Here, it is assumed that the music review data 112d is text data in which a music reviewer reviews and criticizes a music album included in the MP3 file. A plurality of morpheme analyzers 130 that process the music review data 112d may be provided. In that case, each morpheme analysis unit may perform morpheme analysis using different dictionaries or the like so that the analysis results are different. Accordingly, various morphological analyzes can be performed, and the keyword extraction unit 150 in the subsequent stage can extract keywords rich in variations based on different analysis results. Similarly, a plurality of morphological analysis units 128 for processing the music bibliographic data 112c may be provided.

  As described above, the low-level metadata extraction unit 120 extracts a plurality of low-level metadata obtained by analyzing various content data as physical quantities, so that the intermediate-level metadata extraction unit 140 and the high-level metadata extraction unit 160 in the subsequent stage are extracted. Can expand the range of metadata extraction processing options, and can also dynamically change the combination of low-level metadata.

  The intermediate level metadata extraction unit 140 includes a beat information extraction unit 142, an intermediate level music feature amount extraction unit 144, an image feature amount extraction unit 146, a keyword extraction unit 148, and a keyword extraction unit 150, and uniquely sets the low level metadata. Generate converted medium level metadata. The beat information extraction unit 142 extracts medium level metadata from the beat feature value and the time-pitch feature value extracted from the music data 112a. Similarly, the medium level music feature amount extraction unit 144 extracts medium level metadata from the beat feature amount and the time-pitch feature amount extracted from the music data 112a. The beat information extracting unit 142 and the medium level music feature amount extracting unit 144 generate medium level metadata based on different types of low level metadata extracted from one music data 112a. By using a plurality of types of low-level metadata, the accuracy of the generated intermediate-level metadata can be increased. The image feature amount extraction unit 146 extracts medium level metadata from the image analysis data analyzed by the image analysis unit 126. The keyword extraction unit 148 extracts medium level metadata from the morpheme analysis data by the morpheme analysis unit 128. Similarly, the keyword extraction unit 150 extracts medium level metadata from the morpheme analysis data obtained by the morpheme analysis unit 130.

  The beat information extraction unit 142 extracts the beat position of the music signal with high accuracy from the beat feature amount extracted by the beat feature amount analysis unit 122 and the time-pitch feature amount extracted by the time-pitch analysis unit 124. . Since the basic period of music is extracted as the beat feature amount, and the pitch amount for each time is extracted as the time-pitch feature amount, the beat information extraction unit 142 uses the extracted basic cycle as the pitch amount for each time. The correct beat position can be extracted by performing correction or the like. The beat position extracted by the beat information extraction unit 142 is supplied to the subsequent high-level music feature amount extraction unit 162. The extracted beat position may be used for remix processing for connecting different music. By extracting accurate beat positions, smooth remix processing can be realized.

  The medium level music feature amount extraction unit 144 uses the beat feature amount extracted by the beat feature amount analysis unit 122 and the time-pitch feature amount extracted by the time-pitch analysis unit 124 to obtain note information and sound for each measure. Medium-level music features such as music strength, tempo, and music structure. Conventionally, these intermediate level music feature values are derived only from the time-music feature value extracted by the time-pitch analysis unit 124, but the intermediate level music feature value extraction unit 144 is extracted as a beat feature value. By further utilizing the basic period of the music, it is possible to improve the extraction accuracy such as the music structure and tempo.

  The image feature amount extraction unit 146 extracts image feature amounts such as the position, size, and hue of the object from the image analysis data extracted by the image analysis unit 126.

  The keyword extraction unit 148 extracts keywords from the morpheme analysis data extracted by the morpheme analysis unit 128. The keyword extraction unit 150 also extracts keywords from the morpheme analysis data extracted by the morpheme analysis unit 130. The keyword extraction unit 150 holds, for example, terms commonly used in music reviews as a table. Such terms may be keywords such as “ballad”, “easy to hear”, “settled”, etc., and the keyword extraction unit 150 extracts the terms included in the table from the morphological analysis extracted by the morphological analysis unit 130. Extract from the data.

  The high level metadata extraction unit 160 includes a high level music feature amount extraction unit 162, a face extraction unit 164, and an evaluation extraction unit 166. The high-level metadata extraction unit 160 uses the high-level metadata from the medium-level metadata extracted by the beat information extraction unit 142, the medium level music feature amount extraction unit 144, the image feature amount extraction unit 146, and the keyword extraction unit 148. To extract. The face extraction unit 164 extracts high-level metadata from the medium level metadata extracted by the image feature amount extraction unit 146. The evaluation extraction unit 166 extracts high-level metadata from the medium level metadata extracted by the keyword extraction unit 148 and the keyword extraction unit 150.

  The high-level metadata extraction unit 160 extracts, as high-level metadata, feature quantities for the entire music such as music genre, musical instrument sound, mood, sound quality, speed, and sound quality as high-level music feature quantities. Further, as the high level feature quantity, it is possible to extract the ratio of the energy of the rhythm instrument to the total energy, the number of notes per unit time, and the like. The high level metadata extraction unit 160 receives the medium level metadata extracted by the beat information extraction unit 142, the medium level music feature amount extraction unit 144, the image feature amount extraction unit 146, and the keyword extraction unit 148. Extract music features with high accuracy.

  For example, the high-level music feature amount extraction unit 162 uses the accurate beat information supplied from the beat information extraction unit 142 and the accurate medium-level music feature amount supplied from the middle-level music feature amount extraction unit 144 to determine the music structure and tempo. Therefore, it is possible to accurately estimate and extract the speed and mood of music. The high-level music feature amount extraction unit 162 can also estimate and extract music genres from these medium-level metadata. As for the genre, a genre is included in the keyword supplied from the keyword extraction unit 148, and a new accurate genre may be determined from the estimated genre and the genre included in the keyword. For example, even if the genre of “ballad” is extracted as a keyword, if the music structure or tempo is estimated to be slightly intense rock music, the genre of the music is locked. A high-level music feature amount may be extracted. In this case, a genre between ballad and rock, for example, pop may be extracted as a genre. In this way, it is possible to extract highly accurate metadata by matching the bibliographic data with the result of analyzing and recognizing actual music. Further, according to the image feature amount supplied from the image feature amount extraction unit 146, for example, when a bright color is used for a jacket photo, the high-level music feature amount extraction unit 162 estimates that the music will be fun. However, the high-level music feature value may be extracted from other elements such as the tempo and the musical instrument used, assuming that the music has a bright tone. Conventionally, no attempt has been made to use the color tone of a jacket photo or the feature amount of an object included in the jacket photo as metadata. On the other hand, in this embodiment, the high-level music feature amount extraction unit 162 can extract new metadata by generating high-level metadata including the image feature amount supplied from the image feature amount extraction unit 146. In addition, it is possible to increase the accuracy of metadata extraction.

  The face extraction unit 164 extracts a face from the image feature amount. For example, the extraction result may be used in an application that determines who the extracted face is by, for example, face recognition processing.

  The evaluation extraction unit 166 associates the keywords extracted by the keyword extraction unit 148 and the keyword extraction unit 150. For example, when the music review data 112d includes an album list by the same artist, a process in which the bibliographic data and the album list in the music bibliographic data 112c are linked may be performed.

  In the above, this invention was demonstrated based on the Example. This embodiment is an exemplification, and it will be understood by those skilled in the art that various modifications can be made to the combination of each component and each processing process, and such modifications are also within the scope of the present invention. .

  In the embodiment, the example of generating the metadata of the MP3 music file has been described. However, the content data is not limited to this, and may be a simple audio file or video data. All targets for which metadata needs to be created may be included.

  In the embodiment, the example in which the information processing apparatus 10 functions as a metadata generation apparatus has been described. However, for example, the content providing server 18 may function as a metadata generation apparatus. In this case, the content providing server 18 may distribute the metadata generated from the content data to the information processing apparatus 10 together with the content data.

It is a figure which shows the use environment of the information processing system concerning the Example of this invention. It is a figure which shows the content delivery system concerning the Example of this invention. It is a figure which shows the functional block diagram of information processing apparatus. It is a figure which shows the internal structure of the main controller which performs a metadata production | generation process. It is a figure which shows the detail of the main controller shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Information processing system, 10 ... Information processing apparatus, 12 ... Display apparatus, 100 ... Main controller, 110 ... Content data, 112a ... Music data, 112b ... Jacket picture Image data, 112c ... Music bibliographic data, 112d ... Music review data, 120 ... Low level metadata extraction unit, 122 ... Beat feature amount analysis unit, 124 ... Time-pitch analysis unit, 126 ... Image analysis unit, 128 ... Morphological analysis unit, 130 ... Morphological analysis unit, 140 ... Medium level metadata extraction unit, 142 ... Beat information extraction unit, 144 ... Medium level Music feature extraction unit, 146... Image feature extraction unit, 148... Keyword extraction unit, 150... Keyword extraction unit, 160. 162 ... high-level music characteristic quantity extracting section, 164 ... face extraction unit, 166 ··· evaluation extractor.

Claims (6)

A first metadata extraction unit for extracting first level metadata from content data;
A metadata generation device that hierarchically generates metadata related to content, including a second metadata extraction unit that extracts second level metadata from first level metadata;
The first metadata extraction unit extracts a plurality of types of first level metadata from first data included in content data,
The second metadata extraction unit extracts second level metadata from a plurality of types of first level metadata.   The metadata generation apparatus according to claim 1, further comprising a third metadata extraction unit that extracts third level metadata from the second level metadata. The first metadata extraction unit extracts a plurality of types of first-level metadata from first data included in content data and second data having different types from the first data,
The second metadata extraction unit is configured to extract a second level from a plurality of types of metadata extracted based on different types of first data and second data included in the content data in the first metadata extraction unit. The metadata generation apparatus according to claim 1, wherein metadata is extracted. The first metadata extraction unit extracts a plurality of types of first-level metadata from first data included in content data and second data having different types from the first data,
The second metadata extraction unit includes a plurality of types of first data from a plurality of types of metadata extracted based on different types of first data and second data included in the content data in the first metadata extraction unit. Extract two levels of metadata,
3. The third metadata extraction unit according to claim 2, wherein the third metadata extraction unit extracts third level metadata from a plurality of types of second level metadata extracted by the second metadata extraction unit. Metadata generation device. Extracting first level metadata from content data;
A metadata generation method for hierarchically generating metadata related to content, comprising: extracting second level metadata from first level metadata;
The step of extracting the first level metadata extracts a plurality of types of first level metadata from one type of data included in the content data,
The step of extracting the second level metadata includes extracting one type of second level metadata from a plurality of types of first level metadata. On the computer,
A function of extracting a plurality of types of first level metadata from one type of data included in the content data;
A program that realizes a function of extracting one type of second level metadata from a plurality of types of first level metadata, and hierarchically generates metadata related to the content.

Images